Zobrazeno 1 - 10
of 912
pro vyhledávání: '"Szatkowski, P."'
Continual learning (CL) has emerged as a critical area in machine learning, enabling neural networks to learn from evolving data distributions while mitigating catastrophic forgetting. However, recent research has identified the stability gap -- a ph
Externí odkaz:
http://arxiv.org/abs/2411.04723
Static sparse training aims to train sparse models from scratch, achieving remarkable results in recent years. A key design choice is given by the sparse initialization, which determines the trainable sub-network through a binary mask. Existing metho
Externí odkaz:
http://arxiv.org/abs/2406.01755
Autor:
Szatkowski, Filip, Yang, Fei, Twardowski, Bartłomiej, Trzciński, Tomasz, van de Weijer, Joost
Continual learning is crucial for applications in dynamic environments, where machine learning models must adapt to changing data distributions while retaining knowledge of previous tasks. Despite significant advancements, catastrophic forgetting - w
Externí odkaz:
http://arxiv.org/abs/2403.07404
Autor:
Bombiński, Przemysław, Szatkowski, Patryk, Sobieski, Bartłomiej, Kwieciński, Tymoteusz, Płotka, Szymon, Adamek, Mariusz, Banasiuk, Marcin, Furmanek, Mariusz I., Biecek, Przemysław
Lung mask creation lacks well-defined criteria and standardized guidelines, leading to a high degree of subjectivity between annotators. In this study, we assess the underestimation of lung regions on chest X-ray segmentation masks created according
Externí odkaz:
http://arxiv.org/abs/2402.11510
Autor:
Litwin, Przemyslaw, Wronski, Jakub, Markowski, Konrad, Lopez-Mago, Dorilian, Masajada, Jan, Szatkowski, Mateusz
The need set by a computational industry to increase processing power, while simultaneously reducing the energy consumption of data centers became a challenge for modern computational systems. In this work, we propose an optical communication solutio
Externí odkaz:
http://arxiv.org/abs/2401.03521
Transformer models can face practical limitations due to their high computational requirements. At the same time, such models exhibit significant activation sparsity, which can be leveraged to reduce the inference cost by converting parts of the netw
Externí odkaz:
http://arxiv.org/abs/2310.04361
Publikováno v:
Scientific Reports, Vol 14, Iss 1, Pp 1-18 (2024)
Abstract The article presents the results of a three-year field study that was conducted in Poland to evaluate the yield and quality of seeds and oil from traditional (SAM) and canola-quality white mustard (SAC) in response to different N fertilizer
Externí odkaz:
https://doaj.org/article/b8ecc55e754443c6aeaad88a97fd7046
Autor:
Szatkowski, Filip, Pyla, Mateusz, Przewięźlikowski, Marcin, Cygert, Sebastian, Twardowski, Bartłomiej, Trzciński, Tomasz
In this work, we investigate exemplar-free class incremental learning (CIL) with knowledge distillation (KD) as a regularization strategy, aiming to prevent forgetting. KD-based methods are successfully used in CIL, but they often struggle to regular
Externí odkaz:
http://arxiv.org/abs/2308.09544
Autor:
Baniecki, Hubert, Sobieski, Bartlomiej, Szatkowski, Patryk, Bombinski, Przemyslaw, Biecek, Przemyslaw
Publikováno v:
Artificial Intelligence in Medicine, vol. 159, 103026, 2025
Time-to-event prediction, e.g. cancer survival analysis or hospital length of stay, is a highly prominent machine learning task in medical and healthcare applications. However, only a few interpretable machine learning methods comply with its challen
Externí odkaz:
http://arxiv.org/abs/2303.09817
Implicit Neural Representations (INRs) are nowadays used to represent multimedia signals across various real-life applications, including image super-resolution, image compression, or 3D rendering. Existing methods that leverage INRs are predominantl
Externí odkaz:
http://arxiv.org/abs/2302.04959