Applying generative neural networks for fast simulations of the ALICE (CERN) experiment
Authors:
Maksymilian Wojnar
Abstract:
This thesis investigates the application of state-of-the-art advances in generative neural networks for fast simulation of the Zero Degree Calorimeter (ZDC) neutron detector in the ALICE experiment at CERN. Traditional simulation methods using the GEANT Monte Carlo toolkit, while accurate, are computationally demanding. With increasing computational needs at CERN, efficient simulation techniques a…
▽ More
This thesis investigates the application of state-of-the-art advances in generative neural networks for fast simulation of the Zero Degree Calorimeter (ZDC) neutron detector in the ALICE experiment at CERN. Traditional simulation methods using the GEANT Monte Carlo toolkit, while accurate, are computationally demanding. With increasing computational needs at CERN, efficient simulation techniques are essential. The thesis provides a comprehensive literature review on the application of neural networks in computer vision, fast simulations using machine learning, and generative neural networks in high-energy physics. The theory of the analyzed models is also discussed, along with technical aspects and the challenges associated with a practical implementation. The experiments evaluate various neural network architectures, including convolutional neural networks, vision transformers, and MLP-Mixers, as well as generative frameworks such as autoencoders, generative adversarial networks, vector quantization models, and diffusion models. Key contributions include the implementation and evaluation of these models, a significant improvement in the Wasserstein metric compared to existing methods with a low generation time of 5 milliseconds per sample, and the formulation of a list of recommendations for developing models for fast ZDC simulation. Open-source code and detailed hyperparameter settings are provided for reproducibility. Additionally, the thesis outlines future research directions to further enhance simulation fidelity and efficiency.
△ Less
Submitted 10 July, 2024;
originally announced July 2024.
FTMRate: Collision-Immune Distance-based Data Rate Selection for IEEE 802.11 Networks
Authors:
Wojciech Ciezobka,
Maksymilian Wojnar,
Katarzyna Kosek-Szott,
Szymon Szott,
Krzysztof Rusek
Abstract:
Data rate selection algorithms for Wi-Fi devices are an important area of research because they directly impact performance. Most of the proposals are based on measuring the transmission success probability for a given data rate. In dense scenarios, however, this probing approach will fail because frame collisions are misinterpreted as erroneous data rate selection. We propose FTMRate which uses t…
▽ More
Data rate selection algorithms for Wi-Fi devices are an important area of research because they directly impact performance. Most of the proposals are based on measuring the transmission success probability for a given data rate. In dense scenarios, however, this probing approach will fail because frame collisions are misinterpreted as erroneous data rate selection. We propose FTMRate which uses the fine timing measurement (FTM) feature, recently introduced in IEEE 802.11. FTM allows stations to measure their distance from the AP. We argue that knowledge of the distance from the receiver can be useful in determining which data rate to use. We apply statistical learning (a form of machine learning) to estimate the distance based on measurements, estimate channel quality from the distance, and select data rates based on channel quality. We evaluate three distinct estimation approaches: exponential smoothing, Kalman filter, and particle filter. We present a performance evaluation of the three variants of FTMRate and show, in several dense and mobile (though line-of-sight only) scenarios, that it can outperform two benchmarks and provide close to optimal results in IEEE 802.11ax networks.
△ Less
Submitted 9 August, 2023; v1 submitted 20 April, 2023;
originally announced April 2023.
