-
Multimodal Banking Dataset: Understanding Client Needs through Event Sequences
Authors:
Mollaev Dzhambulat,
Alexander Kostin,
Postnova Maria,
Ivan Karpukhin,
Ivan A Kireev,
Gleb Gusev,
Andrey Savchenko
Abstract:
Financial organizations collect a huge amount of data about clients that typically has a temporal (sequential) structure and is collected from various sources (modalities). Due to privacy issues, there are no large-scale open-source multimodal datasets of event sequences, which significantly limits the research in this area. In this paper, we present the industrial-scale publicly available multimo…
▽ More
Financial organizations collect a huge amount of data about clients that typically has a temporal (sequential) structure and is collected from various sources (modalities). Due to privacy issues, there are no large-scale open-source multimodal datasets of event sequences, which significantly limits the research in this area. In this paper, we present the industrial-scale publicly available multimodal banking dataset, MBD, that contains more than 1.5M corporate clients with several modalities: 950M bank transactions, 1B geo position events, 5M embeddings of dialogues with technical support and monthly aggregated purchases of four bank's products. All entries are properly anonymized from real proprietary bank data. Using this dataset, we introduce a novel benchmark with two business tasks: campaigning (purchase prediction in the next month) and matching of clients. We provide numerical results that demonstrate the superiority of our multi-modal baselines over single-modal techniques for each task. As a result, the proposed dataset can open new perspectives and facilitate the future development of practically important large-scale multimodal algorithms for event sequences.
HuggingFace Link: https://huggingface.co/datasets/ai-lab/MBD
Github Link: https://github.com/Dzhambo/MBD
△ Less
Submitted 26 September, 2024;
originally announced September 2024.
-
TransCAD: A Hierarchical Transformer for CAD Sequence Inference from Point Clouds
Authors:
Elona Dupont,
Kseniya Cherenkova,
Dimitrios Mallis,
Gleb Gusev,
Anis Kacem,
Djamila Aouada
Abstract:
3D reverse engineering, in which a CAD model is inferred given a 3D scan of a physical object, is a research direction that offers many promising practical applications. This paper proposes TransCAD, an end-to-end transformer-based architecture that predicts the CAD sequence from a point cloud. TransCAD leverages the structure of CAD sequences by using a hierarchical learning strategy. A loop refi…
▽ More
3D reverse engineering, in which a CAD model is inferred given a 3D scan of a physical object, is a research direction that offers many promising practical applications. This paper proposes TransCAD, an end-to-end transformer-based architecture that predicts the CAD sequence from a point cloud. TransCAD leverages the structure of CAD sequences by using a hierarchical learning strategy. A loop refiner is also introduced to regress sketch primitive parameters. Rigorous experimentation on the DeepCAD and Fusion360 datasets show that TransCAD achieves state-of-the-art results. The result analysis is supported with a proposed metric for CAD sequence, the mean Average Precision of CAD Sequence, that addresses the limitations of existing metrics.
△ Less
Submitted 18 July, 2024; v1 submitted 17 July, 2024;
originally announced July 2024.
-
Interaction-controlled transport in a two-dimensional massless-massive Dirac system: Transition from degenerate to nondegenerate regimes
Authors:
A. D. Levin,
G. M. Gusev,
F. G. G. Hernandez,
E. B. Olshanetsky,
V. M. Kovalev,
M. V. Entin,
N. N. Mikhailov
Abstract:
The resistivity of two-dimensional (2D) metals generally exhibits insensitivity to electron-electron scattering. However, it's worth noting that Galilean invariance may not hold true in systems characterized by a spectrum containing multiple electronic branches or in scenarios involving electron-hole plasma. In the context of our study, we focus on 2D electrons confined within a triple quantum wel…
▽ More
The resistivity of two-dimensional (2D) metals generally exhibits insensitivity to electron-electron scattering. However, it's worth noting that Galilean invariance may not hold true in systems characterized by a spectrum containing multiple electronic branches or in scenarios involving electron-hole plasma. In the context of our study, we focus on 2D electrons confined within a triple quantum well (TQW) based on HgTe. This system displays a coexistence of energy bands featuring both linear and parabolic-like spectra at low energy and, therefore, lacks the Galilean invariance. This research employs a combined theoretical and experimental approach to investigate the transport properties of this two-component system across various regimes. By manipulating carrier density and temperature, we tune our system from a fully degenerate regime, where resistance follows a temperature-dependent behavior proportional to $T^2$, to a regime where both types of electrons adhere to Boltzmann statistics. In the non-degenerate regime, electron interactions lead to resistance that is weakly dependent on temperature. Notably, our experimental observations closely align with the theoretical predictions derived in this study. This work establishes the HgTe-based TQW as a promising platform for exploring different interaction dominant scenarios for the massless-massive Dirac system.9 pages, 8 figures
△ Less
Submitted 3 May, 2024;
originally announced May 2024.
-
Hydrodynamics of electron-hole fluid photogenerated in a mesoscopic two-dimensional channel
Authors:
M. A. T. Patricio,
G. M. Jacobsen,
M. D. Teodoro,
G. M. Gusev,
A. K. Bakarov,
Yu. A. Pusep
Abstract:
The dynamics of the diffusion flow of holes photoinjected into a mesoscopic GaAs channel of variable width, where they, together with background electrons, form a hydrodynamic electron-hole fluid, is studied using time-resolved microphotoluminescence. It is found that the rate of recombination of photoinjected holes, which is proportional to the rate of their flow, decreases when holes pass throug…
▽ More
The dynamics of the diffusion flow of holes photoinjected into a mesoscopic GaAs channel of variable width, where they, together with background electrons, form a hydrodynamic electron-hole fluid, is studied using time-resolved microphotoluminescence. It is found that the rate of recombination of photoinjected holes, which is proportional to the rate of their flow, decreases when holes pass through the expanded sections of the channel. In fact, this is the Venturi effect, which consists in a decrease in the velocity of the fluid in the expanded sections of the pipe. Moreover, a non-uniform diffusion velocity profile is observed, similar to the parabolic Hagen-Poiseuille velocity profile, which indicates a viscous hydrodynamic flow. It is shown that in argeement with a theory, the magnetic field strongly suppresses the viscosity of the electron-hole fluid. Additional evidence of the viscous nature of the studied electron-hole fluid is the observed increase in the recombination rate with increasing temperature, which is similar to the decrease in the electrical resistance of viscous electrons with temperature.
△ Less
Submitted 6 March, 2024;
originally announced March 2024.
-
Interaction dominated transport in 2D conductors: from degenerate to partially-degenerate regime
Authors:
G. M. Gusev,
A. D. Levin,
E. B. Olshanetsky,
Z. D. Kvon,
V. M. Kovalev,
M. V. Entin,
N. N. Mikhailov
Abstract:
In this study, we investigate the conductivity of a two-dimensional (2D) system in HgTe quantum well comprising two types of carriers with linear and quadratic spectra, respectively. The interactions between the two-dimensional Dirac holes and the heavy holes lead to the breakdown of Galilean invariance, resulting in interaction-limited resistivity. Our exploration of the transport properties span…
▽ More
In this study, we investigate the conductivity of a two-dimensional (2D) system in HgTe quantum well comprising two types of carriers with linear and quadratic spectra, respectively. The interactions between the two-dimensional Dirac holes and the heavy holes lead to the breakdown of Galilean invariance, resulting in interaction-limited resistivity. Our exploration of the transport properties spans from low temperatures, where both subsystems are fully degenerate, to higher temperatures, where the Dirac holes remain degenerate while the heavy holes follow Boltzmann statistics, creating a partially degenerate regime. Through a developed theory, we successfully predict the behavior of resistivity as $ρ\sim T^2$ and $ρ\sim T^{3}$ for the fully degenerate and partially degenerate regimes, respectively, which is in reasonable agreement with experimental observations. Notably, at elevated temperatures, the interaction-limited resistivity surpasses the resistivity caused by impurity scattering by a factor of 5-6. These findings imply that the investigated system serves as a versatile experimental platform for exploring various interaction-limited transport regimes in two component plasma.
△ Less
Submitted 2 January, 2024;
originally announced January 2024.
-
Geometric engineering of viscous magnetotransport in a two-dimensional electron system
Authors:
A. D. Levin,
G. M. Gusev,
A. S. Yaroshevich,
Z. D. Kvon,
A. K. Bakarov
Abstract:
In this study, we present our experimental investigation on the magnetotransport properties of a two-dimensional electron system in GaAs quantum wells utilizing a variety of device geometries, including obstacles with thin barriers and periodic width variations. Our primary focus is to explore the impact of these geometries on the electron viscous flow parameters, enabling precise manipulation of…
▽ More
In this study, we present our experimental investigation on the magnetotransport properties of a two-dimensional electron system in GaAs quantum wells utilizing a variety of device geometries, including obstacles with thin barriers and periodic width variations. Our primary focus is to explore the impact of these geometries on the electron viscous flow parameters, enabling precise manipulation of hydrodynamic effects under controlled conditions. Through an analysis of the large negative magnetoresistivity and zero field resistivity, we deduce the scattering times for electron-electron and electron-phonon interactions, as well as the effective channel width. Our findings confirm that the system under investigation serves as a tunable experimental platform for investigating hydrodynamic transport regimes at temperatures above 10 K.
△ Less
Submitted 22 September, 2023;
originally announced September 2023.
-
SHARP Challenge 2023: Solving CAD History and pArameters Recovery from Point clouds and 3D scans. Overview, Datasets, Metrics, and Baselines
Authors:
Dimitrios Mallis,
Sk Aziz Ali,
Elona Dupont,
Kseniya Cherenkova,
Ahmet Serdar Karadeniz,
Mohammad Sadil Khan,
Anis Kacem,
Gleb Gusev,
Djamila Aouada
Abstract:
Recent breakthroughs in geometric Deep Learning (DL) and the availability of large Computer-Aided Design (CAD) datasets have advanced the research on learning CAD modeling processes and relating them to real objects. In this context, 3D reverse engineering of CAD models from 3D scans is considered to be one of the most sought-after goals for the CAD industry. However, recent efforts assume multipl…
▽ More
Recent breakthroughs in geometric Deep Learning (DL) and the availability of large Computer-Aided Design (CAD) datasets have advanced the research on learning CAD modeling processes and relating them to real objects. In this context, 3D reverse engineering of CAD models from 3D scans is considered to be one of the most sought-after goals for the CAD industry. However, recent efforts assume multiple simplifications limiting the applications in real-world settings. The SHARP Challenge 2023 aims at pushing the research a step closer to the real-world scenario of CAD reverse engineering through dedicated datasets and tracks. In this paper, we define the proposed SHARP 2023 tracks, describe the provided datasets, and propose a set of baseline methods along with suitable evaluation metrics to assess the performance of the track solutions. All proposed datasets along with useful routines and the evaluation metrics are publicly available.
△ Less
Submitted 30 August, 2023;
originally announced August 2023.
-
SepicNet: Sharp Edges Recovery by Parametric Inference of Curves in 3D Shapes
Authors:
Kseniya Cherenkova,
Elona Dupont,
Anis Kacem,
Ilya Arzhannikov,
Gleb Gusev,
Djamila Aouada
Abstract:
3D scanning as a technique to digitize objects in reality and create their 3D models, is used in many fields and areas. Though the quality of 3D scans depends on the technical characteristics of the 3D scanner, the common drawback is the smoothing of fine details, or the edges of an object. We introduce SepicNet, a novel deep network for the detection and parametrization of sharp edges in 3D shape…
▽ More
3D scanning as a technique to digitize objects in reality and create their 3D models, is used in many fields and areas. Though the quality of 3D scans depends on the technical characteristics of the 3D scanner, the common drawback is the smoothing of fine details, or the edges of an object. We introduce SepicNet, a novel deep network for the detection and parametrization of sharp edges in 3D shapes as primitive curves. To make the network end-to-end trainable, we formulate the curve fitting in a differentiable manner. We develop an adaptive point cloud sampling technique that captures the sharp features better than uniform sampling. The experiments were conducted on a newly introduced large-scale dataset of 50k 3D scans, where the sharp edge annotations were extracted from their parametric CAD models, and demonstrate significant improvement over state-of-the-art methods.
△ Less
Submitted 13 April, 2023;
originally announced April 2023.
-
CADOps-Net: Jointly Learning CAD Operation Types and Steps from Boundary-Representations
Authors:
Elona Dupont,
Kseniya Cherenkova,
Anis Kacem,
Sk Aziz Ali,
Ilya Arzhannikov,
Gleb Gusev,
Djamila Aouada
Abstract:
3D reverse engineering is a long sought-after, yet not completely achieved goal in the Computer-Aided Design (CAD) industry. The objective is to recover the construction history of a CAD model. Starting from a Boundary Representation (B-Rep) of a CAD model, this paper proposes a new deep neural network, CADOps-Net, that jointly learns the CAD operation types and the decomposition into different CA…
▽ More
3D reverse engineering is a long sought-after, yet not completely achieved goal in the Computer-Aided Design (CAD) industry. The objective is to recover the construction history of a CAD model. Starting from a Boundary Representation (B-Rep) of a CAD model, this paper proposes a new deep neural network, CADOps-Net, that jointly learns the CAD operation types and the decomposition into different CAD operation steps. This joint learning allows to divide a B-Rep into parts that were created by various types of CAD operations at the same construction step; therefore providing relevant information for further recovery of the design history. Furthermore, we propose the novel CC3D-Ops dataset that includes over $37k$ CAD models annotated with CAD operation type labels and step labels. Compared to existing datasets, the complexity and variety of CC3D-Ops models are closer to those used for industrial purposes. Our experiments, conducted on the proposed CC3D-Ops and the publicly available Fusion360 datasets, demonstrate the competitive performance of CADOps-Net with respect to state-of-the-art, and confirm the importance of the joint learning of CAD operation types and steps.
△ Less
Submitted 22 August, 2022;
originally announced August 2022.
-
Quantum transport of Dirac fermions in HgTe gapless quantum wells
Authors:
G. M. Gusev,
A. D. Levin,
D. A. Kozlov,
Z. D. Kvon,
N. N. Mikhailov
Abstract:
We study transport properties of HgTe quantum wells with critical well thickness, where the band gap is closed, and the low energy spectrum is described by a single Dirac cone. In this work, we examined both macroscopic and micron-sized (mesocopic) samples. In micron-sized samples, we observe a magnetic field induced, quantized resistance ($\sim h/2e^{2}$) at Landau filling factor $ν=0$, correspon…
▽ More
We study transport properties of HgTe quantum wells with critical well thickness, where the band gap is closed, and the low energy spectrum is described by a single Dirac cone. In this work, we examined both macroscopic and micron-sized (mesocopic) samples. In micron-sized samples, we observe a magnetic field induced, quantized resistance ($\sim h/2e^{2}$) at Landau filling factor $ν=0$, corresponding to the formation of helical edge states centered at the charge neutrality point (CNP). In macroscopic samples, the resistance near zero Landau level (LL) reveals strong oscillations, which we attribute to scattering between the edge $ν=0$ state and bulk $ν\neq 0$ hole LL. We provide a model taking an empirical approach to construct a LL diagram based on a reservoir scenario, formed by the heavy holes.
△ Less
Submitted 14 June, 2022;
originally announced June 2022.
-
Towards Computationally Feasible Deep Active Learning
Authors:
Akim Tsvigun,
Artem Shelmanov,
Gleb Kuzmin,
Leonid Sanochkin,
Daniil Larionov,
Gleb Gusev,
Manvel Avetisian,
Leonid Zhukov
Abstract:
Active learning (AL) is a prominent technique for reducing the annotation effort required for training machine learning models. Deep learning offers a solution for several essential obstacles to deploying AL in practice but introduces many others. One of such problems is the excessive computational resources required to train an acquisition model and estimate its uncertainty on instances in the un…
▽ More
Active learning (AL) is a prominent technique for reducing the annotation effort required for training machine learning models. Deep learning offers a solution for several essential obstacles to deploying AL in practice but introduces many others. One of such problems is the excessive computational resources required to train an acquisition model and estimate its uncertainty on instances in the unlabeled pool. We propose two techniques that tackle this issue for text classification and tagging tasks, offering a substantial reduction of AL iteration duration and the computational overhead introduced by deep acquisition models in AL. We also demonstrate that our algorithm that leverages pseudo-labeling and distilled models overcomes one of the essential obstacles revealed previously in the literature. Namely, it was shown that due to differences between an acquisition model used to select instances during AL and a successor model trained on the labeled data, the benefits of AL can diminish. We show that our algorithm, despite using a smaller and faster acquisition model, is capable of training a more expressive successor model with higher performance.
△ Less
Submitted 7 May, 2022;
originally announced May 2022.
-
Diffusion of photo-excited holes in viscous electron fluid
Authors:
Yu. A. Pusep,
M. D. Teodoro,
V. Laurindo Jr.,
E. R. Cardozo de Oliveira,
G. M. Gusev,
A. K. Bakarov
Abstract:
The diffusion of photo-generated holes is studied in a high-mobility mesoscopic GaAs\ channel where electrons exhibit hydrodynamic properties. It is shown that the injection of holes into such an electron system leads to the formation of a hydrodynamic three-component mixture consisted of electrons and photo-generated heavy and light holes. The obtained results are analyzed within the framework of…
▽ More
The diffusion of photo-generated holes is studied in a high-mobility mesoscopic GaAs\ channel where electrons exhibit hydrodynamic properties. It is shown that the injection of holes into such an electron system leads to the formation of a hydrodynamic three-component mixture consisted of electrons and photo-generated heavy and light holes. The obtained results are analyzed within the framework of ambipolar diffusion, which reveals characteristics of a viscous flow. Both hole types exhibit similar hydrodynamic characteristics. In such a way the diffusion lengths, ambipolar diffusion coefficient and the effective viscosity of the electron-hole system are determined.
△ Less
Submitted 31 March, 2022;
originally announced April 2022.
-
Thermoelectric transport in a three-dimensional HgTe topological insulator
Authors:
G. M. Gusev,
Z. D. Kvon,
A. D. Levin,
N. N. Mikhailov
Abstract:
The thermoelectric response of 80-nm-thick strained HgTe films of a three-dimensional topological insulator (3D TI) has been studied experimentally. An ambipolar thermopower is observed where the Fermi energy moves from conducting to the valence bulk band. The comparison between theory and experiment shows that the thermopower is mostly due to the phonon drag contribution. In the region where the…
▽ More
The thermoelectric response of 80-nm-thick strained HgTe films of a three-dimensional topological insulator (3D TI) has been studied experimentally. An ambipolar thermopower is observed where the Fermi energy moves from conducting to the valence bulk band. The comparison between theory and experiment shows that the thermopower is mostly due to the phonon drag contribution. In the region where the 2D Dirac electrons coexist with bulk hole states, the Seebeck coefficient is modified due to 2D electron - 3D hole scattering.
△ Less
Submitted 11 December, 2021;
originally announced December 2021.
-
Engineering topological phases in triple HgTe/CdTe quantum wells
Authors:
G. J. Ferreira,
D. R. Candido,
F. G. G. Hernandez,
G. M. Gusev,
E. B. Olshanetsky,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
Quantum wells formed by layers of HgTe between Hg$_{1-x}$Cd$_x$Te barriers lead to two-dimensional (2D) topological insulators, as predicted by the BHZ model. Here, we theoretically and experimentally investigate the characteristics of triple HgTe quantum wells. We describe such heterostructure with a three dimensional $8\times 8$ Kane model, and use its eigenstates to derive an effective 2D Hamil…
▽ More
Quantum wells formed by layers of HgTe between Hg$_{1-x}$Cd$_x$Te barriers lead to two-dimensional (2D) topological insulators, as predicted by the BHZ model. Here, we theoretically and experimentally investigate the characteristics of triple HgTe quantum wells. We describe such heterostructure with a three dimensional $8\times 8$ Kane model, and use its eigenstates to derive an effective 2D Hamiltonian for the system. From these we obtain a phase diagram as a function of the well and barrier widths and we identify the different topological phases composed by zero, one, two, and three sets of edge states hybridized along the quantum wells. The phase transitions are characterized by a change of the spin Chern numbers and their corresponding band inversions. Complementary, transport measurements are experimentally investigated on a sample close to the transition line between the phases with one and two sets of edges states. Accordingly, for this sample we predict a gapless spectrum with low energy bulk conduction bands given by one parabolic and one Dirac band, and with edge states immersed in the bulk valance bands. Consequently, we show that under these conditions, local and non-local transport measurements are inconclusive to characterize a sole edge state conductivity due to bulk conductivity. On the other hand, Shubnikov-de Haas (SdH) oscillations show an excellent agreement with our theory. Particularly, we show that the measured SdH oscillation frequencies agrees with our model and show clear signatures of the coexistence of a parabolic and Dirac bands.
△ Less
Submitted 2 December, 2021;
originally announced December 2021.
-
Transport through the network of topological channels in HgTe based quantum well
Authors:
G. M. Gusev,
Z. D. Kvon,
D. A. Kozlov,
E. B. Olshanetsky,
M. V. Entin,
N. N. Mikhailov
Abstract:
Topological insulators represent a new quantum state of matter which is characterized by edge or surface states and an insulating band gap in the bulk. In a two dimensional (2D) system based on the HgTe quantum well (QW) of critical width random deviations of the well width from its average value result in local crossovers from zero gap 2D Dirac fermion system to either the 2D topological insulato…
▽ More
Topological insulators represent a new quantum state of matter which is characterized by edge or surface states and an insulating band gap in the bulk. In a two dimensional (2D) system based on the HgTe quantum well (QW) of critical width random deviations of the well width from its average value result in local crossovers from zero gap 2D Dirac fermion system to either the 2D topological insulator or the ordinary insulator, forming a complicated in-plane network of helical channels along the zero-gap lines. We have studied experimentally the transport properties of the critical width HgTe quantum wells near the Dirac point, where the conductance is determined by a percolation along the zero-gap lines. The experimental results confirm the presence of percolating conducting channels of a finite width. Our work establishes the critical width HgTe QW as a promising platform for the study of the interplay between topology and localization.
△ Less
Submitted 30 November, 2021;
originally announced November 2021.
-
Magnetohydrodynamics and electro-electron interaction of massless Dirac fermions
Authors:
D. A. Khudaiberdiev,
G. M. Gusev,
E. B. Olshanetsky,
Z. D. Kvon,
N. N. Mikhailov
Abstract:
The magnetotransport properties of massless Dirac fermions in a gapless HgTe quantum well are investigated. In samples with narrow channels, a large negative magnetoresistance with a Lorentzian profile is observed, which is interpreted as a manifestation of electron viscosity due to electron-electron interaction. Comparison of experiment with theory yields the shear stress relaxation time of the D…
▽ More
The magnetotransport properties of massless Dirac fermions in a gapless HgTe quantum well are investigated. In samples with narrow channels, a large negative magnetoresistance with a Lorentzian profile is observed, which is interpreted as a manifestation of electron viscosity due to electron-electron interaction. Comparison of experiment with theory yields the shear stress relaxation time of the Dirac fermions caused by electron-electron scattering.
△ Less
Submitted 30 July, 2021;
originally announced July 2021.
-
Adversarial Attacks on Deep Models for Financial Transaction Records
Authors:
Ivan Fursov,
Matvey Morozov,
Nina Kaploukhaya,
Elizaveta Kovtun,
Rodrigo Rivera-Castro,
Gleb Gusev,
Dmitry Babaev,
Ivan Kireev,
Alexey Zaytsev,
Evgeny Burnaev
Abstract:
Machine learning models using transaction records as inputs are popular among financial institutions. The most efficient models use deep-learning architectures similar to those in the NLP community, posing a challenge due to their tremendous number of parameters and limited robustness. In particular, deep-learning models are vulnerable to adversarial attacks: a little change in the input harms the…
▽ More
Machine learning models using transaction records as inputs are popular among financial institutions. The most efficient models use deep-learning architectures similar to those in the NLP community, posing a challenge due to their tremendous number of parameters and limited robustness. In particular, deep-learning models are vulnerable to adversarial attacks: a little change in the input harms the model's output.
In this work, we examine adversarial attacks on transaction records data and defences from these attacks. The transaction records data have a different structure than the canonical NLP or time series data, as neighbouring records are less connected than words in sentences, and each record consists of both discrete merchant code and continuous transaction amount. We consider a black-box attack scenario, where the attack doesn't know the true decision model, and pay special attention to adding transaction tokens to the end of a sequence. These limitations provide more realistic scenario, previously unexplored in NLP world.
The proposed adversarial attacks and the respective defences demonstrate remarkable performance using relevant datasets from the financial industry. Our results show that a couple of generated transactions are sufficient to fool a deep-learning model. Further, we improve model robustness via adversarial training or separate adversarial examples detection. This work shows that embedding protection from adversarial attacks improves model robustness, allowing a wider adoption of deep models for transaction records in banking and finance.
△ Less
Submitted 15 June, 2021;
originally announced June 2021.
-
Viscous magnetotransport and Gurzhi effect in bilayer electron system
Authors:
G. M. Gusev,
A. S. Jaroshevich,
A. D. Levin,
Z. D. Kvon,
A. K. Bakarov
Abstract:
We observe a large negative magnetoresistance and a decrease of resistivity with increasing temperature, known as the Gurzhi effect, in a bilayer electron (BL) system formed by a wide GaAs quantum well. A hydrodynamic model for the single fluid transport parameters in narrow channels is employed and successfully describes our experimental findings.
We find that the electron-electron scattering i…
▽ More
We observe a large negative magnetoresistance and a decrease of resistivity with increasing temperature, known as the Gurzhi effect, in a bilayer electron (BL) system formed by a wide GaAs quantum well. A hydrodynamic model for the single fluid transport parameters in narrow channels is employed and successfully describes our experimental findings.
We find that the electron-electron scattering in the bilayer is more intensive in comparison with a single-band well (SW).
The hydrodynamic assumption implies a strong dependence on boundary conditions, which can be characterized by slip length, describing the behavior of a liquid near the edge. Our results reveal that slip length in a BL is shorter than in a SW, and that the BL system goes deeper into the hydrodynamic regime. This is in agreement with the model proposed where the slip length is of the order of the electron-electron mean free path.
△ Less
Submitted 10 February, 2021;
originally announced February 2021.
-
Multiple crossing of Landau levels of two-dimensional fermions in double HgTe quantum wells
Authors:
G. M. Gusev,
E. B. Olshanetsky,
F. G. G. Hernandez,
O. E. Raichev,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
The double quantum well systems consisting of two HgTe layers separated by a tunnel-transparent barrier are expected to manifest a variety of phase states including two-dimensional gapless semimetal and two-dimensional topological insulator. The presence of several subbands in such systems leads to a rich filling factor diagram in the quantum Hall regime. We have performed magnetotransport measure…
▽ More
The double quantum well systems consisting of two HgTe layers separated by a tunnel-transparent barrier are expected to manifest a variety of phase states including two-dimensional gapless semimetal and two-dimensional topological insulator. The presence of several subbands in such systems leads to a rich filling factor diagram in the quantum Hall regime. We have performed magnetotransport measurements of the HgTe-based double quantum wells in both gapless and gapped state and observed numerous crossings between the Landau levels belonging to different subbands. We analyze the Landau level crossing patterns and compare them to the results of theoretical calculations.
△ Less
Submitted 13 January, 2021;
originally announced January 2021.
-
PvDeConv: Point-Voxel Deconvolution for Autoencoding CAD Construction in 3D
Authors:
Kseniya Cherenkova,
Djamila Aouada,
Gleb Gusev
Abstract:
We propose a Point-Voxel DeConvolution (PVDeConv) module for 3D data autoencoder. To demonstrate its efficiency we learn to synthesize high-resolution point clouds of 10k points that densely describe the underlying geometry of Computer Aided Design (CAD) models. Scanning artifacts, such as protrusions, missing parts, smoothed edges and holes, inevitably appear in real 3D scans of fabricated CAD ob…
▽ More
We propose a Point-Voxel DeConvolution (PVDeConv) module for 3D data autoencoder. To demonstrate its efficiency we learn to synthesize high-resolution point clouds of 10k points that densely describe the underlying geometry of Computer Aided Design (CAD) models. Scanning artifacts, such as protrusions, missing parts, smoothed edges and holes, inevitably appear in real 3D scans of fabricated CAD objects. Learning the original CAD model construction from a 3D scan requires a ground truth to be available together with the corresponding 3D scan of an object. To solve the gap, we introduce a new dedicated dataset, the CC3D, containing 50k+ pairs of CAD models and their corresponding 3D meshes. This dataset is used to learn a convolutional autoencoder for point clouds sampled from the pairs of 3D scans - CAD models. The challenges of this new dataset are demonstrated in comparison with other generative point cloud sampling models trained on ShapeNet. The CC3D autoencoder is efficient with respect to memory consumption and training time as compared to stateof-the-art models for 3D data generation.
△ Less
Submitted 12 January, 2021;
originally announced January 2021.
-
Topological insulators based on HgTe
Authors:
Z. D. Kvon,
D. A. Kozlov,
E. B. Olshanetsky,
G. M. Gusev,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
The most interesting experimental results obtained in studies of 2D and 3D topological insulators (TIs) based on HgTe quantum wells and films are reviewed. In the case of 2D TIs, these include the observation of nonlocal ballistic and diffusion transport, the magnetic breakdown of 2D TIs, and an anomalous temperature dependence of edge-channel resistance. In 3D TIs, a record-setting high mobility…
▽ More
The most interesting experimental results obtained in studies of 2D and 3D topological insulators (TIs) based on HgTe quantum wells and films are reviewed. In the case of 2D TIs, these include the observation of nonlocal ballistic and diffusion transport, the magnetic breakdown of 2D TIs, and an anomalous temperature dependence of edge-channel resistance. In 3D TIs, a record-setting high mobility of surface two-dimensional Dirac fermions (DFs) has been attained. This enabled determining all the main TI parameters (the bulk gap and the density of DFs on both of its surfaces) and provided information on the phase of the Shubnikov - de Haas oscillations of DFs, which indicates the rigid topological coupling between the fermion spin and momentum. Prospects for further research are discussed in the conclusion.
△ Less
Submitted 26 December, 2020;
originally announced December 2020.
-
SHARP 2020: The 1st Shape Recovery from Partial Textured 3D Scans Challenge Results
Authors:
Alexandre Saint,
Anis Kacem,
Kseniya Cherenkova,
Konstantinos Papadopoulos,
Julian Chibane,
Gerard Pons-Moll,
Gleb Gusev,
David Fofi,
Djamila Aouada,
Bjorn Ottersten
Abstract:
The SHApe Recovery from Partial textured 3D scans challenge, SHARP 2020, is the first edition of a challenge fostering and benchmarking methods for recovering complete textured 3D scans from raw incomplete data. SHARP 2020 is organised as a workshop in conjunction with ECCV 2020. There are two complementary challenges, the first one on 3D human scans, and the second one on generic objects. Challen…
▽ More
The SHApe Recovery from Partial textured 3D scans challenge, SHARP 2020, is the first edition of a challenge fostering and benchmarking methods for recovering complete textured 3D scans from raw incomplete data. SHARP 2020 is organised as a workshop in conjunction with ECCV 2020. There are two complementary challenges, the first one on 3D human scans, and the second one on generic objects. Challenge 1 is further split into two tracks, focusing, first, on large body and clothing regions, and, second, on fine body details. A novel evaluation metric is proposed to quantify jointly the shape reconstruction, the texture reconstruction and the amount of completed data. Additionally, two unique datasets of 3D scans are proposed, to provide raw ground-truth data for the benchmarks. The datasets are released to the scientific community. Moreover, an accompanying custom library of software routines is also released to the scientific community. It allows for processing 3D scans, generating partial data and performing the evaluation. Results of the competition, analysed in comparison to baselines, show the validity of the proposed evaluation metrics, and highlight the challenging aspects of the task and of the datasets. Details on the SHARP 2020 challenge can be found at https://cvi2.uni.lu/sharp2020/.
△ Less
Submitted 26 October, 2020;
originally announced October 2020.
-
Electrical control of spin relaxation anisotropy during drift transport in a two-dimensional electron gas
Authors:
F. G. G. Hernandez,
G. J. Ferreira,
M. Luengo-Kovac,
V. Sih,
N. M. Kawahala,
G. M. Gusev,
A. K. Bakarov
Abstract:
Spin relaxation was studied in a two-dimensional electron gas confined in a wide GaAs quantum well. Recently, the control of the spin relaxation anisotropy by diffusive motion was first shown in D. Iizasa et al., arXiv:2006.08253 (2020). Here, we demonstrate electrical control by drift transport in a system with two-subbands occupied. The combined effect of in-plane and gate voltages was investiga…
▽ More
Spin relaxation was studied in a two-dimensional electron gas confined in a wide GaAs quantum well. Recently, the control of the spin relaxation anisotropy by diffusive motion was first shown in D. Iizasa et al., arXiv:2006.08253 (2020). Here, we demonstrate electrical control by drift transport in a system with two-subbands occupied. The combined effect of in-plane and gate voltages was investigated using time-resolved Kerr rotation. The measured relaxation time present strong anisotropy with respect to the transport direction. For an in-plane accelerating electric field along $\left[110\right]$, the lifetime was strongly suppressed irrespective of the applied gate voltage. Remarkably, for transport along $\left[1\bar{1}0\right]$, the data shows spin lifetime that was gate-dependent and longer than in the $\left[110\right]$ direction regardless of the in-plane voltage. In agreement, independent results of anisotropic spin precession frequencies are also presented. Nevertheless, the long spin lifetime, strong anisotropy and drift response seen in the data are beyond the existing models for spin drift and diffusion.
△ Less
Submitted 31 August, 2020; v1 submitted 20 July, 2020;
originally announced July 2020.
-
Manifestations of classical size effect and electronic viscosity in the magnetoresistance of narrow two-dimensional conductors: Theory and experiment
Authors:
O. E. Raichev,
G. M. Gusev,
A. D. Levin,
A. K. Bakarov
Abstract:
We develop a classical kinetic theory of magnetotransport of 2D electrons in narrow channels with partly diffusive boundary scattering and apply it to description of magnetoresistance measured in the temperature interval 4.2-30 K in long mesoscopic bars fabricated from high-purity GaAs quantum well structures. Both experiment and theory demonstrate a number of characteristic features in the longit…
▽ More
We develop a classical kinetic theory of magnetotransport of 2D electrons in narrow channels with partly diffusive boundary scattering and apply it to description of magnetoresistance measured in the temperature interval 4.2-30 K in long mesoscopic bars fabricated from high-purity GaAs quantum well structures. Both experiment and theory demonstrate a number of characteristic features in the longitudinal and Hall resistances caused by the size effect in two dimensions owing to the high ballisticity of the transport. In addition to the features described previously, we also reveal a change in the slope of the first derivative of magnetoresistance when the cyclotron orbit diameter equals to half of the channel width. These features are suppressed with increasing temperature as a result of the electronic viscosity due to electron-electron interaction. By comparing theory and experiment, we determine the characteristic time of relaxation of angular distribution of electrons caused by electron-electron scattering.
△ Less
Submitted 22 June, 2020;
originally announced June 2020.
-
Experimental analysis of the spin-orbit coupling dependence on the drift velocity of a spin packet
Authors:
N. M . Kawahala,
F. C. D. Moraes,
G. M. Gusev,
A. K. Bakarov,
F. G. G. Hernandez
Abstract:
Spin transport was studied in a two-dimensional electron gas hosted in a wide GaAs quantum well occupying two subbands. Using space and time Kerr rotation microscopy to image drifting spin packets under an in-plane accelerating electric field, optical injection and detection of spin polarization were achieved in a pump-probe configuration. The experimental data exhibited high spin mobility and lon…
▽ More
Spin transport was studied in a two-dimensional electron gas hosted in a wide GaAs quantum well occupying two subbands. Using space and time Kerr rotation microscopy to image drifting spin packets under an in-plane accelerating electric field, optical injection and detection of spin polarization were achieved in a pump-probe configuration. The experimental data exhibited high spin mobility and long spin lifetimes allowing to obtain the spin-orbit fields as a function of the spin velocities. Surprisingly, above moderate electric fields of 0.4V/cm with velocities higher than 2$μ$m/ns, we observed a dependence of both bulk and structure-related spin-orbit interactions on the velocity magnitude. A remarkable feature is the increase of the cubic Dresselhaus term to approximately half of the linear coupling when the velocity is raised to 10$μ$m/ns. In contrast, the Rashba coupling for both subbands decreases to about half of its value in the same range. These results yield new information for the application of drift models in spin-orbit fields and about limitations for the operation of spin transistors.
△ Less
Submitted 8 June, 2020; v1 submitted 30 May, 2020;
originally announced June 2020.
-
Stokes flow around an obstacle in viscous two-dimensional electron liquid
Authors:
G. M. Gusev,
A. S. Jaroshevich,
A. D. Levin,
Z. D. Kvon,
A. K. Bakarov
Abstract:
The electronic analog of the Poiseuille flow is the transport in a narrow channel with disordered edges that scatter electrons in a diffuse way. In the hydrodynamic regime, the resistivity decreases with temperature, referred to as the Gurzhi effect, distinct from conventional Ohmic behaviour. We studied experimentally an electronic analog of the Stokes flow around a disc immersed in a two-dimensi…
▽ More
The electronic analog of the Poiseuille flow is the transport in a narrow channel with disordered edges that scatter electrons in a diffuse way. In the hydrodynamic regime, the resistivity decreases with temperature, referred to as the Gurzhi effect, distinct from conventional Ohmic behaviour. We studied experimentally an electronic analog of the Stokes flow around a disc immersed in a two-dimensional viscous liquid. The circle obstacle results in an additive contribution to resistivity. If specular boundary conditions apply, it is no longer possible to detect Poiseuille type flow and the Gurzhi effect. However, in flow through a channel with a circular obstacle, the resistivity decreases with temperature. By tuning the temperature, we observed the transport signatures of the ballistic and hydrodynamic regimes on the length scale of disc size. Our experimental results confirm theoretical predictions.
△ Less
Submitted 12 May, 2020;
originally announced May 2020.
-
Microwave Photoresistance of a Two-Dimensional Topological Insulator in a HgTe Quantum Well
Authors:
A. S. Yaroshevich,
Z. D. Kvon,
G. M. Gusev,
N. N. Mikhailov
Abstract:
The microwave photoresistance of a two-dimensional topological insulator in a HgTe quantum well with an inverted spectrum has been experimentally studied under irradiation at frequencies of 110-169 GHz. Two mechanisms of formation of this photoresistance have been revealed. The first mechanism is due to transitions between the dispersion branches of edge current states, whereas the second mechanis…
▽ More
The microwave photoresistance of a two-dimensional topological insulator in a HgTe quantum well with an inverted spectrum has been experimentally studied under irradiation at frequencies of 110-169 GHz. Two mechanisms of formation of this photoresistance have been revealed. The first mechanism is due to transitions between the dispersion branches of edge current states, whereas the second mechanism is caused by the action of radiation on the bulk of the quantum well.
△ Less
Submitted 4 May, 2020;
originally announced May 2020.
-
Two-dimensional topological insulator state in double HgTe quantum well
Authors:
G. M. Gusev,
E. B. Olshanetsky,
F. G. G. Hernandez,
O. E. Raichev,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
The two-dimensional topological insulator phase has been observed previously in single HgTe-based quantum wells with inverted subband ordering. In double quantum wells (DQWs), coupling between the layers introduces additional degrees of freedom leading to a rich phase picture. By studying local and nonlocal resistance in HgTe-based DQWs, we observe both the gapless semimetal phase and the topologi…
▽ More
The two-dimensional topological insulator phase has been observed previously in single HgTe-based quantum wells with inverted subband ordering. In double quantum wells (DQWs), coupling between the layers introduces additional degrees of freedom leading to a rich phase picture. By studying local and nonlocal resistance in HgTe-based DQWs, we observe both the gapless semimetal phase and the topological insulator phase, depending on parameters of the samples and according to theoretical predictions. Our work establishes the DQWs as a promising platform for realization of multilayer topological insulators.
△ Less
Submitted 8 April, 2020;
originally announced April 2020.
-
CoLES: Contrastive Learning for Event Sequences with Self-Supervision
Authors:
Dmitrii Babaev,
Ivan Kireev,
Nikita Ovsov,
Mariya Ivanova,
Gleb Gusev,
Ivan Nazarov,
Alexander Tuzhilin
Abstract:
We address the problem of self-supervised learning on discrete event sequences generated by real-world users. Self-supervised learning incorporates complex information from the raw data in low-dimensional fixed-length vector representations that could be easily applied in various downstream machine learning tasks. In this paper, we propose a new method "CoLES", which adapts contrastive learning, p…
▽ More
We address the problem of self-supervised learning on discrete event sequences generated by real-world users. Self-supervised learning incorporates complex information from the raw data in low-dimensional fixed-length vector representations that could be easily applied in various downstream machine learning tasks. In this paper, we propose a new method "CoLES", which adapts contrastive learning, previously used for audio and computer vision domains, to the discrete event sequences domain in a self-supervised setting. We deployed CoLES embeddings based on sequences of transactions at the large European financial services company. Usage of CoLES embeddings significantly improves the performance of the pre-existing models on downstream tasks and produces significant financial gains, measured in hundreds of millions of dollars yearly. We also evaluated CoLES on several public event sequences datasets and showed that CoLES representations consistently outperform other methods on different downstream tasks.
△ Less
Submitted 22 July, 2022; v1 submitted 19 February, 2020;
originally announced February 2020.
-
Minimal Variance Sampling in Stochastic Gradient Boosting
Authors:
Bulat Ibragimov,
Gleb Gusev
Abstract:
Stochastic Gradient Boosting (SGB) is a widely used approach to regularization of boosting models based on decision trees. It was shown that, in many cases, random sampling at each iteration can lead to better generalization performance of the model and can also decrease the learning time. Different sampling approaches were proposed, where probabilities are not uniform, and it is not currently cle…
▽ More
Stochastic Gradient Boosting (SGB) is a widely used approach to regularization of boosting models based on decision trees. It was shown that, in many cases, random sampling at each iteration can lead to better generalization performance of the model and can also decrease the learning time. Different sampling approaches were proposed, where probabilities are not uniform, and it is not currently clear which approach is the most effective. In this paper, we formulate the problem of randomization in SGB in terms of optimization of sampling probabilities to maximize the estimation accuracy of split scoring used to train decision trees. This optimization problem has a closed-form nearly optimal solution, and it leads to a new sampling technique, which we call Minimal Variance Sampling (MVS). The method both decreases the number of examples needed for each iteration of boosting and increases the quality of the model significantly as compared to the state-of-the art sampling methods. The superiority of the algorithm was confirmed by introducing MVS as a new default option for subsampling in CatBoost, a gradient boosting library achieving state-of-the-art quality on various machine learning tasks.
△ Less
Submitted 29 October, 2019;
originally announced October 2019.
-
Mesoscopic transport in two-dimensional topological insulators
Authors:
G. M. Gusev,
Z. D Kvon,
E. B. Olshanetsky,
N. N. Mikhailov
Abstract:
Topological states of matter have attracted a lot of attention due to their many intriguing transport properties. In particular, two-dimensional topological insulators (2D TI) possess gapless counter propagating conducting edge channels, with opposite spin, that are topologically protected from backscattering. Two basic features are supposed to confirm the existence of the ballistic edge channels…
▽ More
Topological states of matter have attracted a lot of attention due to their many intriguing transport properties. In particular, two-dimensional topological insulators (2D TI) possess gapless counter propagating conducting edge channels, with opposite spin, that are topologically protected from backscattering. Two basic features are supposed to confirm the existence of the ballistic edge channels in the submicrometer limit: the 4-terminal conductance is expected to be quantized at the universal value $2e^{2}/h$, and a nonlocal signal should appear due to a net current along the sample edge, carried by the helical states. On the other hand for longer channels the conductance has been found to deviate from the quantized value. This article reviewer the experimental and theoretical work related to the transport in two-dimensional topological insulators (2D-TI), based on HgTe quantum wells in zero magnetic field. We provide an overview of the basic mechanisms predicting a deviation from the quantized transport due to backscattering (accompanied by spin-flips) between the helical channels. We discuss the details of the model, which takes into account the edge and bulk contribution to the total current and reproduces the experimental results.
△ Less
Submitted 10 October, 2019;
originally announced October 2019.
-
Latent Distribution Assumption for Unbiased and Consistent Consensus Modelling
Authors:
Valentina Fedorova,
Gleb Gusev,
Pavel Serdyukov
Abstract:
We study the problem of aggregation noisy labels. Usually, it is solved by proposing a stochastic model for the process of generating noisy labels and then estimating the model parameters using the observed noisy labels. A traditional assumption underlying previously introduced generative models is that each object has one latent true label. In contrast, we introduce a novel latent distribution as…
▽ More
We study the problem of aggregation noisy labels. Usually, it is solved by proposing a stochastic model for the process of generating noisy labels and then estimating the model parameters using the observed noisy labels. A traditional assumption underlying previously introduced generative models is that each object has one latent true label. In contrast, we introduce a novel latent distribution assumption, implying that a unique true label for an object might not exist, but rather each object might have a specific distribution generating a latent subjective label each time the object is observed. Our experiments showed that the novel assumption is more suitable for difficult tasks, when there is an ambiguity in choosing a "true" label for certain objects.
△ Less
Submitted 20 June, 2019;
originally announced June 2019.
-
Aggregation of pairwise comparisons with reduction of biases
Authors:
Nadezhda Bugakova,
Valentina Fedorova,
Gleb Gusev,
Alexey Drutsa
Abstract:
We study the problem of ranking from crowdsourced pairwise comparisons. Answers to pairwise tasks are known to be affected by the position of items on the screen, however, previous models for aggregation of pairwise comparisons do not focus on modeling such kind of biases. We introduce a new aggregation model factorBT for pairwise comparisons, which accounts for certain factors of pairwise tasks t…
▽ More
We study the problem of ranking from crowdsourced pairwise comparisons. Answers to pairwise tasks are known to be affected by the position of items on the screen, however, previous models for aggregation of pairwise comparisons do not focus on modeling such kind of biases. We introduce a new aggregation model factorBT for pairwise comparisons, which accounts for certain factors of pairwise tasks that are known to be irrelevant to the result of comparisons but may affect workers' answers due to perceptual reasons. By modeling biases that influence workers, factorBT is able to reduce the effect of biased pairwise comparisons on the resulted ranking. Our empirical studies on real-world data sets showed that factorBT produces more accurate ranking from crowdsourced pairwise comparisons than previously established models.
△ Less
Submitted 9 June, 2019;
originally announced June 2019.
-
Electronic thermal conductivity in 2D topological insulator in a HgTe quantum well
Authors:
G. M. Gusev,
Z. D. Kvon,
A. D. Levin,
E. B. Olshanetsky,
O. E. Raichev,
N. N. Mikhailovand,
S. A. Dvoretsky
Abstract:
We have measured the differential resistance in a two-dimensional topological insulator (2DTI) in a HgTe quantum well, as a function of the applied dc current. The transport near the charge neutrality point is characterized by a pair of counter propagating gapless edge modes. In the presence of an electric field, the energy is transported by counter propagating channels in the opposite direction.…
▽ More
We have measured the differential resistance in a two-dimensional topological insulator (2DTI) in a HgTe quantum well, as a function of the applied dc current. The transport near the charge neutrality point is characterized by a pair of counter propagating gapless edge modes. In the presence of an electric field, the energy is transported by counter propagating channels in the opposite direction. We test a hot carrier effect model and demonstrate that the energy transfer complies with the Wiedemann Franz law near the charge neutrality point in the edge transport regime.
△ Less
Submitted 29 January, 2019;
originally announced January 2019.
-
Thermoelectric transport in two-dimensional topological insulator state based on HgTe quantum well
Authors:
G. M. Gusev,
O. E. Raichev,
E. B. Olshanetsky,
A. D. Levin,
Z. D. Kvon,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
The thermoelectric response of HgTe quantum wells in the state of two-dimensional topological insulator (2D TI) has been studied experimentally. Ambipolar thermopower, typical for an electron-hole system, has been observed across the charge neutrality point, where the carrier type changes from electrons to holes according to the resistance measurements. The hole-type thermopower is much stronger t…
▽ More
The thermoelectric response of HgTe quantum wells in the state of two-dimensional topological insulator (2D TI) has been studied experimentally. Ambipolar thermopower, typical for an electron-hole system, has been observed across the charge neutrality point, where the carrier type changes from electrons to holes according to the resistance measurements. The hole-type thermopower is much stronger than the electron-type one. The thermopower linearly increases with temperature. We present a theoretical model which accounts for both the edge and bulk contributions to the electrical conductivity and thermoelectric effect in a 2D TI, including the effects of edge to bulk leakage. The model, contrary to previous theoretical studies, demonstrates that the 2D TI is not expected to show anomalies of thermopower near the band conductivity threshold, which is consistent with our experimental results. Based on the experimental data and theoretical analysis, we conclude that the observed thermopower is mostly of the bulk origin, while the resistance is determined by both the edge and bulk transport.
△ Less
Submitted 21 December, 2018;
originally announced December 2018.
-
Thermopower of a Two-Dimensional Semimetal in a HgTe Quantum Well
Authors:
E. B. Olshanetsky,
Z. D. Kvon,
M. V. Entin,
L. I. Magarill,
A. Levin,
G. M. Gusev,
N. N. Mikhailov
Abstract:
The thermopower in a two-dimensional semimetal existing in HgTe quantum wells 18-21 nm thick has been studied experimentally and theoretically for the first time. It has been found theoretically and experimentally that the thermopower has two components - diffusion and phonon drag and that the second component is several times larger than the first. It has been concluded that the electron-hole sca…
▽ More
The thermopower in a two-dimensional semimetal existing in HgTe quantum wells 18-21 nm thick has been studied experimentally and theoretically for the first time. It has been found theoretically and experimentally that the thermopower has two components - diffusion and phonon drag and that the second component is several times larger than the first. It has been concluded that the electron-hole scattering plays an important role in both mechanisms of the thermopower.
△ Less
Submitted 19 December, 2018;
originally announced December 2018.
-
Viscous transport and Hall viscosity in a two-dimensional electron system
Authors:
G. M. Gusev,
A. D. Levin,
E. V. Levinson,
A. K. Bakarov
Abstract:
Hall viscosity is a nondissipative response function describing momentum transport in two-dimensional (2D) systems with broken time-reversal symmetry. In the classical regime, Hall viscosity contributes to the viscous flow of 2D electrons in the presence of a magnetic field. We observe a pronounced, negative Hall resistivity at low magnetic field in a mesoscopic size, two-dimensional electron syst…
▽ More
Hall viscosity is a nondissipative response function describing momentum transport in two-dimensional (2D) systems with broken time-reversal symmetry. In the classical regime, Hall viscosity contributes to the viscous flow of 2D electrons in the presence of a magnetic field. We observe a pronounced, negative Hall resistivity at low magnetic field in a mesoscopic size, two-dimensional electron system, which is attributed to Hall viscosity in the inhomogeneous charge flow. Experimental results supported by a theoretical analysis confirm that the conditions for observation of Hall viscosity are correlated with predictions.
△ Less
Submitted 15 October, 2018;
originally announced October 2018.
-
Vorticity induced negative nonlocal resistance in viscous two-dimensional electron system
Authors:
A. D. Levin,
G. M. Gusev,
E. V. Levinson,
Z. D. Kvon,
A. K. Bakarov
Abstract:
We report non-local electrical measurements in a mesoscopic size two-dimensional (2D) electron gas in a GaAs quantum well in a hydrodynamic regime. Viscous electric flow is expected to be dominant when electron-electron collisions occur more often than the impurity or phonon scattering events. We observe a negative nonlocal resistance and attribute it to the formation of whirlpools in the electron…
▽ More
We report non-local electrical measurements in a mesoscopic size two-dimensional (2D) electron gas in a GaAs quantum well in a hydrodynamic regime. Viscous electric flow is expected to be dominant when electron-electron collisions occur more often than the impurity or phonon scattering events. We observe a negative nonlocal resistance and attribute it to the formation of whirlpools in the electron flow. We use the different nonlocal transport geometries and compare the results with a theory demonstrating the significance of hydrodynamics in mesoscopic samples.
△ Less
Submitted 15 October, 2018;
originally announced October 2018.
-
Online Evaluation for Effective Web Service Development
Authors:
Roman Budylin,
Alexey Drutsa,
Gleb Gusev,
Pavel Serdyukov,
Igor Yashkov
Abstract:
Development of the majority of the leading web services and software products today is generally guided by data-driven decisions based on evaluation that ensures a steady stream of updates, both in terms of quality and quantity. Large internet companies use online evaluation on a day-to-day basis and at a large scale. The number of smaller companies using A/B testing in their development cycle is…
▽ More
Development of the majority of the leading web services and software products today is generally guided by data-driven decisions based on evaluation that ensures a steady stream of updates, both in terms of quality and quantity. Large internet companies use online evaluation on a day-to-day basis and at a large scale. The number of smaller companies using A/B testing in their development cycle is also growing. Web development across the board strongly depends on quality of experimentation platforms. In this tutorial, we overview state-of-the-art methods underlying everyday evaluation pipelines at some of the leading Internet companies. Software engineers, designers, analysts, service or product managers --- beginners, advanced specialists, and researchers --- can learn how to make web service development data-driven and do it effectively.
△ Less
Submitted 3 September, 2018;
originally announced September 2018.
-
A survey on Deep Learning Advances on Different 3D Data Representations
Authors:
Eman Ahmed,
Alexandre Saint,
Abd El Rahman Shabayek,
Kseniya Cherenkova,
Rig Das,
Gleb Gusev,
Djamila Aouada,
Bjorn Ottersten
Abstract:
3D data is a valuable asset the computer vision filed as it provides rich information about the full geometry of sensed objects and scenes. Recently, with the availability of both large 3D datasets and computational power, it is today possible to consider applying deep learning to learn specific tasks on 3D data such as segmentation, recognition and correspondence. Depending on the considered 3D d…
▽ More
3D data is a valuable asset the computer vision filed as it provides rich information about the full geometry of sensed objects and scenes. Recently, with the availability of both large 3D datasets and computational power, it is today possible to consider applying deep learning to learn specific tasks on 3D data such as segmentation, recognition and correspondence. Depending on the considered 3D data representation, different challenges may be foreseen in using existent deep learning architectures. In this work, we provide a comprehensive overview about various 3D data representations highlighting the difference between Euclidean and non-Euclidean ones. We also discuss how Deep Learning methods are applied on each representation, analyzing the challenges to overcome.
△ Less
Submitted 6 April, 2019; v1 submitted 4 August, 2018;
originally announced August 2018.
-
Optically-detected long-lived spin coherence in multilayer systems: Double and triple quantum wells
Authors:
S. Ullah,
G. M. Gusev,
A. K. Bakarov,
F. G. G. Hernandez
Abstract:
In this contribution, we investigated the spin coherence of high-mobility dense two-dimensional electron gases confined in multilayer systems. The dynamics of optically-induced spin polarization was experimentally studied employing the time-resolved Kerr rotation and resonant spin amplification techniques. For both the double and triple quantum wells doped beyond the metal-insulator transition, wh…
▽ More
In this contribution, we investigated the spin coherence of high-mobility dense two-dimensional electron gases confined in multilayer systems. The dynamics of optically-induced spin polarization was experimentally studied employing the time-resolved Kerr rotation and resonant spin amplification techniques. For both the double and triple quantum wells doped beyond the metal-insulator transition, where the spin coherence is greatly suppressed, we found remarkably long spin lifetimes limited by the Dyakonov-Perel mechanism and spin hopping process between the donor sites as well as the spread of ensemble g-factor. The double quantum well structure yields a spin lifetime of 6.25 ns at T = 5 K while the triple quantum well shows a spin lifetime exceeding 25 ns at T = 8 K.
△ Less
Submitted 8 June, 2018;
originally announced June 2018.
-
Viscous electron flow in mesoscopic two-dimensional electron gas
Authors:
G. M. Gusev,
A. D. Levin,
E. V. Levinson,
A. K. Bakarov
Abstract:
We report electrical and magneto transport measurements in mesoscopic size, two-dimensional (2D) electron gas in a GaAs quantum well. Remarkably, we find that the probe configuration and sample geometry strongly affects the temperature evolution of local resistance. We attribute all transport properties to the presence of hydrodynamic effects. Experimental results confirm the theoretically predict…
▽ More
We report electrical and magneto transport measurements in mesoscopic size, two-dimensional (2D) electron gas in a GaAs quantum well. Remarkably, we find that the probe configuration and sample geometry strongly affects the temperature evolution of local resistance. We attribute all transport properties to the presence of hydrodynamic effects. Experimental results confirm the theoretically predicted significance of viscous flow in mesoscopic devices.
△ Less
Submitted 11 June, 2018; v1 submitted 26 February, 2018;
originally announced February 2018.
-
Tailoring multilayer quantum wells for spin devices
Authors:
S. Ullah,
G. M. Gusev,
A. K. Bakarov,
F. G. G. Hernandez
Abstract:
The electron spin dynamics in multilayer GaAs/AlGaAs quantum wells, containing high-mobility dense two-dimensional electron gases, have been studied using time-resolved Kerr rotation and resonant spin amplification techniques. The electron spin dynamics was regulated through the wave function engineering and quantum confinement in multilayer quantum wells. We observed the spin coherence with a rem…
▽ More
The electron spin dynamics in multilayer GaAs/AlGaAs quantum wells, containing high-mobility dense two-dimensional electron gases, have been studied using time-resolved Kerr rotation and resonant spin amplification techniques. The electron spin dynamics was regulated through the wave function engineering and quantum confinement in multilayer quantum wells. We observed the spin coherence with a remarkably long dephasing time T2* > 13 ns for the structure doped beyond metal-insulator transition. Dyakonov-Perel spin relaxation mechanism, as well as the inhomogeneity of electron g-factor, was suggested as the major limiting factors for the spin coherence time. In the metallic regime, we found that the electron-electron collisions become dominant over microscopic scattering on the electron spin relaxation with the Dyakonov-Perel mechanism. Furthermore, the data analysis indicated that in our structure, due to the spin relaxation anisotropy, Dyakonov-Perel spin relaxation mechanism is efficient for the spins oriented in-plane and suppressed along the quantum well growth direction resulting in the enhancement of T2*. Our findings, namely, long-lived spin coherence persisting up to about room temperature, spin polarization decay time with and without a magnetic field, the spin-orbit field, single electron relaxation time, transport scattering time, and the electron-electron Coulomb scattering time highlight the attractiveness of n-doped multilayer systems for spin devices.
△ Less
Submitted 13 January, 2018;
originally announced January 2018.
-
Robustness of spin coherence of the exciton bound to neutral donor states in multilayer system
Authors:
S. Ullah,
F. C. D. Moraes,
G. M. Gusev,
A. K. Bakarov,
F. G. G. Hernandez
Abstract:
We address the temperature influence on the precessional motion of electron spins under transverse magnetic field, studied in a GaAs/AlGaAs triple quantum wells, using pump-probe Kerr rotation. In the presence of an applied in-plane magnetic field the TRKR measurements show the robustness of carrier's spin polarization against temperature which can be easily traced in an extended range up to 250 K…
▽ More
We address the temperature influence on the precessional motion of electron spins under transverse magnetic field, studied in a GaAs/AlGaAs triple quantum wells, using pump-probe Kerr rotation. In the presence of an applied in-plane magnetic field the TRKR measurements show the robustness of carrier's spin polarization against temperature which can be easily traced in an extended range up to 250 K. By tuning the pump-probe wavelength to the exciton bound to a neutral donor transition, we observed a remarkably long-lasting spin coherence (with dephasing time T2* > 14 ns) limited by the spin hopping process and exchange interaction between the donor sites as well as the ensemble spread of g-factor. The temperature dependent spin dephasing time revealed a double linear dependence due to the different relaxation mechanisms active at respective temperature ranges. We observed that the increase of sample temperature from 5 K to 250 K, leads to a strong T2* reduction by almost 98%/97% for the excitation wavelengths of 823/821 nm. Furthermore, we noticed that the temperature increase not only causes the reduction of spin lifetime but can also lead to the variation of electron g-factor. Additionally, the spin dynamics was studied through the dependencies on the applied magnetic field and optical pump power.
△ Less
Submitted 17 May, 2018; v1 submitted 13 January, 2018;
originally announced January 2018.
-
Two dimensional topological insulator in quantizing magnetic fields
Authors:
E. B. Olshanetsky,
Z. D. Kvon,
G. M. Gusev,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
The effect of quantizing magnetic field on the electron transport is investigated in a two dimensional topological insulator (2D TI) based on a 8 nm (013) HgTe quantum well (QW). The local resistance behavior is indicative of a metal-insulator transition at $B\approx 6$ T. On the whole the experimental data agrees with the theory according to which the helical edge states transport in a 2D TI pers…
▽ More
The effect of quantizing magnetic field on the electron transport is investigated in a two dimensional topological insulator (2D TI) based on a 8 nm (013) HgTe quantum well (QW). The local resistance behavior is indicative of a metal-insulator transition at $B\approx 6$ T. On the whole the experimental data agrees with the theory according to which the helical edge states transport in a 2D TI persists from zero up to a critical magnetic field $B_c$ after which a gap opens up in the 2D TI spectrum.
△ Less
Submitted 6 December, 2017;
originally announced December 2017.
-
Robust helical edge transport at $ν=0$ quantum Hall state
Authors:
G. M. Gusev,
D. A. Kozlov,
A. D. Levin,
Z. D. Kvon,
N. N. Mikhailov,
S. A. Dvoretsky
Abstract:
Among the most interesting predictions in two-dimensional materials with a Dirac cone is the existence of the zeroth Landau level (LL), equally filled by electrons and holes with opposite chirality. The gapless edge states with helical spin structure emerge from Zeeman splitting at the LL filling factor $ν=0$ gapped quantum Hall state. We present observations of a giant nonlocal four-terminal tran…
▽ More
Among the most interesting predictions in two-dimensional materials with a Dirac cone is the existence of the zeroth Landau level (LL), equally filled by electrons and holes with opposite chirality. The gapless edge states with helical spin structure emerge from Zeeman splitting at the LL filling factor $ν=0$ gapped quantum Hall state. We present observations of a giant nonlocal four-terminal transport in zero-gap HgTe quantum wells at the $ν=0$ quantum Hall state. Our experiment clearly demonstrates the existence of the robust helical edge state in a system with single valley Dirac cone materials.
△ Less
Submitted 24 July, 2017;
originally announced July 2017.
-
CatBoost: unbiased boosting with categorical features
Authors:
Liudmila Prokhorenkova,
Gleb Gusev,
Aleksandr Vorobev,
Anna Veronika Dorogush,
Andrey Gulin
Abstract:
This paper presents the key algorithmic techniques behind CatBoost, a new gradient boosting toolkit. Their combination leads to CatBoost outperforming other publicly available boosting implementations in terms of quality on a variety of datasets. Two critical algorithmic advances introduced in CatBoost are the implementation of ordered boosting, a permutation-driven alternative to the classic algo…
▽ More
This paper presents the key algorithmic techniques behind CatBoost, a new gradient boosting toolkit. Their combination leads to CatBoost outperforming other publicly available boosting implementations in terms of quality on a variety of datasets. Two critical algorithmic advances introduced in CatBoost are the implementation of ordered boosting, a permutation-driven alternative to the classic algorithm, and an innovative algorithm for processing categorical features. Both techniques were created to fight a prediction shift caused by a special kind of target leakage present in all currently existing implementations of gradient boosting algorithms. In this paper, we provide a detailed analysis of this problem and demonstrate that proposed algorithms solve it effectively, leading to excellent empirical results.
△ Less
Submitted 20 January, 2019; v1 submitted 28 June, 2017;
originally announced June 2017.
-
Riemannian Optimization for Skip-Gram Negative Sampling
Authors:
Alexander Fonarev,
Oleksii Hrinchuk,
Gleb Gusev,
Pavel Serdyukov,
Ivan Oseledets
Abstract:
Skip-Gram Negative Sampling (SGNS) word embedding model, well known by its implementation in "word2vec" software, is usually optimized by stochastic gradient descent. However, the optimization of SGNS objective can be viewed as a problem of searching for a good matrix with the low-rank constraint. The most standard way to solve this type of problems is to apply Riemannian optimization framework to…
▽ More
Skip-Gram Negative Sampling (SGNS) word embedding model, well known by its implementation in "word2vec" software, is usually optimized by stochastic gradient descent. However, the optimization of SGNS objective can be viewed as a problem of searching for a good matrix with the low-rank constraint. The most standard way to solve this type of problems is to apply Riemannian optimization framework to optimize the SGNS objective over the manifold of required low-rank matrices. In this paper, we propose an algorithm that optimizes SGNS objective using Riemannian optimization and demonstrates its superiority over popular competitors, such as the original method to train SGNS and SVD over SPPMI matrix.
△ Less
Submitted 26 April, 2017;
originally announced April 2017.
-
Gate control of the spin mobility through the modification of the spin-orbit interaction in two-dimensional systems
Authors:
M. Luengo-Kovac,
F. C. D. Moraes,
G. J. Ferreira,
A. S. L. Ribeiro,
G. M. Gusev,
A. K. Bakarov,
V. Sih,
F. G. G. Hernandez
Abstract:
Spin drag measurements were performed in a two-dimensional electron system set close to the crossed spin helix regime and coupled by strong intersubband scattering. In a sample with uncommon combination of long spin lifetime and high charge mobility, the drift transport allows us to determine the spin-orbit field and the spin mobility anisotropies. We used a random walk model to describe the syste…
▽ More
Spin drag measurements were performed in a two-dimensional electron system set close to the crossed spin helix regime and coupled by strong intersubband scattering. In a sample with uncommon combination of long spin lifetime and high charge mobility, the drift transport allows us to determine the spin-orbit field and the spin mobility anisotropies. We used a random walk model to describe the system dynamics and found excellent agreement for the Rashba and Dresselhaus couplings. The proposed two-subband system displays a large tuning lever arm for the Rashba constant with gate voltage, which provides a new path towards a spin transistor. Furthermore, the data shows large spin mobility controlled by the spin-orbit constants setting the field along the direction perpendicular to the drift velocity. This work directly reveals the resistance experienced in the transport of a spin-polarized packet as a function of the strength of anisotropic spin-orbit fields.
△ Less
Submitted 18 May, 2017; v1 submitted 24 March, 2017;
originally announced March 2017.
-
Large anisotropic spin relaxation time of exciton bound to donor states in triple quantum wells
Authors:
S. Ullah,
G. M. Gusev,
A. K. Bakarov,
F. G. G. Hernandez
Abstract:
We have studied the spin dynamics of a dense two-dimensional electron gas confined in a GaAs/AlGaAs triple quantum well by using time-resolved Kerr rotation and resonant spin amplification. Strong anisotropy of the spin relaxation time up to a factor of 10 was found between the electron spins oriented in-plane and out-of-plane when the excitation energy is tuned to an exciton bound to neutral dono…
▽ More
We have studied the spin dynamics of a dense two-dimensional electron gas confined in a GaAs/AlGaAs triple quantum well by using time-resolved Kerr rotation and resonant spin amplification. Strong anisotropy of the spin relaxation time up to a factor of 10 was found between the electron spins oriented in-plane and out-of-plane when the excitation energy is tuned to an exciton bound to neutral donor transition. We model this anisotropy using an internal magnetic field and the inhomogeneity of the electron g-factor. The data analysis allows us to determine the direction and magnitude of this internal field in the range of a few mT for our studied structure, which decreases with the sample temperature and optical power. The dependence of the anisotropic spin relaxation was directly measured as a function of several experimental parameters: excitation wavelength, sample temperature, pump-probe time delay, and pump power.
△ Less
Submitted 2 May, 2017; v1 submitted 21 March, 2017;
originally announced March 2017.