Showing 1–50 of 145 results for author: Lake, B

Neural networks that overcome classic challenges through practice

Authors: Kazuki Irie, Brenden M. Lake

Abstract: Since the earliest proposals for neural network models of the mind and brain, critics have pointed out key weaknesses in these models compared to human cognitive abilities. Here we review recent work that has used metalearning to help overcome some of these challenges. We characterize their successes as addressing an important developmental problem: they provide machines with an incentive to impro… ▽ More Since the earliest proposals for neural network models of the mind and brain, critics have pointed out key weaknesses in these models compared to human cognitive abilities. Here we review recent work that has used metalearning to help overcome some of these challenges. We characterize their successes as addressing an important developmental problem: they provide machines with an incentive to improve X (where X represents the desired capability) and opportunities to practice it, through explicit optimization for X; unlike conventional approaches that hope for achieving X through generalization from related but different objectives. We review applications of this principle to four classic challenges: systematicity, catastrophic forgetting, few-shot learning and multi-step reasoning; we also discuss related aspects of human development in natural environments. △ Less

Submitted 14 October, 2024; originally announced October 2024.

H-ARC: A Robust Estimate of Human Performance on the Abstraction and Reasoning Corpus Benchmark

Authors: Solim LeGris, Wai Keen Vong, Brenden M. Lake, Todd M. Gureckis

Abstract: The Abstraction and Reasoning Corpus (ARC) is a visual program synthesis benchmark designed to test challenging out-of-distribution generalization in humans and machines. Since 2019, limited progress has been observed on the challenge using existing artificial intelligence methods. Comparing human and machine performance is important for the validity of the benchmark. While previous work explored… ▽ More The Abstraction and Reasoning Corpus (ARC) is a visual program synthesis benchmark designed to test challenging out-of-distribution generalization in humans and machines. Since 2019, limited progress has been observed on the challenge using existing artificial intelligence methods. Comparing human and machine performance is important for the validity of the benchmark. While previous work explored how well humans can solve tasks from the ARC benchmark, they either did so using only a subset of tasks from the original dataset, or from variants of ARC, and therefore only provided a tentative estimate of human performance. In this work, we obtain a more robust estimate of human performance by evaluating 1729 humans on the full set of 400 training and 400 evaluation tasks from the original ARC problem set. We estimate that average human performance lies between 73.3% and 77.2% correct with a reported empirical average of 76.2% on the training set, and between 55.9% and 68.9% correct with a reported empirical average of 64.2% on the public evaluation set. However, we also find that 790 out of the 800 tasks were solvable by at least one person in three attempts, suggesting that the vast majority of the publicly available ARC tasks are in principle solvable by typical crowd-workers recruited over the internet. Notably, while these numbers are slightly lower than earlier estimates, human performance still greatly exceeds current state-of-the-art approaches for solving ARC. To facilitate research on ARC, we publicly release our dataset, called H-ARC (human-ARC), which includes all of the submissions and action traces from human participants. △ Less

Submitted 2 September, 2024; originally announced September 2024.

Comments: 12 pages, 7 figures

Growth of Ba_2CoWO_6 Single Crystals and their Magnetic, Thermodynamic and Electronic Properties

Authors: Abanoub R. N. Hanna, A. T. M. N. Islam, C. Ritter, S. Luther, R. Feyerherm, B. Lake

Abstract: This study explores the bulk crystal growth, structural characterization, and physical property measurements of the cubic double perovskite Ba_2CoWO_6(BCWO). In BCWO, Co+2 ions form a face-centered cubic (FCC) lattice with non-distorted cobalt octahedra. The compound exhibits long-range antiferromagnetic order below TN = 14 K. Magnetization data indicated a slight anisotropy along with a spin-flop… ▽ More This study explores the bulk crystal growth, structural characterization, and physical property measurements of the cubic double perovskite Ba_2CoWO_6(BCWO). In BCWO, Co+2 ions form a face-centered cubic (FCC) lattice with non-distorted cobalt octahedra. The compound exhibits long-range antiferromagnetic order below TN = 14 K. Magnetization data indicated a slight anisotropy along with a spin-flop transition at 10 kOe , a saturation field of 310 kOe and an ordered moment of 2.17 Mu_B at T = 1.6 K. Heat capacity measurements indicate an effective j = 1/2 ground state configuration, resulting from the combined effects of the crystal electric field and spin-orbit interaction. Surface photovoltage analysis reveals two optical gaps in the UV-Visible region, suggesting potential applications in photocatalysis and photovoltaics. The magnetic and optical properties highlight the significant role of orbital contributions within BCWO, indicating various other potential applications. △ Less

Submitted 16 August, 2024; originally announced August 2024.

Comments: 15 pages, 7 figures

Beyond the Doors of Perception: Vision Transformers Represent Relations Between Objects

Authors: Michael A. Lepori, Alexa R. Tartaglini, Wai Keen Vong, Thomas Serre, Brenden M. Lake, Ellie Pavlick

Abstract: Though vision transformers (ViTs) have achieved state-of-the-art performance in a variety of settings, they exhibit surprising failures when performing tasks involving visual relations. This begs the question: how do ViTs attempt to perform tasks that require computing visual relations between objects? Prior efforts to interpret ViTs tend to focus on characterizing relevant low-level visual featur… ▽ More Though vision transformers (ViTs) have achieved state-of-the-art performance in a variety of settings, they exhibit surprising failures when performing tasks involving visual relations. This begs the question: how do ViTs attempt to perform tasks that require computing visual relations between objects? Prior efforts to interpret ViTs tend to focus on characterizing relevant low-level visual features. In contrast, we adopt methods from mechanistic interpretability to study the higher-level visual algorithms that ViTs use to perform abstract visual reasoning. We present a case study of a fundamental, yet surprisingly difficult, relational reasoning task: judging whether two visual entities are the same or different. We find that pretrained ViTs fine-tuned on this task often exhibit two qualitatively different stages of processing despite having no obvious inductive biases to do so: 1) a perceptual stage wherein local object features are extracted and stored in a disentangled representation, and 2) a relational stage wherein object representations are compared. In the second stage, we find evidence that ViTs can learn to represent somewhat abstract visual relations, a capability that has long been considered out of reach for artificial neural networks. Finally, we demonstrate that failure points at either stage can prevent a model from learning a generalizable solution to our fairly simple tasks. By understanding ViTs in terms of discrete processing stages, one can more precisely diagnose and rectify shortcomings of existing and future models. △ Less

Submitted 22 June, 2024; originally announced June 2024.

Goals as Reward-Producing Programs

Authors: Guy Davidson, Graham Todd, Julian Togelius, Todd M. Gureckis, Brenden M. Lake

Abstract: People are remarkably capable of generating their own goals, beginning with child's play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behavior, models are still far from capturing the richness of everyday human goals. Here, we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-… ▽ More People are remarkably capable of generating their own goals, beginning with child's play and continuing into adulthood. Despite considerable empirical and computational work on goals and goal-oriented behavior, models are still far from capturing the richness of everyday human goals. Here, we bridge this gap by collecting a dataset of human-generated playful goals (in the form of scorable, single-player games), modeling them as reward-producing programs, and generating novel human-like goals through program synthesis. Reward-producing programs capture the rich semantics of goals through symbolic operations that compose, add temporal constraints, and allow for program execution on behavioral traces to evaluate progress. To build a generative model of goals, we learn a fitness function over the infinite set of possible goal programs and sample novel goals with a quality-diversity algorithm. Human evaluators found that model-generated goals, when sampled from partitions of program space occupied by human examples, were indistinguishable from human-created games. We also discovered that our model's internal fitness scores predict games that are evaluated as more fun to play and more human-like. △ Less

Submitted 10 September, 2024; v1 submitted 21 May, 2024; originally announced May 2024.

Comments: Project website and goal program viewer: https://exps.gureckislab.org/guydav/goal_programs_viewer/main/

(C$_5$H$_9$NH$_3$)$_2$CuBr$_4$: a metal-organic two-ladder quantum magnet

Authors: J. Philippe, F. Elson, M. P. N. Casati, S. Sanz, M. Metzelaars, O. Shliakhtun, O. K. Forslund, J. Lass, T. Shiroka, A. Linden, D. G. Mazzone, J. Ollivier, S. Shin, M. Medarde, B. Lake, M. Mansson, M. Bartkowiak, B. Normand, P. Kögerler, Y. Sassa, M. Janoschek, G. Simutis

Abstract: Low-dimensional quantum magnets are a versatile materials platform for studying the emergent many-body physics and collective excitations that can arise even in systems with only short-range interactions. Understanding their low-temperature structure and spin Hamiltonian is key to explaining their magnetic properties, including unconventional quantum phases, phase transitions, and excited states.… ▽ More Low-dimensional quantum magnets are a versatile materials platform for studying the emergent many-body physics and collective excitations that can arise even in systems with only short-range interactions. Understanding their low-temperature structure and spin Hamiltonian is key to explaining their magnetic properties, including unconventional quantum phases, phase transitions, and excited states. We study the metal-organic coordination compound (C$_5$H$_9$NH$_3$)$_2$CuBr$_4$ and its deuterated counterpart, which upon its discovery was identified as a candidate two-leg quantum ($S = 1/2$) spin ladder in the strong-leg coupling regime. By growing large single crystals and probing them with both bulk and microscopic techniques, we deduce that two previously unknown structural phase transitions take place between 136 K and 113 K. The low-temperature structure has a monoclinic unit cell giving rise to two inequivalent spin ladders. We further confirm the absence of long-range magnetic order down to 30 mK and discuss the implications of this two-ladder structure for the magnetic properties of (C$_5$H$_9$NH$_3$)$_2$CuBr$_4$. △ Less

Submitted 6 September, 2024; v1 submitted 12 April, 2024; originally announced April 2024.

Journal ref: Phys. Rev. B 110, 094101 (2024)

CoLLEGe: Concept Embedding Generation for Large Language Models

Authors: Ryan Teehan, Brenden Lake, Mengye Ren

Abstract: Current language models are unable to quickly learn new concepts on the fly, often requiring a more involved finetuning process to learn robustly. Prompting in-context is not robust to context distractions, and often fails to confer much information about the new concepts. Classic methods for few-shot word learning in NLP, relying on global word vectors, are less applicable to large language model… ▽ More Current language models are unable to quickly learn new concepts on the fly, often requiring a more involved finetuning process to learn robustly. Prompting in-context is not robust to context distractions, and often fails to confer much information about the new concepts. Classic methods for few-shot word learning in NLP, relying on global word vectors, are less applicable to large language models. In this paper, we introduce a novel approach named CoLLEGe (Concept Learning with Language Embedding Generation) to modernize few-shot concept learning. CoLLEGe is a meta-learning framework capable of generating flexible embeddings for new concepts using a small number of example sentences or definitions. Our primary meta-learning objective is simply to facilitate a language model to make next word predictions in forthcoming sentences, making it compatible with language model pretraining. We design a series of tasks to test new concept learning in challenging real-world scenarios, including new word acquisition, definition inference, and verbal reasoning, and demonstrate that our method succeeds in each setting without task-specific training. Code and data for our project can be found at https://college-concept-learning.github.io/ △ Less

Submitted 16 October, 2024; v1 submitted 22 March, 2024; originally announced March 2024.

Compositional learning of functions in humans and machines

Authors: Yanli Zhou, Brenden M. Lake, Adina Williams

Abstract: The ability to learn and compose functions is foundational to efficient learning and reasoning in humans, enabling flexible generalizations such as creating new dishes from known cooking processes. Beyond sequential chaining of functions, existing linguistics literature indicates that humans can grasp more complex compositions with interacting functions, where output production depends on context… ▽ More The ability to learn and compose functions is foundational to efficient learning and reasoning in humans, enabling flexible generalizations such as creating new dishes from known cooking processes. Beyond sequential chaining of functions, existing linguistics literature indicates that humans can grasp more complex compositions with interacting functions, where output production depends on context changes induced by different function orderings. Extending the investigation into the visual domain, we developed a function learning paradigm to explore the capacity of humans and neural network models in learning and reasoning with compositional functions under varied interaction conditions. Following brief training on individual functions, human participants were assessed on composing two learned functions, in ways covering four main interaction types, including instances in which the application of the first function creates or removes the context for applying the second function. Our findings indicate that humans can make zero-shot generalizations on novel visual function compositions across interaction conditions, demonstrating sensitivity to contextual changes. A comparison with a neural network model on the same task reveals that, through the meta-learning for compositionality (MLC) approach, a standard sequence-to-sequence Transformer can mimic human generalization patterns in composing functions. △ Less

Submitted 18 March, 2024; originally announced March 2024.

Comments: 7 pages, 6 figures

A systematic investigation of learnability from single child linguistic input

Authors: Yulu Qin, Wentao Wang, Brenden M. Lake

Abstract: Language models (LMs) have demonstrated remarkable proficiency in generating linguistically coherent text, sparking discussions about their relevance to understanding human language learnability. However, a significant gap exists between the training data for these models and the linguistic input a child receives. LMs are typically trained on data that is orders of magnitude larger and fundamental… ▽ More Language models (LMs) have demonstrated remarkable proficiency in generating linguistically coherent text, sparking discussions about their relevance to understanding human language learnability. However, a significant gap exists between the training data for these models and the linguistic input a child receives. LMs are typically trained on data that is orders of magnitude larger and fundamentally different from child-directed speech (Warstadt and Bowman, 2022; Warstadt et al., 2023; Frank, 2023a). Addressing this discrepancy, our research focuses on training LMs on subsets of a single child's linguistic input. Previously, Wang, Vong, Kim, and Lake (2023) found that LMs trained in this setting can form syntactic and semantic word clusters and develop sensitivity to certain linguistic phenomena, but they only considered LSTMs and simpler neural networks trained from just one single-child dataset. Here, to examine the robustness of learnability from single-child input, we systematically train six different model architectures on five datasets (3 single-child and 2 baselines). We find that the models trained on single-child datasets showed consistent results that matched with previous work, underscoring the robustness of forming meaningful syntactic and semantic representations from a subset of a child's linguistic input. △ Less

Submitted 10 May, 2024; v1 submitted 12 February, 2024; originally announced February 2024.

Comments: Please cite as Qin, Y., Wang, W., and Lake, B. M. (2024). A systematic investigation of learnability from single child linguistic input. In Proceedings of the 46th Annual Conference of the Cognitive Science Society

Comparing Abstraction in Humans and Large Language Models Using Multimodal Serial Reproduction

Authors: Sreejan Kumar, Raja Marjieh, Byron Zhang, Declan Campbell, Michael Y. Hu, Umang Bhatt, Brenden Lake, Thomas L. Griffiths

Abstract: Humans extract useful abstractions of the world from noisy sensory data. Serial reproduction allows us to study how people construe the world through a paradigm similar to the game of telephone, where one person observes a stimulus and reproduces it for the next to form a chain of reproductions. Past serial reproduction experiments typically employ a single sensory modality, but humans often commu… ▽ More Humans extract useful abstractions of the world from noisy sensory data. Serial reproduction allows us to study how people construe the world through a paradigm similar to the game of telephone, where one person observes a stimulus and reproduces it for the next to form a chain of reproductions. Past serial reproduction experiments typically employ a single sensory modality, but humans often communicate abstractions of the world to each other through language. To investigate the effect language on the formation of abstractions, we implement a novel multimodal serial reproduction framework by asking people who receive a visual stimulus to reproduce it in a linguistic format, and vice versa. We ran unimodal and multimodal chains with both humans and GPT-4 and find that adding language as a modality has a larger effect on human reproductions than GPT-4's. This suggests human visual and linguistic representations are more dissociable than those of GPT-4. △ Less

Submitted 5 February, 2024; originally announced February 2024.

Self-supervised learning of video representations from a child's perspective

Authors: A. Emin Orhan, Wentao Wang, Alex N. Wang, Mengye Ren, Brenden M. Lake

Abstract: Children learn powerful internal models of the world around them from a few years of egocentric visual experience. Can such internal models be learned from a child's visual experience with highly generic learning algorithms or do they require strong inductive biases? Recent advances in collecting large-scale, longitudinal, developmentally realistic video datasets and generic self-supervised learni… ▽ More Children learn powerful internal models of the world around them from a few years of egocentric visual experience. Can such internal models be learned from a child's visual experience with highly generic learning algorithms or do they require strong inductive biases? Recent advances in collecting large-scale, longitudinal, developmentally realistic video datasets and generic self-supervised learning (SSL) algorithms are allowing us to begin to tackle this nature vs. nurture question. However, existing work typically focuses on image-based SSL algorithms and visual capabilities that can be learned from static images (e.g. object recognition), thus ignoring temporal aspects of the world. To close this gap, here we train self-supervised video models on longitudinal, egocentric headcam recordings collected from a child over a two year period in their early development (6-31 months). The resulting models are highly effective at facilitating the learning of action concepts from a small number of labeled examples; they have favorable data size scaling properties; and they display emergent video interpolation capabilities. Video models also learn more accurate and more robust object representations than image-based models trained with the exact same data. These results suggest that important temporal aspects of a child's internal model of the world may be learnable from their visual experience using highly generic learning algorithms and without strong inductive biases. △ Less

Submitted 16 October, 2024; v1 submitted 31 January, 2024; originally announced February 2024.

Comments: v3 updates results with significantly improved models; v2 was published as a conference paper at CogSci 2024; code & models available from https://github.com/eminorhan/video-models

Spinon heat transport in the three-dimensional quantum magnet PbCuTe$_2$O$_6$

Authors: Xiaochen Hong, Matthias Gillig, Abanoub R. N. Hanna, Shravani Chillal, A. T. M. Nazmul Islam, Bella Lake, Bernd Büchner, Christian Hess

Abstract: Quantum spin liquids (QSL) are novel phases of matter which remain quantum disordered even at the lowest temperature. They are characterized by emergent gauge fields and fractionalized quasiparticles. Here we show that the sub-Kelvin thermal transport of the three-dimensional $S=1/2$ hyper-hyperkagome quantum magnet PbCuTe$_2$O$_6$ is governed by a sizeable charge-neutral fermionic contribution wh… ▽ More Quantum spin liquids (QSL) are novel phases of matter which remain quantum disordered even at the lowest temperature. They are characterized by emergent gauge fields and fractionalized quasiparticles. Here we show that the sub-Kelvin thermal transport of the three-dimensional $S=1/2$ hyper-hyperkagome quantum magnet PbCuTe$_2$O$_6$ is governed by a sizeable charge-neutral fermionic contribution which is compatible with the itinerant fractionalized excitations of a spinon Fermi surface. We demonstrate that this hallmark feature of the QSL state is remarkably robust against sample crystallinity, large magnetic field, and field-induced magnetic order, ruling out the imitation of QSL features by extrinsic effects. Our findings thus reveal the characteristic low-energy features of PbCuTe$_2$O$_6$ which qualify this compound as a true QSL material. △ Less

Submitted 20 November, 2023; originally announced November 2023.

Deep Neural Networks Can Learn Generalizable Same-Different Visual Relations

Authors: Alexa R. Tartaglini, Sheridan Feucht, Michael A. Lepori, Wai Keen Vong, Charles Lovering, Brenden M. Lake, Ellie Pavlick

Abstract: Although deep neural networks can achieve human-level performance on many object recognition benchmarks, prior work suggests that these same models fail to learn simple abstract relations, such as determining whether two objects are the same or different. Much of this prior work focuses on training convolutional neural networks to classify images of two same or two different abstract shapes, testi… ▽ More Although deep neural networks can achieve human-level performance on many object recognition benchmarks, prior work suggests that these same models fail to learn simple abstract relations, such as determining whether two objects are the same or different. Much of this prior work focuses on training convolutional neural networks to classify images of two same or two different abstract shapes, testing generalization on within-distribution stimuli. In this article, we comprehensively study whether deep neural networks can acquire and generalize same-different relations both within and out-of-distribution using a variety of architectures, forms of pretraining, and fine-tuning datasets. We find that certain pretrained transformers can learn a same-different relation that generalizes with near perfect accuracy to out-of-distribution stimuli. Furthermore, we find that fine-tuning on abstract shapes that lack texture or color provides the strongest out-of-distribution generalization. Our results suggest that, with the right approach, deep neural networks can learn generalizable same-different visual relations. △ Less

Submitted 14 October, 2023; originally announced October 2023.

Magnetic structure and phase diagram of the Heisenberg-Ising spin chain antiferromagnetic PbCo$_{2}$V$_{2}$O$_{8}$

Authors: K. Puzniak, C. Aguilar-Maldonado, R. Feyerherm, K. Prokeš, A. T. M. N. Islam, Y. Skourski, L. Keller, B. Lake

Abstract: The effective spin-1/2 antiferromagnetic Heisenberg-Ising chain materials, ACo$_2$V$_2$O$_8$, A = Sr, Ba, are a rich source of exotic fundamental phenomena and have been investigated for their model magnetic properties both in zero and non-zero magnetic fields. Here we investigate a new member of the family, namely PbCo$_2$V$_2$O$_8$. We synthesize powder and single crystal samples of PbCo$_2$V… ▽ More The effective spin-1/2 antiferromagnetic Heisenberg-Ising chain materials, ACo$_2$V$_2$O$_8$, A = Sr, Ba, are a rich source of exotic fundamental phenomena and have been investigated for their model magnetic properties both in zero and non-zero magnetic fields. Here we investigate a new member of the family, namely PbCo$_2$V$_2$O$_8$. We synthesize powder and single crystal samples of PbCo$_2$V$_2$O$_8$ and determine its magnetic structure using neutron diffraction. Furthermore, the magnetic field/temperature phase diagrams for magnetic field applied along the c, a, and [110] crystallographic directions in the tetragonal unit cell are determined via magnetization and heat capacity measurements. A complex series of phases and quantum phase transitions are discovered that depend strongly on both the magnitude and direction of the field. Our results show that \pcvo is an effective spin-1/2 antiferromagnetic Heisenberg-Ising chain with properties that are in general comparable to those of SrCo$_2$V$_2$O$_8$ and BaCo$_2$V$_2$O$_8$. One interesting departure from the results of these related compounds, is however, the discovery of a new field-induced phase for the field direction $H\|$[110] which has not been previously observed. △ Less

Submitted 28 September, 2023; originally announced September 2023.

Spin dynamics of the $E_8$ particles

Authors: Xiao Wang, Konrad Puzniak, Karin Schmalzl, C. Balz, M. Matsuda, Akira Okutani, M. Hagiwara, Jie Ma, Jianda Wu, Bella Lake

Abstract: In this article, we report on inelastic neutron scattering measurements on a quasi-1D antiferromagnet BaCo$_2$V$_2$O$_8$ under a transverse magnetic field applied along the (0,1,0) direction. Combining results of inelastic neutron scattering experiments, analytical analysis, and numerical simulations, we precisely studied the $E_8$ excitations appearing in the whole Brillouin zone at… ▽ More In this article, we report on inelastic neutron scattering measurements on a quasi-1D antiferromagnet BaCo$_2$V$_2$O$_8$ under a transverse magnetic field applied along the (0,1,0) direction. Combining results of inelastic neutron scattering experiments, analytical analysis, and numerical simulations, we precisely studied the $E_8$ excitations appearing in the whole Brillouin zone at $B_c^{1D}\approx 4.7$ T. The energy scan at $Q=(0,0,2)$ reveals a match between the data and the theoretical prediction of energies of multiple $E_8$ excitations. Furthermore, dispersions of the lightest three $E_8$ particles have been clearly observed, confirming the existence of the $E_8$ particles in BaCo$_2$V$_2$O$_8$. Our results lay down a concrete ground to systematically study the physics of the exotic $E_8$ particles. △ Less

Submitted 31 July, 2023; originally announced August 2023.

Comments: 10 pages, 4 figures

Journal ref: Science Bulletin (2024)

Classical spin models of the windmill lattice and their relevance for PbCuTe$_2$O$_6$

Authors: Anna Fancelli, Johannes Reuther, Bella Lake

Abstract: We investigate classical Heisenberg models on the distorted windmill lattice and discuss their applicability to the spin-$1/2$ spin liquid candidate PbCuTe$_2$O$_6$. We first consider a general Heisenberg model on this lattice with antiferromagnetic interactions $J_n$ ($n=1,2,3,4$) up to fourth neighbors. Setting $J_1=J_2$ (as approximately realized in PbCuTe$_2$O$_6$) we map out the classical gro… ▽ More We investigate classical Heisenberg models on the distorted windmill lattice and discuss their applicability to the spin-$1/2$ spin liquid candidate PbCuTe$_2$O$_6$. We first consider a general Heisenberg model on this lattice with antiferromagnetic interactions $J_n$ ($n=1,2,3,4$) up to fourth neighbors. Setting $J_1=J_2$ (as approximately realized in PbCuTe$_2$O$_6$) we map out the classical ground state phase diagram in the remaining parameter space and identify a competition between $J_3$ and $J_4$ that opens up interesting magnetic scenarios. Particularly, these couplings tune the ground states from coplanar commensurate or non-coplanar incommensurate magnetically ordered states to highly degenerate ground state manifolds with subextensive or extensive degeneracies. In the latter case, we uncover an unusual classical spin liquid defined on a lattice of corner sharing octahedra. We then focus on the particular set of interaction parameters $J_n$ that has previously been proposed for PbCuTe$_2$O$_6$ and investigate the system's incommensurate magnetic ground state order and finite temperature multistage ordering mechanism. We perform extensive finite temperature simulations of the system's dynamical spin structure factor and compare it with published neutron scattering data for PbCuTe$_2$O$_6$ at low temperatures. Our results demonstrate that thermal fluctuations in the classical model can largely explain the signal distribution in the measured spin structure factor but we also identify distinct differences. Our investigations make use of a variety of different analytical and numerical approaches for classical spin systems, such as Luttinger-Tisza, classical Monte Carlo, iterative minimization, and molecular dynamics simulations. △ Less

Submitted 15 March, 2024; v1 submitted 20 June, 2023; originally announced June 2023.

Comments: 17 pages, 9 figures

Compositional diversity in visual concept learning

Authors: Yanli Zhou, Reuben Feinman, Brenden M. Lake

Abstract: Humans leverage compositionality to efficiently learn new concepts, understanding how familiar parts can combine together to form novel objects. In contrast, popular computer vision models struggle to make the same types of inferences, requiring more data and generalizing less flexibly than people do. Here, we study these distinctively human abilities across a range of different types of visual co… ▽ More Humans leverage compositionality to efficiently learn new concepts, understanding how familiar parts can combine together to form novel objects. In contrast, popular computer vision models struggle to make the same types of inferences, requiring more data and generalizing less flexibly than people do. Here, we study these distinctively human abilities across a range of different types of visual composition, examining how people classify and generate ``alien figures'' with rich relational structure. We also develop a Bayesian program induction model which searches for the best programs for generating the candidate visual figures, utilizing a large program space containing different compositional mechanisms and abstractions. In few shot classification tasks, we find that people and the program induction model can make a range of meaningful compositional generalizations, with the model providing a strong account of the experimental data as well as interpretable parameters that reveal human assumptions about the factors invariant to category membership (here, to rotation and changing part attachment). In few shot generation tasks, both people and the models are able to construct compelling novel examples, with people behaving in additional structured ways beyond the model capabilities, e.g. making choices that complete a set or reconfiguring existing parts in highly novel ways. To capture these additional behavioral patterns, we develop an alternative model based on neuro-symbolic program induction: this model also composes new concepts from existing parts yet, distinctively, it utilizes neural network modules to successfully capture residual statistical structure. Together, our behavioral and computational findings show how people and models can produce a rich variety of compositional behavior when classifying and generating visual objects. △ Less

Submitted 30 May, 2023; originally announced May 2023.

Comments: 40 pages, 23 figures

Learning high-level visual representations from a child's perspective without strong inductive biases

Authors: A. Emin Orhan, Brenden M. Lake

Abstract: Young children develop sophisticated internal models of the world based on their visual experience. Can such models be learned from a child's visual experience without strong inductive biases? To investigate this, we train state-of-the-art neural networks on a realistic proxy of a child's visual experience without any explicit supervision or domain-specific inductive biases. Specifically, we train… ▽ More Young children develop sophisticated internal models of the world based on their visual experience. Can such models be learned from a child's visual experience without strong inductive biases? To investigate this, we train state-of-the-art neural networks on a realistic proxy of a child's visual experience without any explicit supervision or domain-specific inductive biases. Specifically, we train both embedding models and generative models on 200 hours of headcam video from a single child collected over two years and comprehensively evaluate their performance in downstream tasks using various reference models as yardsticks. On average, the best embedding models perform at a respectable 70% of a high-performance ImageNet-trained model, despite substantial differences in training data. They also learn broad semantic categories and object localization capabilities without explicit supervision, but they are less object-centric than models trained on all of ImageNet. Generative models trained with the same data successfully extrapolate simple properties of partially masked objects, like their rough outline, texture, color, or orientation, but struggle with finer object details. We replicate our experiments with two other children and find remarkably consistent results. Broadly useful high-level visual representations are thus robustly learnable from a representative sample of a child's visual experience without strong inductive biases. △ Less

Submitted 22 September, 2023; v1 submitted 24 May, 2023; originally announced May 2023.

Comments: 32 pages, 19 figures, 3 tables; code & all pretrained models available from https://github.com/eminorhan/silicon-menagerie

Magnetic excitation spectrum and Hamiltonian of the quantum spin chain BaCuTe2O6

Authors: A. Samartzis, S. Chillal, H. O. Jeschke, D. J. Voneshen, Z. Lu, A. T. M. N. Islam, B. Lake

Abstract: The magnetic excitation spectrum and Hamiltonian of the quantum magnet BaCuTe2O6 is studied by inelastic neutron scattering (INS) and density functional theory (DFT). INS on powder and single crystal samples reveals overlapping spinon continuua - the spectrum of an antiferromagnetic spin-1/2 spin chain - due to equivalent chains running along the a, b, and c directions. Long-range magnetic order o… ▽ More The magnetic excitation spectrum and Hamiltonian of the quantum magnet BaCuTe2O6 is studied by inelastic neutron scattering (INS) and density functional theory (DFT). INS on powder and single crystal samples reveals overlapping spinon continuua - the spectrum of an antiferromagnetic spin-1/2 spin chain - due to equivalent chains running along the a, b, and c directions. Long-range magnetic order onsets below TN = 6.3 K due to interchain interactions, and is accompanied by the emergence of sharp spin-wave excitations which replace the continuua at low energies. The spin-wave spectrum is highly complex and was successfully modelled achieving excellent agreement with the data. The extracted interactions reveal an intrachain interaction, J3 = 2.9 meV, while the antiferromagnetic hyperkagome interaction J2, is the sub-leading interaction responsible for coupling the chains together in a frustrated way. DFT calculations reveal a similar picture for BaCuTe2O6 of dominant J3 and sub-leading J2 antiferromagnetic interactions and also indicate a high sensitivity of the interactions to small changes of structure which could explain the very different Hamiltonians observed in the sister compounds SrCuTe2O6 and PbCuTe2O6. △ Less

Submitted 5 May, 2023; originally announced May 2023.

Comments: 12 pages, 8 figures

Journal ref: Physical Review B 107, 184435 (2023)

Analysis of COVID-19 first wave in the US based on demographic, mobility, and environmental variables

Authors: Dario Spiller, Gabriele Santin, Alessandro Sebastianelli, Lorenzo Lucchini, Riccardo Gallotti, Brennan Lake, Silvia Liberata Ullo, Bertrand Le Saux, Bruno Lepri

Abstract: COVID-19 had a strong and disruptive impact on our society, and yet further analyses on most relevant factors explaining the spread of the pandemic are needed. Interdisciplinary studies linking epidemiological, mobility, environmental, and socio-demographic data analysis can help understanding how historical conditions, concurrent social policies and environmental factors impacted on the evolution… ▽ More COVID-19 had a strong and disruptive impact on our society, and yet further analyses on most relevant factors explaining the spread of the pandemic are needed. Interdisciplinary studies linking epidemiological, mobility, environmental, and socio-demographic data analysis can help understanding how historical conditions, concurrent social policies and environmental factors impacted on the evolution of the pandemic crisis. This work deals with a regression analysis linking COVID-19 mortality to socio-demographic, mobility, and environmental data in the US during the first half of 2020, i.e., during the COVID-19 pandemic first wave. This study can provide very useful insights about risk factors enhancing mortality rates before non-pharmaceutical interventions or vaccination campaigns took place. Our cross-sectional ecological regression analysis demonstrates that, when considering the entire US area, the socio-demographic variables globally play the most important role with respect to environmental and mobility variables in describing COVID-19 mortality. Compared to the complete generalized linear model considering all socio-demographic, mobility, and environmental data, the regression based only on socio-demographic data provides a better approximation and proves to be a better explanatory model when compared to the mobility-based and environmental-based models. However, when looking at single entries within each of the three groups, we see that the mobility data can become relevant descriptive predictors at local scale, as in New Jersey where the time spent at work is one of the most relevant explanatory variables, while environmental data play contradictory roles. △ Less

Submitted 22 February, 2023; originally announced February 2023.

Comments: Submitted to the Scientific Reports, COVID-19 Collection, for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible

Field-induced effects in the spin liquid candidate PbCuTe$_{2}$O$_{6}$

Authors: Paul Eibisch, Christian Thurn, Arif Ata, Ulrich Tutsch, Yohei Saito, Steffi Hartmann, Bernd Wolf, Abanoub R. N. Hanna, A. T. M. Nazmul Islam, Shravani Chillal, Bella Lake, Michael Lang

Abstract: PbCuTe$_2$O$_6$ is considered as one of the rare candidate materials for a three-dimensional quantum spin liquid (QSL). This assessment was based on the results of various magnetic experiments, performed mainly on polycrystalline material. More recent measurements on single crystals revealed an even more exotic behavior, yielding ferroelectric order below $T_{\text{FE}}\approx 1\,\text{K}$, accomp… ▽ More PbCuTe$_2$O$_6$ is considered as one of the rare candidate materials for a three-dimensional quantum spin liquid (QSL). This assessment was based on the results of various magnetic experiments, performed mainly on polycrystalline material. More recent measurements on single crystals revealed an even more exotic behavior, yielding ferroelectric order below $T_{\text{FE}}\approx 1\,\text{K}$, accompanied by distinct lattice distortions, and a somewhat modified magnetic response which is still consistent with a QSL. Here we report on low-temperature measurements of various thermodynamic, magnetic and dielectric properties of single crystalline PbCuTe$_2$O$_6$ in magnetic fields $B\leq 14.5\,\text{T}$. The combination of these various probes allows us to construct a detailed $B$-$T$ phase diagram including a ferroelectric phase for $B \leq$ $8\,\text{T}$ and a $B$-induced magnetic phase at $B \geq$ $11\,\text{T}$. These phases are preceded by or coincide with a structural transition from a cubic high-temperature phase into a distorted non-cubic low-temperature state. The phase diagram discloses two quantum critical points (QCPs) in the accessible field range, a ferroelectric QCP at $B_{c1}$ = $7.9\,\text{T}$ and a magnetic QCP at $B_{c2}$ = $11\,\text{T}$. Field-induced lattice distortions, observed in the state at $T>$ $1\,\text{K}$ and which are assigned to the effect of spin-orbit interaction of the Cu$^{2+}$-ions, are considered as the key mechanism by which the magnetic field couples to the dielectric degrees of freedom in this material. △ Less

Submitted 9 May, 2023; v1 submitted 27 February, 2023; originally announced February 2023.

Multiple-level Point Embedding for Solving Human Trajectory Imputation with Prediction

Authors: Kyle K. Qin, Yongli Ren, Wei Shao, Brennan Lake, Filippo Privitera, Flora D. Salim

Abstract: Sparsity is a common issue in many trajectory datasets, including human mobility data. This issue frequently brings more difficulty to relevant learning tasks, such as trajectory imputation and prediction. Nowadays, little existing work simultaneously deals with imputation and prediction on human trajectories. This work plans to explore whether the learning process of imputation and prediction cou… ▽ More Sparsity is a common issue in many trajectory datasets, including human mobility data. This issue frequently brings more difficulty to relevant learning tasks, such as trajectory imputation and prediction. Nowadays, little existing work simultaneously deals with imputation and prediction on human trajectories. This work plans to explore whether the learning process of imputation and prediction could benefit from each other to achieve better outcomes. And the question will be answered by studying the coexistence patterns between missing points and observed ones in incomplete trajectories. More specifically, the proposed model develops an imputation component based on the self-attention mechanism to capture the coexistence patterns between observations and missing points among encoder-decoder layers. Meanwhile, a recurrent unit is integrated to extract the sequential embeddings from newly imputed sequences for predicting the following location. Furthermore, a new implementation called Imputation Cycle is introduced to enable gradual imputation with prediction enhancement at multiple levels, which helps to accelerate the speed of convergence. The experimental results on three different real-world mobility datasets show that the proposed approach has significant advantages over the competitive baselines across both imputation and prediction tasks in terms of accuracy and stability. △ Less

Submitted 12 January, 2023; v1 submitted 11 January, 2023; originally announced January 2023.

Comments: 22 pages; accepted by ACM Transactions on Spatial Algorithms and Systems

MSC Class: 68T07 ACM Class: H.0

Characterizing collective physical distancing in the U.S. during the first nine months of the COVID-19 pandemic

Authors: Brennan Klein, Timothy LaRock, Stefan McCabe, Leo Torres, Lisa Friedland, Maciej Kos, Filippo Privitera, Brennan Lake, Moritz U. G. Kraemer, John S. Brownstein, Richard Gonzalez, David Lazer, Tina Eliassi-Rad, Samuel V. Scarpino, Alessandro Vespignani, Matteo Chinazzi

Abstract: The COVID-19 pandemic offers an unprecedented natural experiment providing insights into the emergence of collective behavioral changes of both exogenous (government mandated) and endogenous (spontaneous reaction to infection risks) origin. Here, we characterize collective physical distancing -- mobility reductions, minimization of contacts, shortening of contact duration -- in response to the COV… ▽ More The COVID-19 pandemic offers an unprecedented natural experiment providing insights into the emergence of collective behavioral changes of both exogenous (government mandated) and endogenous (spontaneous reaction to infection risks) origin. Here, we characterize collective physical distancing -- mobility reductions, minimization of contacts, shortening of contact duration -- in response to the COVID-19 pandemic in the pre-vaccine era by analyzing de-identified, privacy-preserving location data for a panel of over 5.5 million anonymized, opted-in U.S. devices. We define five indicators of users' mobility and proximity to investigate how the emerging collective behavior deviates from the typical pre-pandemic patterns during the first nine months of the COVID-19 pandemic. We analyze both the dramatic changes due to the government mandated mitigation policies and the more spontaneous societal adaptation into a new (physically distanced) normal in the fall 2020. The indicators defined here allow the quantification of behavior changes across the rural/urban divide and highlight the statistical association of mobility and proximity indicators with metrics characterizing the pandemic's social and public health impact such as unemployment and deaths. This study provides a framework to study massive social distancing phenomena with potential uses in analyzing and monitoring the effects of pandemic mitigation plans at the national and international level. △ Less

Submitted 17 December, 2022; originally announced December 2022.

Finite temperature tensor network algorithm for frustrated two-dimensional quantum materials

Authors: Philipp Schmoll, Christian Balz, Bella Lake, Jens Eisert, Augustine Kshetrimayum

Abstract: Aimed at a more realistic classical description of natural quantum systems, we present a two-dimensional tensor network algorithm to study finite temperature properties of frustrated model quantum systems and real quantum materials. For this purpose, we introduce the infinite projected entangled simplex operator ansatz to study thermodynamic properties. To obtain state-of-the-art benchmarking resu… ▽ More Aimed at a more realistic classical description of natural quantum systems, we present a two-dimensional tensor network algorithm to study finite temperature properties of frustrated model quantum systems and real quantum materials. For this purpose, we introduce the infinite projected entangled simplex operator ansatz to study thermodynamic properties. To obtain state-of-the-art benchmarking results, we explore the highly challenging spin-1/2 Heisenberg anti-ferromagnet on the Kagome lattice, a system for which we investigate the melting of the magnetization plateaus at finite magnetic field and temperature. Making close connection to actual experimental data of real quantum materials, we go on to studying the finite temperature properties of Ca$_{10}$Cr$_7$O$_{28}$. We compare the magnetization curve of this material in the presence of an external magnetic field at finite temperature with classically simulated data. As a first theoretical tool that incorporates both thermal fluctuations as well as quantum correlations in the study of this material, our work contributes to settling the existing controversy between the experimental data and previous theoretical works on the magnetization process. △ Less

Submitted 31 October, 2022; originally announced November 2022.

Comments: 9 pages, 11 figures

Journal ref: Phys. Rev. B 109, 235119 (2024)

Pinch points and half-moons in dipolar-octupolar Nd$_2$Hf$_2$O$_7$

Authors: A. Samartzis, J. Xu, V. K. Anand, A. T. M. N. Islam, J. Ollivier, Y. Su, B. Lake

Abstract: While it is established that the pinch point scattering pattern in spin ice arises from an emergent coulomb phase associated with magnetic moment that is divergence-free, more complex Hamiltonians can introduce a divergence-full part. If these two parts remain decoupled, they give rise to the co-existence of distinct features. Here we show that the moment in ${\rm Nd_2Hf_2O_7}$ forms a static long… ▽ More While it is established that the pinch point scattering pattern in spin ice arises from an emergent coulomb phase associated with magnetic moment that is divergence-free, more complex Hamiltonians can introduce a divergence-full part. If these two parts remain decoupled, they give rise to the co-existence of distinct features. Here we show that the moment in ${\rm Nd_2Hf_2O_7}$ forms a static long-range ordered ground state, a flat, gapped pinch point excitation and dispersive excitations. These results confirm recent theories which predict that the dispersive modes, which arise from the divergence-full moment, host a pinch point pattern of their own, observed experimentally as `half-moons'. △ Less

Submitted 8 September, 2022; v1 submitted 25 August, 2022; originally announced August 2022.

Comments: 8 pages, 5 figures

Journal ref: Physical Review B Letters 106, L100401 (2022)

A route towards engineering many-body localization in real materials

Authors: A. Nietner, A. Kshetrimayum, J. Eisert, B. Lake

Abstract: The interplay of interactions and disorder in a quantum many body system may lead to the elusive phenomenon of many body localization (MBL). It has been observed under precisely controlled conditions in synthetic quantum many-body systems, but to detect it in actual quantum materials seems challenging. In this work, we present a path to synthesize real materials that show signatures of many body l… ▽ More The interplay of interactions and disorder in a quantum many body system may lead to the elusive phenomenon of many body localization (MBL). It has been observed under precisely controlled conditions in synthetic quantum many-body systems, but to detect it in actual quantum materials seems challenging. In this work, we present a path to synthesize real materials that show signatures of many body localization by mixing different species of materials in the laboratory. To provide evidence for the functioning of our approach, we perform a detailed tensor-network based numerical analysis to study the effects of various doping ratios of the constituting materials. Moreover, in order to provide guidance to experiments, we investigate different choices of actual candidate materials. To address the challenge of how to achieve stability under heating, we study the effect of the electron-phonon coupling, focusing on effectively one dimensional materials embedded in one, two and three dimensional lattices. We analyze how this coupling affects the MBL and provide an intuitive microscopic description of the interplay between the electronic degrees of freedom and the lattice vibrations. Our work provides a guideline for the necessary conditions on the properties of the ingredient materials and, as such, serves as a road map to experimentally synthesizing real quantum materials exhibiting signatures of MBL. △ Less

Submitted 21 July, 2022; originally announced July 2022.

Comments: 12 pages, 7 figures

Improving Systematic Generalization Through Modularity and Augmentation

Authors: Laura Ruis, Brenden Lake

Abstract: Systematic generalization is the ability to combine known parts into novel meaning; an important aspect of efficient human learning, but a weakness of neural network learning. In this work, we investigate how two well-known modeling principles -- modularity and data augmentation -- affect systematic generalization of neural networks in grounded language learning. We analyze how large the vocabular… ▽ More Systematic generalization is the ability to combine known parts into novel meaning; an important aspect of efficient human learning, but a weakness of neural network learning. In this work, we investigate how two well-known modeling principles -- modularity and data augmentation -- affect systematic generalization of neural networks in grounded language learning. We analyze how large the vocabulary needs to be to achieve systematic generalization and how similar the augmented data needs to be to the problem at hand. Our findings show that even in the controlled setting of a synthetic benchmark, achieving systematic generalization remains very difficult. After training on an augmented dataset with almost forty times more adverbs than the original problem, a non-modular baseline is not able to systematically generalize to a novel combination of a known verb and adverb. When separating the task into cognitive processes like perception and navigation, a modular neural network is able to utilize the augmented data and generalize more systematically, achieving 70% and 40% exact match increase over state-of-the-art on two gSCAN tests that have not previously been improved. We hope that this work gives insight into the drivers of systematic generalization, and what we still need to improve for neural networks to learn more like humans do. △ Less

Submitted 22 February, 2022; originally announced February 2022.

A Developmentally-Inspired Examination of Shape versus Texture Bias in Machines

Authors: Alexa R. Tartaglini, Wai Keen Vong, Brenden M. Lake

Abstract: Early in development, children learn to extend novel category labels to objects with the same shape, a phenomenon known as the shape bias. Inspired by these findings, Geirhos et al. (2019) examined whether deep neural networks show a shape or texture bias by constructing images with conflicting shape and texture cues. They found that convolutional neural networks strongly preferred to classify fam… ▽ More Early in development, children learn to extend novel category labels to objects with the same shape, a phenomenon known as the shape bias. Inspired by these findings, Geirhos et al. (2019) examined whether deep neural networks show a shape or texture bias by constructing images with conflicting shape and texture cues. They found that convolutional neural networks strongly preferred to classify familiar objects based on texture as opposed to shape, suggesting a texture bias. However, there are a number of differences between how the networks were tested in this study versus how children are typically tested. In this work, we re-examine the inductive biases of neural networks by adapting the stimuli and procedure from Geirhos et al. (2019) to more closely follow the developmental paradigm and test on a wide range of pre-trained neural networks. Across three experiments, we find that deep neural networks exhibit a preference for shape rather than texture when tested under conditions that more closely replicate the developmental procedure. △ Less

Submitted 17 May, 2022; v1 submitted 16 February, 2022; originally announced February 2022.

Comments: 7 pages, 4 figures

Quantum wake dynamics in Heisenberg antiferromagnetic chains

Authors: Allen Scheie, Pontus Laurell, Bella Lake, Stephen E. Nagler, Matthew B. Stone, Jean-Sebastian Caux, D. Alan Tennant

Abstract: Traditional spectroscopy, by its very nature, characterizes properties of physical systems in the momentum and frequency domains. The most interesting and potentially practically useful quantum many-body effects however emerge from the deep composition of local, short-time correlations. Here, using inelastic neutron scattering and methods of integrability, we experimentally observe and theoretical… ▽ More Traditional spectroscopy, by its very nature, characterizes properties of physical systems in the momentum and frequency domains. The most interesting and potentially practically useful quantum many-body effects however emerge from the deep composition of local, short-time correlations. Here, using inelastic neutron scattering and methods of integrability, we experimentally observe and theoretically describe a local, coherent, long-lived, quasiperiodically oscillating magnetic state emerging out of the distillation of propagating excitations following a local quantum quench in a Heisenberg antiferromagnetic chain. This "quantum wake" displays similarities to Floquet states, discrete time crystals and nonlinear Luttinger liquids. △ Less

Submitted 18 January, 2022; v1 submitted 10 January, 2022; originally announced January 2022.

Comments: 7 pages 4 figures, with 4 pages supplemental information

Journal ref: Nature Communications 13, 5796 (2022)

Crystal growth, characterization and phase transition of PbCuTe$_2$O$_6$

Authors: A. R. N. Hanna, A. T. M. N. Islam, R. Feyerherm, K. Siemensmeyer, K. Karmakar, S. Chillal, B. Lake

Abstract: Single crystals of the three-dimensional frustrated magnet and spin liquid candidate compound PbCuTe$_2$O$_6$ were grown using both the Travelling Solvent Floating Zone (TSFZ) and the Top-Seeded Solution Growth (TSSG) techniques. The growth conditions were optimized by investigating the thermal properties. The quality of the crystals was checked by polarized optical microscopy, X-ray Laue and X-ra… ▽ More Single crystals of the three-dimensional frustrated magnet and spin liquid candidate compound PbCuTe$_2$O$_6$ were grown using both the Travelling Solvent Floating Zone (TSFZ) and the Top-Seeded Solution Growth (TSSG) techniques. The growth conditions were optimized by investigating the thermal properties. The quality of the crystals was checked by polarized optical microscopy, X-ray Laue and X-ray powder diffraction, and compared to the polycrystalline samples. Excellent quality crystals were obtained by the TSSG method. Magnetic measurements of these crystals revealed a small anisotropy for different crystallographic directions in comparison with the previously reported data. The heat capacity of both single crystal and powder samples reveal a transition anomaly around 1 K. Curiously the position and magnitude of the transition are strongly dependent on the crystallite size and it is almost entirely absent for the smallest crystallites. A structural transition is suggested which accompanies the reported ferroelectric transition, and a scenario whereby it becomes energetically unfavourable in small crystallites is proposed. △ Less

Submitted 28 September, 2021; v1 submitted 15 July, 2021; originally announced July 2021.

Comments: 10 pages, 8 figures

Weak three-dimensional coupling of Heisenberg quantum spin chains in SrCuTe$_{2}$O$_{6}$

Authors: S. Chillal, A. T. M. N. Islam, P. Steffens, R. Bewley, B. Lake

Abstract: The magnetic Hamiltonian of the Heisenberg quantum antiferromagnet SrCuTe$_{2}$O$_{6}$ is studied by inelastic neutron scattering technique on powder and single crystalline samples above and below the magnetic transition temperatures at 8 K and 2 K. The high temperature spectra reveal a characteristic diffuse scattering corresponding to a multi-spinon continuum confirming the dominant quantum spin… ▽ More The magnetic Hamiltonian of the Heisenberg quantum antiferromagnet SrCuTe$_{2}$O$_{6}$ is studied by inelastic neutron scattering technique on powder and single crystalline samples above and below the magnetic transition temperatures at 8 K and 2 K. The high temperature spectra reveal a characteristic diffuse scattering corresponding to a multi-spinon continuum confirming the dominant quantum spin-chain behavior due to the third neighbour interaction J$_{intra}$ = 4.22 meV (49 K). The low temperature spectra exhibits sharper excitations at energies below 1.25 meV which can be explained by considering a combination of weak antiferromagnetic first nearest neighbour interchain coupling J$_1$ = 0.17 meV (1.9 K) and even weaker ferromagnetic second nearest neighbour J$_2$ = -0.037 meV (-0.4 K) or a weak ferromagnetic J$_2$ = -0.11 meV (-1.3 K) and antiferromagnetic J$_6$ = 0.16 meV (1.85 K) giving rise to the long-range magnetic order and spin-wave excitations at low energies. These results suggest that SrCuTe$_{2}$O$_{6}$ is a highly one-dimensional Heisenberg system with three mutually perpendicular spin-chains coupled by a weak ferromagnetic J$_2$ in addition to the antiferromagnetic J$_1$ or J$_6$ presenting a contrasting scenario from the highly frustrated hyper-hyperkagome lattice (equally strong antiferromagnetic J$_1$ and J$_2$) found in the iso-structural PbCuTe$_{2}$O$_{6}$. △ Less

Submitted 16 July, 2021; v1 submitted 12 July, 2021; originally announced July 2021.

Comments: 9 pages, 5+2 figures

Journal ref: Phys. Rev. B 104, 144402 (2021)

Improving Coherence and Consistency in Neural Sequence Models with Dual-System, Neuro-Symbolic Reasoning

Authors: Maxwell Nye, Michael Henry Tessler, Joshua B. Tenenbaum, Brenden M. Lake

Abstract: Human reasoning can often be understood as an interplay between two systems: the intuitive and associative ("System 1") and the deliberative and logical ("System 2"). Neural sequence models -- which have been increasingly successful at performing complex, structured tasks -- exhibit the advantages and failure modes of System 1: they are fast and learn patterns from data, but are often inconsistent… ▽ More Human reasoning can often be understood as an interplay between two systems: the intuitive and associative ("System 1") and the deliberative and logical ("System 2"). Neural sequence models -- which have been increasingly successful at performing complex, structured tasks -- exhibit the advantages and failure modes of System 1: they are fast and learn patterns from data, but are often inconsistent and incoherent. In this work, we seek a lightweight, training-free means of improving existing System 1-like sequence models by adding System 2-inspired logical reasoning. We explore several variations on this theme in which candidate generations from a neural sequence model are examined for logical consistency by a symbolic reasoning module, which can either accept or reject the generations. Our approach uses neural inference to mediate between the neural System 1 and the logical System 2. Results in robust story generation and grounded instruction-following show that this approach can increase the coherence and accuracy of neurally-based generations. △ Less

Submitted 15 December, 2021; v1 submitted 6 July, 2021; originally announced July 2021.

Comments: NeurIPS 2021

Non-Abelian statistics in light scattering processes across interacting Haldane chains

Authors: Vladimir Gnezdilov, Vladimir Kurnosov, Yurii Pashkevich, Anup Kumar Bera, A. T. M. Nazmul Islam, Bella Lake, Bodo Lobbenmeier, Dirk Wulferding, Peter Lemmens

Abstract: The $S=1$ Haldane state is constructed from a product of local singlet dimers in the bulk and topological states at the edges of a chain. It is a fundamental representative of topological quantum matter. Its well-known representative, the quasi-one-dimensional SrNi$_2$V$_2$O$_8$ shows both conventional as well as unconventional magnetic Raman scattering. The former is observed as one- and two-trip… ▽ More The $S=1$ Haldane state is constructed from a product of local singlet dimers in the bulk and topological states at the edges of a chain. It is a fundamental representative of topological quantum matter. Its well-known representative, the quasi-one-dimensional SrNi$_2$V$_2$O$_8$ shows both conventional as well as unconventional magnetic Raman scattering. The former is observed as one- and two-triplet excitations with small linewidths and energies corresponding to the Haldane gap $Δ_H$ and the exchange coupling $J_c$ along the chain, respectively. Well-defined magnetic quasiparticles are assumed to be stabilized by interchain interactions and uniaxial single-ion anisotropy. Unconventional scattering exists as broad continua of scattering with an intensity $I(T)$ that shows a mixed bosonic / fermionic statistic. Such a mixed statistic has also been observed in Kitaev spin liquids and could point to a non-Abelian symmetry. As the ground state in the bulk of SrNi$_2$V$_2$O$_8$ is topologically trivial, we suggest its fractionalization to be due to light-induced interchain exchange processes. These processes are supposed to be enhanced due to a proximity to an Ising ordered state with a quantum critical point. A comparison with SrCo$_2$V$_2$O$_8$, the $S=1/2$ analogue to our title compound, supports these statements. △ Less

Submitted 16 June, 2021; originally announced June 2021.

Comments: 3 figures, 1 table

Journal ref: Phys. Rev. B 104, 165118, (2021)

Neutron diffraction of field-induced magnon condensation in the spin-dimerized antiferromagnet Sr$_{3}$Cr$_{2}$O$_{8}$

Authors: Alsu Gazizulina, Diana Lucia Quintero-Castro, Zhe Wang, Fabienne Duc, Frederic Bourdarot, Karel Prokes, Wolfgang Schmidt, Ramzy Daou, Sergei Zherlitsyn, Nazmul Islam, Nils Henrik Kolnes, Abhijit Bhat Kademane, Andreas Schilling, Bella Lake

Abstract: In this work, we investigate the evolution and settling of magnon condensation in the spin-1/2 dimer system Sr$_{3}$Cr$_{2}$O$_{8}$ using a combination of magnetostriction in pulsed fields and inelastic neutron scattering in a continuous magnetic field. The magnetic structure in the Bose-Einstein condensation (BEC) phase was probed by neutron diffraction in pulsed magnetic fields up to 39~T. The m… ▽ More In this work, we investigate the evolution and settling of magnon condensation in the spin-1/2 dimer system Sr$_{3}$Cr$_{2}$O$_{8}$ using a combination of magnetostriction in pulsed fields and inelastic neutron scattering in a continuous magnetic field. The magnetic structure in the Bose-Einstein condensation (BEC) phase was probed by neutron diffraction in pulsed magnetic fields up to 39~T. The magnetic structure in this phase was confirmed to be an XY-antiferromagnetic structure validated by irreducible representational analysis. The magnetic phase diagram as a function of an applied magnetic field for this system is presented. Furthermore, zero-field neutron diffraction results indicate that dimerization plays an important role in stabilizing the low-temperature crystal structure. △ Less

Submitted 4 June, 2021; originally announced June 2021.

Comments: 6 pages, 4 figures

Journal ref: Phys. Rev. B 104, 064430 (2021)

Flexible Compositional Learning of Structured Visual Concepts

Authors: Yanli Zhou, Brenden M. Lake

Abstract: Humans are highly efficient learners, with the ability to grasp the meaning of a new concept from just a few examples. Unlike popular computer vision systems, humans can flexibly leverage the compositional structure of the visual world, understanding new concepts as combinations of existing concepts. In the current paper, we study how people learn different types of visual compositions, using abst… ▽ More Humans are highly efficient learners, with the ability to grasp the meaning of a new concept from just a few examples. Unlike popular computer vision systems, humans can flexibly leverage the compositional structure of the visual world, understanding new concepts as combinations of existing concepts. In the current paper, we study how people learn different types of visual compositions, using abstract visual forms with rich relational structure. We find that people can make meaningful compositional generalizations from just a few examples in a variety of scenarios, and we develop a Bayesian program induction model that provides a close fit to the behavioral data. Unlike past work examining special cases of compositionality, our work shows how a single computational approach can account for many distinct types of compositional generalization. △ Less

Submitted 20 May, 2021; originally announced May 2021.

Comments: Please cite as: Zhou, Y. and Lake, B. M. (2021). Flexible compositional learning of structured visual concepts. In Proceedings of the 43rd Annual Conference of the Cognitive Science Society

Spin liquid and ferroelectricity close to a quantum critical point in PbCuTe$_2$O$_6$

Authors: Christian Thurn, Paul Eibisch, Arif Ata, Maximilian Winkler, Peter Lunkenheimer, István Kézsmárki, Ulrich Tutsch, Yohei Saito, Steffi Hartmann, Jan Zimmermann, Abanoub R. N. Hanna, A. T. M. Nazmul Islam, Shravani Chillal, Bella Lake, Bernd Wolf, Michael Lang

Abstract: Geometrical frustration among interacting spins combined with strong quantum fluctuations destabilize long-range magnetic order in favour of more exotic states such as spin liquids. By following this guiding principle, a number of spin liquid candidate systems were identified in quasi-two-dimensional (quasi-2D) systems. For 3D, however, the situation is less favourable as quantum fluctuations are… ▽ More Geometrical frustration among interacting spins combined with strong quantum fluctuations destabilize long-range magnetic order in favour of more exotic states such as spin liquids. By following this guiding principle, a number of spin liquid candidate systems were identified in quasi-two-dimensional (quasi-2D) systems. For 3D, however, the situation is less favourable as quantum fluctuations are reduced and competing states become more relevant. Here we report a comprehensive study of thermodynamic, magnetic and dielectric properties on single crystalline and pressed-powder samples of PbCuTe$_2$O$_6$, a candidate material for a 3D frustrated quantum spin liquid featuring a hyperkagome lattice. Whereas the low-temperature properties of the powder samples are consistent with the recently proposed quantum spin liquid state, an even more exotic behaviour is revealed for the single crystals. These crystals show ferroelectric order at $T_{\text{FE}} \approx 1\,\text{K}$, accompanied by strong lattice distortions, and a modified magnetic response -- still consistent with a quantum spin liquid -- but with clear indications for quantum critical behaviour. △ Less

Submitted 19 November, 2021; v1 submitted 31 March, 2021; originally announced March 2021.

Comments: 59 pages, 15 figures, This version of the article has been accepted for publication, after peer review but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online

Journal ref: npj Quantum Materials 6, 95 (2021)

Fast and flexible: Human program induction in abstract reasoning tasks

Authors: Aysja Johnson, Wai Keen Vong, Brenden M. Lake, Todd M. Gureckis

Abstract: The Abstraction and Reasoning Corpus (ARC) is a challenging program induction dataset that was recently proposed by Chollet (2019). Here, we report the first set of results collected from a behavioral study of humans solving a subset of tasks from ARC (40 out of 1000). Although this subset of tasks contains considerable variation, our results showed that humans were able to infer the underlying pr… ▽ More The Abstraction and Reasoning Corpus (ARC) is a challenging program induction dataset that was recently proposed by Chollet (2019). Here, we report the first set of results collected from a behavioral study of humans solving a subset of tasks from ARC (40 out of 1000). Although this subset of tasks contains considerable variation, our results showed that humans were able to infer the underlying program and generate the correct test output for a novel test input example, with an average of 80% of tasks solved per participant, and with 65% of tasks being solved by more than 80% of participants. Additionally, we find interesting patterns of behavioral consistency and variability within the action sequences during the generation process, the natural language descriptions to describe the transformations for each task, and the errors people made. Our findings suggest that people can quickly and reliably determine the relevant features and properties of a task to compose a correct solution. Future modeling work could incorporate these findings, potentially by connecting the natural language descriptions we collected here to the underlying semantics of ARC. △ Less

Submitted 9 March, 2021; originally announced March 2021.

Comments: 7 pages, 7 figures, 1 table

Magnetic and electronic ordering phenomena in the [Ru$_2$O$_6$] honeycomb lattice compound AgRuO$_3$

Authors: Walter Schnelle, Beluvalli E. Prasad, Claudia Felser, Martin Jansen, Evgenia V. Komleva, Sergey V. Streltsov, Igor I. Mazin, Dmitry Khalyavin, Pascal Manuel, Sukanya Pal, D. V. S. Muthu, A. K. Sood, Ekaterina S. Klyushina, Bella Lake, Jean-Christophe Orain, Hubertus Luetkens

Abstract: The silver ruthenium oxide AgRuO$_3$ consists of honeycomb [Ru$_2^{5+}$O$_6^{2-}$] layers, and can be considered an analogue of SrRu$_2$O$_6$ with a different intercalation stage. We present measurements of magnetic susceptibility and specific heat on AgRuO$_3$ single crystals which reveal a sharp antiferromagnetic transition at 342(3)K. The electrical transport in single crystals of AgRuO$_3$ is… ▽ More The silver ruthenium oxide AgRuO$_3$ consists of honeycomb [Ru$_2^{5+}$O$_6^{2-}$] layers, and can be considered an analogue of SrRu$_2$O$_6$ with a different intercalation stage. We present measurements of magnetic susceptibility and specific heat on AgRuO$_3$ single crystals which reveal a sharp antiferromagnetic transition at 342(3)K. The electrical transport in single crystals of AgRuO$_3$ is determined by a combination of activated conduction over an intrinsic semiconducting gap of $\approx$ 100 meV and carriers trapped and thermally released from defects. From powder neutron diffraction data a Néel-type antiferromagnetic structure with the Ru moments along the $c$ axis is derived. Raman and muon spin rotation spectroscopy measurements on AgRuO$_3$ powder samples indicate a further weak phase transition or a crossover in the temperature range 125-200 K. The transition does not show up in magnetic susceptibility and its origin is argued to be related to defects but cannot be fully clarified. The experimental findings are complemented by DFT-based electronic structure calculations. It is found that the magnetism in AgRuO$_3$ is similar to that of SrRu$_2$O$_6$, however with stronger intralayer and weaker interlayer magnetic exchange interactions. △ Less

Submitted 8 March, 2021; originally announced March 2021.

Comments: 13 pages, 16 figures, includes supplement 3 pages, 3 figures

Journal ref: Phys. Rev. B 103, 214413 (2021)

Baby Intuitions Benchmark (BIB): Discerning the goals, preferences, and actions of others

Authors: Kanishk Gandhi, Gala Stojnic, Brenden M. Lake, Moira R. Dillon

Abstract: To achieve human-like common sense about everyday life, machine learning systems must understand and reason about the goals, preferences, and actions of other agents in the environment. By the end of their first year of life, human infants intuitively achieve such common sense, and these cognitive achievements lay the foundation for humans' rich and complex understanding of the mental states of ot… ▽ More To achieve human-like common sense about everyday life, machine learning systems must understand and reason about the goals, preferences, and actions of other agents in the environment. By the end of their first year of life, human infants intuitively achieve such common sense, and these cognitive achievements lay the foundation for humans' rich and complex understanding of the mental states of others. Can machines achieve generalizable, commonsense reasoning about other agents like human infants? The Baby Intuitions Benchmark (BIB) challenges machines to predict the plausibility of an agent's behavior based on the underlying causes of its actions. Because BIB's content and paradigm are adopted from developmental cognitive science, BIB allows for direct comparison between human and machine performance. Nevertheless, recently proposed, deep-learning-based agency reasoning models fail to show infant-like reasoning, leaving BIB an open challenge. △ Less

Submitted 11 February, 2022; v1 submitted 23 February, 2021; originally announced February 2021.

Comments: Published in Advances in Neural Information Processing Systems (NeurIPS) 34

Witnessing entanglement in quantum magnets using neutron scattering

Authors: A. Scheie, Pontus Laurell, A. M. Samarakoon, B. Lake, S. E. Nagler, G. E. Granroth, S. Okamoto, G. Alvarez, D. A. Tennant

Abstract: We demonstrate how quantum entanglement can be directly witnessed in the quasi-1D Heisenberg antiferromagnet KCuF$_3$. We apply three entanglement witnesses --- one-tangle, two-tangle, and quantum Fisher information --- to its inelastic neutron spectrum, and compare with spectra simulated by finite-temperature density matrix renormalization group (DMRG) and classical Monte Carlo methods. We find t… ▽ More We demonstrate how quantum entanglement can be directly witnessed in the quasi-1D Heisenberg antiferromagnet KCuF$_3$. We apply three entanglement witnesses --- one-tangle, two-tangle, and quantum Fisher information --- to its inelastic neutron spectrum, and compare with spectra simulated by finite-temperature density matrix renormalization group (DMRG) and classical Monte Carlo methods. We find that each witness provides direct access to entanglement. Of these, quantum Fisher information is the most robust experimentally, and indicates the presence of at least bipartite entanglement up to at least 50 K, corresponding to around 10% of the spinon zone-boundary energy. We apply quantum Fisher information to higher spin-S Heisenberg chains, and show theoretically that the witnessable entanglement gets suppressed to lower temperatures as the quantum number increases. Finally, we outline how these results can be applied to higher dimensional quantum materials to witness and quantify entanglement. △ Less

Submitted 16 February, 2021; originally announced February 2021.

Comments: 9 pages and 5 figures, four pages and six figures of appendices, and five pages supplemental information

Journal ref: Phys. Rev. B 103, 224434 (2021)

Structural and magnetic properties of the new quantum magnet BaCuTe$_2$O$_6$

Authors: A. Samartzis, D. Khalyavin, A. T. M. N. Islam, S. Chillal, K. Siemensmeyer, K. Prokes, D. J. Voneshen, A. Senyshyn, B. Lake

Abstract: We investigate the structural and magnetic properties of the new quantum magnet BaCuTe$_2$O$_6$. This compound is synthesized for the first time in powder and single crystal form. Synchrotron X-ray and neutron diffraction reveal a cubic crystal structure (P4$_1$32) where the magnetic Cu$^{2+}$ ions form a complex network. Physical properties measurements suggest the presence of antiferromagnetic i… ▽ More We investigate the structural and magnetic properties of the new quantum magnet BaCuTe$_2$O$_6$. This compound is synthesized for the first time in powder and single crystal form. Synchrotron X-ray and neutron diffraction reveal a cubic crystal structure (P4$_1$32) where the magnetic Cu$^{2+}$ ions form a complex network. Physical properties measurements suggest the presence of antiferromagnetic interactions with a Curie-Weiss temperature of -33K, while long-range magnetic order occurs at the much lower temperature of ~6.3K. The magnetic structure, solved using neutron diffraction, reveals antiferromagnetic order along chains parallel to the a, b and c crystal axes. This is consistent with the magnetic excitations which resemble the multispinon continuum typical of the spin-1/2 Heisenberg antiferromagnetic chain. A consistent intrachain interaction value of ~34K is achieved from the various techniques. Finally the magnetic structure provides evidence that the chains are coupled together in a non-colinear arrangement by a much weaker antiferromagnetic, frustrated hyperkagome interaction. △ Less

Submitted 15 February, 2021; originally announced February 2021.

Journal ref: Phys. Rev. B 103, 094417 (2021)

Signatures for Berezinsky-Kosterlitz-Thouless critical behaviour in the planar antiferromagnet BaNi$_2$V$_2$O$_8$

Authors: E. S. Klyushina, J. Reuther, L. Weber, A. T. M. N. Islam, J. S. Lord, B. Klemke, M. Månsson, S. Wessel, B. Lake

Abstract: We investigate the critical properties of the spin-$1$ honeycomb antiferromagnet BaNi$_2$V$_2$O$_8$, both below and above the ordering temperature $T_N$ using neutron diffraction and muon spin rotation measurements. Our results characterize BaNi$_2$V$_2$O$_8$ as a two-dimensional (2D) antiferromagnet across the entire temperature range, displaying a series of crossovers from 2D Ising-like to 2D XY… ▽ More We investigate the critical properties of the spin-$1$ honeycomb antiferromagnet BaNi$_2$V$_2$O$_8$, both below and above the ordering temperature $T_N$ using neutron diffraction and muon spin rotation measurements. Our results characterize BaNi$_2$V$_2$O$_8$ as a two-dimensional (2D) antiferromagnet across the entire temperature range, displaying a series of crossovers from 2D Ising-like to 2D XY and then to 2D Heisenberg behavior with increasing temperature. In particular, the extracted critical exponent of the order parameter reveals a narrow temperature regime close to $T_N$, in which the system behaves as a 2D XY antiferromagnet. Above $T_N$, evidence for Berezinsky-Kosterlitz-Thouless behavior driven by vortex excitations is obtained from the scaling of the correlation length. Our experimental results are in accord with classical and quantum Monte Carlo simulations performed for microscopic magnetic model Hamiltonians for BaNi$_2$V$_2$O$_8$. △ Less

Submitted 8 February, 2021; v1 submitted 4 December, 2020; originally announced December 2020.

Comments: 16 pages, 17 figures

Journal ref: Phys. Rev. B 104, 064402 (2021)

Strongly coupled charge, orbital and spin order in TbTe$_{3}$

Authors: S. Chillal, E. Schierle, E. Weschke, F. Yokaichiya, J. -U. Hoffmann, O. S. Volkova, A. N. Vasiliev, A. A. Sinchenko, P. Lejay, A. Hadj-Azzem, P. Monceau, B. Lake

Abstract: We report a ground state with strongly coupled magnetic and charge density wave orders mediated via orbital ordering in the layered compound \tbt. In addition to the commensurate antiferromagnetic (AFM) and charge density wave (CDW) orders, new magnetic peaks are observed whose propagation vector equals the sum of the AFM and CDW propagation vectors, revealing an intricate and highly entwined rela… ▽ More We report a ground state with strongly coupled magnetic and charge density wave orders mediated via orbital ordering in the layered compound \tbt. In addition to the commensurate antiferromagnetic (AFM) and charge density wave (CDW) orders, new magnetic peaks are observed whose propagation vector equals the sum of the AFM and CDW propagation vectors, revealing an intricate and highly entwined relationship. This is especially interesting given that the magnetic and charge orders lie in different layers of the crystal structure where the highly localized magnetic moments of the Tb$^{3+}$ ions are netted in the Tb-Te stacks, while the charge order is formed by the conduction electrons of the adjacent Te-Te layers. Our results, based on neutron diffraction and resonant x-ray scattering reveal that the charge and magnetic subsystems mutually influence each other via the orbital ordering of Tb$^{3+}$ ions. △ Less

Submitted 2 December, 2020; originally announced December 2020.

Comments: 5 pages, 4 figures

Journal ref: Phys. Rev. B 102, 241110 (2020)

Mn-rich MnSb2Te4: A topological insulator with magnetic gap closing at high Curie temperatures of 45-50 K

Authors: S. Wimmer, J. Sánchez-Barriga, P. Küppers, A. Ney, E. Schierle, F. Freyse, O. Caha, J. Michalicka, M. Liebmann, D. Primetzhofer, M. Hoffmann, A. Ernst, M. M. Otrokov, G. Bihlmayer, E. Weschke, B. Lake, E. V. Chulkov, M. Morgenstern, G. Bauer, G. Springholz, O. Rader

Abstract: Ferromagnetic topological insulators exhibit the quantum anomalous Hall effect that might be used for high precision metrology and edge channel spintronics. In conjunction with superconductors, they could host chiral Majorana zero modes which are among the contenders for the realization of topological qubits. Recently, it was discovered that the stable 2+ state of Mn enables the formation of intri… ▽ More Ferromagnetic topological insulators exhibit the quantum anomalous Hall effect that might be used for high precision metrology and edge channel spintronics. In conjunction with superconductors, they could host chiral Majorana zero modes which are among the contenders for the realization of topological qubits. Recently, it was discovered that the stable 2+ state of Mn enables the formation of intrinsic magnetic topological insulators with A1B2C4 stoichiometry. However, the first representative, MnBi2Te4, is antiferromagnetic with 25 K Néel temperature and strongly n-doped. Here, we show that p-type MnSb2Te4, previously considered topologically trivial, is a ferromagnetic topological insulator in the case of a few percent of Mn excess. It shows (i) a ferromagnetic hysteresis with record high Curie temperature of 45-50 K, (ii) out-of-plane magnetic anisotropy and (iii) a two-dimensional Dirac cone with the Dirac point close to the Fermi level which features (iv) out-of-plane spin polarization as revealed by photoelectron spectroscopy and (v) a magnetically induced band gap that closes at the Curie temperature as demonstrated by scanning tunneling spectroscopy. Moreover, it displays (vi) a critical exponent of magnetization beta~1, indicating the vicinity of a quantum critical point. Ab initio band structure calculations reveal that the slight excess of Mn that substitutionally replaces Sb atoms provides the ferromagnetic interlayer coupling. Remaining deviations from the ferromagnetic order, likely related to this substitution, open the inverted bulk band gap and render MnSb2Te4 a robust topological insulator and new benchmark for magnetic topological insulators. △ Less

Submitted 25 April, 2021; v1 submitted 13 November, 2020; originally announced November 2020.

Journal ref: Adv. Mater. (1st September 2021)

CURI: A Benchmark for Productive Concept Learning Under Uncertainty

Authors: Ramakrishna Vedantam, Arthur Szlam, Maximilian Nickel, Ari Morcos, Brenden Lake

Abstract: Humans can learn and reason under substantial uncertainty in a space of infinitely many concepts, including structured relational concepts ("a scene with objects that have the same color") and ad-hoc categories defined through goals ("objects that could fall on one's head"). In contrast, standard classification benchmarks: 1) consider only a fixed set of category labels, 2) do not evaluate composi… ▽ More Humans can learn and reason under substantial uncertainty in a space of infinitely many concepts, including structured relational concepts ("a scene with objects that have the same color") and ad-hoc categories defined through goals ("objects that could fall on one's head"). In contrast, standard classification benchmarks: 1) consider only a fixed set of category labels, 2) do not evaluate compositional concept learning and 3) do not explicitly capture a notion of reasoning under uncertainty. We introduce a new few-shot, meta-learning benchmark, Compositional Reasoning Under Uncertainty (CURI) to bridge this gap. CURI evaluates different aspects of productive and systematic generalization, including abstract understandings of disentangling, productive generalization, learning boolean operations, variable binding, etc. Importantly, it also defines a model-independent "compositionality gap" to evaluate the difficulty of generalizing out-of-distribution along each of these axes. Extensive evaluations across a range of modeling choices spanning different modalities (image, schemas, and sounds), splits, privileged auxiliary concept information, and choices of negatives reveal substantial scope for modeling advances on the proposed task. All code and datasets will be available online. △ Less

Submitted 6 October, 2020; originally announced October 2020.

Enhanced spin correlations in the Bose-Einstein condensate compound Sr3Cr2O8

Authors: T. Nomura, Y. Skourski, D. L. Quintero-Castro, A. A. Zvyagin, A. V. Suslov, D. Gorbunov, S. Yasin, J. Wosnitza, K. Kindo, A. T. M. N. Islam, B. Lake, Y. Kohama, S. Zherlitsyn, M. Jaime

Abstract: Combined experimental and modeling studies of the magnetocaloric effect, ultrasound, and magnetostriction were performed on single-crystal samples of the spin-dimer system Sr$_3$Cr$_2$O$_8$ in large magnetic fields, to probe the spin-correlated regime in the proximity of the field-induced XY-type antiferromagnetic order also referred to as a Bose-Einstein condensate of magnons. The magnetocaloric… ▽ More Combined experimental and modeling studies of the magnetocaloric effect, ultrasound, and magnetostriction were performed on single-crystal samples of the spin-dimer system Sr$_3$Cr$_2$O$_8$ in large magnetic fields, to probe the spin-correlated regime in the proximity of the field-induced XY-type antiferromagnetic order also referred to as a Bose-Einstein condensate of magnons. The magnetocaloric effect, measured under adiabatic conditions, reveals details of the field-temperature ($H,T$) phase diagram, a dome characterized by critical magnetic fields $H_{c1}$ = 30.4 T, $H_{c2}$ = 62 T, and a single maximum ordering temperature $T_{\rm max}(45~$T$)\simeq$8 K. The sample temperature was observed to drop significantly as the magnetic field is increased, even for initial temperatures above $T_{\rm max}$, indicating a significant magnetic entropy associated to the field-induced closure of the spin gap. The ultrasound and magnetostriction experiments probe the coupling between the lattice degrees of freedom and the magnetism in Sr$_3$Cr$_2$O$_8$. Our experimental results are qualitatively reproduced by a minimalistic phenomenological model of the exchange-striction by which sound waves renormalize the effective exchange couplings. △ Less

Submitted 11 August, 2020; originally announced August 2020.

Comments: 9 pages, 14 figures

Magnetic structure of a new quantum magnet SrCuTe$_2$O$_6$

Authors: S. Chillal, A. T. M. N Islam, H. Luetkens, E. Canévet, Y. Skourski, D. Khalyavin, B. Lake

Abstract: SrCuTe$_2$O$_6$ consists of a 3-dimensional arrangement of spin-$\frac{1}{2}$ Cu$^{2+}$ ions. The 1st, 2nd and 3rd neighbor interactions respectively couple Cu$^{2+}$ moments into a network of isolated triangles, a highly frustrated hyperkagome lattice consisting of corner sharing triangles and antiferromagnetic chains. Of these, the chain interaction dominates in SrCuTe$_2$O$_6$ while the other t… ▽ More SrCuTe$_2$O$_6$ consists of a 3-dimensional arrangement of spin-$\frac{1}{2}$ Cu$^{2+}$ ions. The 1st, 2nd and 3rd neighbor interactions respectively couple Cu$^{2+}$ moments into a network of isolated triangles, a highly frustrated hyperkagome lattice consisting of corner sharing triangles and antiferromagnetic chains. Of these, the chain interaction dominates in SrCuTe$_2$O$_6$ while the other two interactions lead to frustrated inter-chain coupling giving rise to long range magnetic order at suppressed temperatures. In this paper, we investigate the magnetic properties in SrCuTe$_2$O$_6$ using muon relaxation spectroscopy and neutron diffraction and present the low temperature magnetic structure. △ Less

Submitted 2 December, 2020; v1 submitted 5 August, 2020; originally announced August 2020.

Comments: 10 pages, 11 figures, 3 tables

Journal ref: Phys. Rev. B 102, 224424 (2020)

Word meaning in minds and machines

Authors: Brenden M. Lake, Gregory L. Murphy

Abstract: Machines have achieved a broad and growing set of linguistic competencies, thanks to recent progress in Natural Language Processing (NLP). Psychologists have shown increasing interest in such models, comparing their output to psychological judgments such as similarity, association, priming, and comprehension, raising the question of whether the models could serve as psychological theories. In this… ▽ More Machines have achieved a broad and growing set of linguistic competencies, thanks to recent progress in Natural Language Processing (NLP). Psychologists have shown increasing interest in such models, comparing their output to psychological judgments such as similarity, association, priming, and comprehension, raising the question of whether the models could serve as psychological theories. In this article, we compare how humans and machines represent the meaning of words. We argue that contemporary NLP systems are fairly successful models of human word similarity, but they fall short in many other respects. Current models are too strongly linked to the text-based patterns in large corpora, and too weakly linked to the desires, goals, and beliefs that people express through words. Word meanings must also be grounded in perception and action and be capable of flexible combinations in ways that current systems are not. We discuss more promising approaches to grounding NLP systems and argue that they will be more successful with a more human-like, conceptual basis for word meaning. △ Less

Submitted 17 April, 2021; v1 submitted 4 August, 2020; originally announced August 2020.

Comments: In press at Psychological Review

Self-supervised learning through the eyes of a child

Authors: A. Emin Orhan, Vaibhav V. Gupta, Brenden M. Lake

Abstract: Within months of birth, children develop meaningful expectations about the world around them. How much of this early knowledge can be explained through generic learning mechanisms applied to sensory data, and how much of it requires more substantive innate inductive biases? Addressing this fundamental question in its full generality is currently infeasible, but we can hope to make real progress in… ▽ More Within months of birth, children develop meaningful expectations about the world around them. How much of this early knowledge can be explained through generic learning mechanisms applied to sensory data, and how much of it requires more substantive innate inductive biases? Addressing this fundamental question in its full generality is currently infeasible, but we can hope to make real progress in more narrowly defined domains, such as the development of high-level visual categories, thanks to improvements in data collecting technology and recent progress in deep learning. In this paper, our goal is precisely to achieve such progress by utilizing modern self-supervised deep learning methods and a recent longitudinal, egocentric video dataset recorded from the perspective of three young children (Sullivan et al., 2020). Our results demonstrate the emergence of powerful, high-level visual representations from developmentally realistic natural videos using generic self-supervised learning objectives. △ Less

Submitted 15 December, 2020; v1 submitted 31 July, 2020; originally announced July 2020.

Comments: Published as a conference paper at NeurIPS 2020; v3 adds a reference, fixes a typo

Learning Task-General Representations with Generative Neuro-Symbolic Modeling

Authors: Reuben Feinman, Brenden M. Lake

Abstract: People can learn rich, general-purpose conceptual representations from only raw perceptual inputs. Current machine learning approaches fall well short of these human standards, although different modeling traditions often have complementary strengths. Symbolic models can capture the compositional and causal knowledge that enables flexible generalization, but they struggle to learn from raw inputs,… ▽ More People can learn rich, general-purpose conceptual representations from only raw perceptual inputs. Current machine learning approaches fall well short of these human standards, although different modeling traditions often have complementary strengths. Symbolic models can capture the compositional and causal knowledge that enables flexible generalization, but they struggle to learn from raw inputs, relying on strong abstractions and simplifying assumptions. Neural network models can learn directly from raw data, but they struggle to capture compositional and causal structure and typically must retrain to tackle new tasks. We bring together these two traditions to learn generative models of concepts that capture rich compositional and causal structure, while learning from raw data. We develop a generative neuro-symbolic (GNS) model of handwritten character concepts that uses the control flow of a probabilistic program, coupled with symbolic stroke primitives and a symbolic image renderer, to represent the causal and compositional processes by which characters are formed. The distributions of parts (strokes), and correlations between parts, are modeled with neural network subroutines, allowing the model to learn directly from raw data and express nonparametric statistical relationships. We apply our model to the Omniglot challenge of human-level concept learning, using a background set of alphabets to learn an expressive prior distribution over character drawings. In a subsequent evaluation, our GNS model uses probabilistic inference to learn rich conceptual representations from a single training image that generalize to 4 unique tasks, succeeding where previous work has fallen short. △ Less

Submitted 23 January, 2021; v1 submitted 25 June, 2020; originally announced June 2020.

Journal ref: International Conference on Learning Representations (ICLR 2021)