检索结果-内蒙古大学图书馆

Resetting induced multimodality

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Pogorzelec, Przemyslaw Dybiec, Bartlomiej Doctoral School of Exact and Natural Sciences Faculty of Physics Astronomy and Applied Computer Science Jagiellonian University Lojasiewicza 11 Kraków30-348 Poland Institute of Theoretical Physics Mark Kac Center for Complex Systems Research Faculty of Physics Astronomy and Applied Computer Science Jagiellonian University Lojasiewicza 11 Kraków30-348 Poland

Properties of stochastic systems are defined by the noise type and deterministic forces acting on the system. In out-of-equilibrium setups, e.g., for motions under action of Lévy noises, the existence of the stationary state is not only determined by the potential but also by the noise. Potential wells need to be steeper than parabolic in order to assure existence of stationary states. The existence of stationary states, in sub-harmonic potential wells, can be restored by stochastic resetting, which is the protocol of starting over at random times. Herein we demonstrate that the combined action of Lévy noise and Poissonian stochastic resetting can result in the phase transition between non-equilibrium stationary states of various multimodality in the overdamped system in super-harmonic potentials. Fine-tuned resetting rates can increase the modality of stationary states, while for high resetting rates the multimodality is destroyed as the stochastic resetting limits the spread of particles. © 2023, CC BY.

关键词： Stochastic systems

Multilayer CVD graphene electrodes using a transfer-free process for the next generation of optically transparent and MRI-compatible neural interfaces

学校读者我要写书评

暂无评论

Microsystems & Nanoengineering 2022年第5期8卷 183-196页

作者： Nasim Bakhshaee Babaroud Merlin Palmar Andrada Iulia Velea Chiara Coletti Sebastian Weingärtner Frans Vos Wouter A.Serdijn Sten Vollebregt Vasiliki Giagka Department of Microelectronics Faculty of Electrical EngineeringMathematics and Computer ScienceDelft University of TechnologyMekelweg 4Delft 2628 CDThe Netherlands Technologies for Bioelectronics Group Department of System Integration and Interconnection TechnologiesFraunhofer Institute for Reliability and Micro-integration IZMGustav-Meyer-Allee 25Berlin 13355Germany Department of Imaging Physics Faculty of Applied ScienceDelft University of TechnologyLorentzweg 1Delft 2628 CJThe Netherlands Erasmus University Medical Center(Erasmus MC) dr.Molewaterplein 40Rotterdam 3015 GDThe Netherlands

Multimodal platforms combining electrical neural recording and stimulation,optogenetics,optical imaging,and magnetic resonance(MRI)imaging are emerging as a promising platform to enhance the depth of characterization in neuroscientific *** conductive,optically transparent,and MRI-compatible electrodes can optimally combine all *** as a suitable electrode candidate material can be grown via chemical vapor deposition(CVD)processes and sandwiched between transparent biocompatible ***,due to the high graphene growth temperature(≥900℃)and the presence of polymers,fabrication is commonly based on a manual transfer process of pre-grown graphene sheets,which causes reliability *** this paper,we present CVD-based multilayer graphene electrodes fabricated using a wafer-scale transfer-free process for use in optically transparent and MRI-compatible neural *** fabricated electrodes feature very low impedances which are comparable to those of noble metal electrodes of the same size and *** also exhibit the highest charge storage capacity(CSC)reported to date among all previously fabricated CVD graphene *** graphene electrodes did not reveal any photo-induced artifact during 10-Hz light pulse ***,we show here,for the first time,that CVD graphene electrodes do not cause any image artifact in a 3T MRI *** results demonstrate that multilayer graphene electrodes are excellent candidates for the next generation of neural interfaces and can substitute the standard conventional metal *** fabricated graphene electrodes enable multimodal neural recording,electrical and optogenetic stimulation,while allowing for optical imaging,as well as,artifact-free MRI studies.

关键词： transparent transfer process

An improved KNN algorithm based on ensemble methods and correlation

学校读者我要写书评

暂无评论

An improved KNN algorithm based on ensemble methods and corr...

International Conference on Information science and Technology (ICIST)

作者： Youness Manzali Khalidou Abdoulaye Barry Mohamed El Far Applied Physics Computer Science and Statistics Laboratory Faculty of Sciences Dhar El Mahraz Sidi Mohammed Ben Abdellah University Fez Morocco

K-Nearest Neighbors is a widely used algorithm due to its simplicity and efficacity. However, KNN suffers from many drawbacks, such as it does not work well with datasets with a high number of features. Also, not all the features contribute to the classification process. To resolve these issues, we present an improved KNN algorithm which uses KNN as a base learner in an ensemble method and correlation for selecting the features subsets. The experiment results show that the proposed algorithm performed better than other machine learning algorithms.

关键词：

DFDFTr + RAC-DAC: Dense Fusion with Deep Fusion Transformer and Realistic Actor-Critic combined with Decoupled Actor-Critic for end-to-end continuous multi-task robotic arm control

学校读者我要写书评

暂无评论

Expert Systems with Applications 2025年 276卷

作者： Dang, Thi Phuc Nguyen, Truong Chinh Dau, Sy Hieu Department of Computer Science Faculty of Information Technology Industrial University of Ho Chi Minh City 12 Nguyen Van Bao Ho Chi Minh7000 Viet Nam Department of Applied Physics Faculty of Applied Science University of Technology - Viet Nam National University HCMC Ho Chi Minh City To Hien Thanh Ho Chi Minh7000 Viet Nam

In this paper, we propose novel methods to address the challenges of intelligent robotic arm operations in complex environments. Our approach includes a deep learning method for 6D object pose estimation and a deep reinforcement learning method for controlling complex tasks. For 6D object pose estimation, we introduce the Dense Fusion combined with Deep Fusion Transformer model (DFDFTr). This method integrates Dense Fusion and Deep Fusion Transformer, applies re-parameterization techniques such as Reparameterized Convolution and Batch Normalization Fusion, and incorporates specialized neural networks for each stage of the model. Experimental results on the LineMOD and Occlusion LineMOD datasets demonstrate that our method outperforms one-stage approaches like DenseFusion and MPF6D in accuracy and achieves results comparable to advanced two-stage methods such as PVN3D and DTr. Leveraging re-parameterization, our method maintains competitive inference speed with significantly fewer parameters than PVN3D and MPF6D, PoseCNN+ICP without sacrificing accuracy. Additionally, it efficiently processes multiple objects, handling 8 objects in 0.1 s compared to 0.07 s for a single object. For continuous task control, we propose a deep reinforcement learning approach using Realistic Actor-Critic combined with Decoupled Actor-Critic (RAC-DAC), integrated with techniques such as Relay Hindsight Experience Replay (RHER), Multi-step Hindsight Experience Replay (MHER), Curiosity-Driven Prioritization (CDP), and Projected Conflicting Gradients (PCGrad). Experimental results show that RAC-DAC-RHER achieves significantly higher success rates than DDPG-RHER and TD3-RHER, particularly in challenging tasks such as object grasping. It also ensures efficient learning and high accuracy. Furthermore, by refining hyperparameters and integrating MHER, CDP, PCGrad, MLP, and the Clipping Mechanism, we enhanced learning performance, mitigated instabilities, and enabled higher learning rates in spars

关键词： Robotic arms

Evaluating Optimal Reference Translations

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zouhar, Vilém Kloudová, Věra Popel, Martin Bojar, Ondřej ETH Zürich Department of Computer Science Switzerland Institute of Formal and Applied Linguistics Faculty of Mathematics and Physics Charles University Czech Republic

The overall translation quality reached by current machine translation (MT) systems for high-resourced language pairs is remarkably good. Standard methods of evaluation are not suitable nor intended to uncover the many translation errors and quality deficiencies that still persist. Furthermore, the quality of standard reference translations is commonly questioned and comparable quality levels have been reached by MT alone in several language pairs. Navigating further research in these high-resource settings is thus difficult. In this article, we propose a methodology for creating more reliable document-level human reference translations, called "optimal reference translations," with the simple aim to raise the bar of what should be deemed "human translation quality." We evaluate the obtained document-level optimal reference translations in comparison with "standard" ones, confirming a significant quality increase and also documenting the relationship between evaluation and translation editing. © 2023, CC BY.

关键词： Quality control

Superconductivity near the Mott-Ioffe-Regel limit in the high-entropy alloy superconductor (ScZrNb)1−x(RhPd)x with a CsCl-type lattice

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gutowska, Sylwia Kawala, Alicja Wiendlocha, Bartlomiej Faculty of Physics and Applied Computer Science AGH University of Science and Technology Aleja Mickiewicza 30 Krakow30-059 Poland Department of Quantum Many-Body Theory Jagiellonian University ul. prof. S. Lojasiewicza 11 Krakow30-348 Poland

Theoretical analysis of the electronic structure of the high-entropy-type superconductor (ScZrNb)1−x(RhPd)x, x ∈ (0.35,0.45) is presented. The studied material is a partially ordered CsCl-type structure, with two sublattices, randomly occupied by Sc, Zr, Nb (first sublattice) and Nb, Rh, and Pd (second sublattice). Calculations were done using the Korringa-Kohn-Rostoker method with the coherent potential approximation (KKR-CPA) and take into account the substitutional disorder. Our total energy calculations confirm the preference for the partially ordered structure over the fully random bcc-type one. Electronic densities of states N(E), dispersion relations, and McMillan-Hopfield parameters η (electronic contribution to electron-phonon coupling) are studied as a function of composition. The computed increasing trends in N(EF) and η with x are opposite to what we expected based on the experimental results, where the decrease in the critical temperature with increasing x was found. Very strong electron scattering due to disorder is observed, as the electronic dispersion relations are strongly smeared. As a result, the computed electronic lifetimes τ are very short, leading to a small mean-free path of electrons of the order of interatomic distance, which puts (ScZrNb)1−x(RhPd)x near the Mott-Ioffe-Regel limit. The trend in τ(x) is similar to the trend observed experimentally in Tc(x), suggesting that disorder may be the factor that influences Tc in this series of alloys. Copyright © 2023, The Authors. All rights reserved.

关键词： Crystal structure

Kerr Geodesics in horizon-penetrating Kerr coordinates: description in terms of Weierstrass functions

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Bakun, Zuzanna Lukanty, Angelika Untilova, Anastasiia Cieślik, Adam Mach, Patryk Faculty of Physics Astronomy and Applied Computer Science Jagiellonian University Lojasiewicza 11 Kraków30-348 Poland Institute of Theoretical Physics Jagiellonian University Lojasiewicza 11 Kraków30-348 Poland

We revisit the theory of timelike and null geodesics in the (extended) Kerr spacetime. This work is a sequel to a recent paper by Cieślik, Hackmann, and Mach, who applied the so-called Biermann–Weierstrass formula to integrate Kerr geodesic equations expressed in Boyer–Lindquist coordinates. We show that a formulation based on the Biermann–Weierstrass theorem can also be applied in horizon-penetrating Kerr coordinates, resulting in solutions that are smooth across Kerr horizons. Horizon-penetrating Kerr coordinates allow for an explicit continuation of timelike and null geodesics between appropriate regions of the maximal analytic extension of the Kerr spacetime. A part of this work is devoted to a graphic visualisation of such geodesics. Copyright © 2024, The Authors. All rights reserved.

关键词： Geodesy

Socio-cognitive Optimization of Time-delay Control Problems using Evolutionary Metaheuristics

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Kipinski, Piotr Guzowski, Hubert Urbanczyk, Aleksandra Smolka, Maciej Kisiel-Dorohinicki, Marek Byrski, Aleksander Oplatkova, Zuzana Kominkova Senkerik, Roman Pekar, Libor Matusu, Radek Gazdos, Frantisek Institute of Computer Science AGH University of Science and Technology Krakow Poland Faculty of Applied Informatics Tomas Bata University in Zlín Czech Republic

Metaheuristics are universal optimization algorithms which should be used for solving difficult problems, unsolvable by classic approaches. In this paper we aim at constructing novel socio-cognitive metaheuristic based on castes, and apply several versions of this algorithm to optimization of time-delay system model. Besides giving the background and the details of the proposed algorithms we apply them to optimization of selected variants of the problem and discuss the results. © 2022, CC BY.

关键词： Time delay

Insight into the structural and magnetotransport properties of epitaxial heterostructures α-Fe2O3/Pt(111): Role of the reversed layer sequence

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Koziol-Rachwal, A. Kwiatek, N. Skowroński, W. Grochot, K. Kanak, J. Madej, E. Freindl, K. Korecki, J. Spiridis, N. Faculty of Physics and Applied Computer Science AGH University of Science and Technology al. Mickiewicza 30 Kraków30-059 Poland Jerzy Haber Institute of Catalysis and Surface Chemistry Polish Academy of Sciences Kraków30-239 Poland Institute of Electronics AGH University of Science and Technology Al. Mickiewicza 30 Kraków30-059 Poland

We report on the chemical structure and spin Hall magnetoresistance (SMR) in epitaxial αFe2O3(hematite)(0001)/Pt(111) bilayers with hematite thicknesses of 6 nm and 15 nm grown by molecular beam epitaxy on a MgO(111) substrate. Unlike previous studies that involved Pt overlayers on hematite, the present hematite films were grown on a stable Pt buffer layer and displayed structural changes as a function of thickness. These structural differences (the presence of a ferrimagnetic phase in the thinner film) significantly affected the magnetotransport properties of the bilayers. We observed a sign change of the SMR from positive to negative when the thickness of hematite increased from 6 nm to 15 nm. For αFe2O3(15 nm)/Pt, we demonstrated room-temperature switching of the Néel order with rectangular, nondecaying switching characteristics. Such structures open the way to extending magnetotransport studies to more complex systems with double asymmetric metal/hematite/Pt interfaces. © 2023, CC BY.

关键词： Hematite