检索结果-内蒙古大学图书馆

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Yubin Shi Yixuan Chen Mingzhi Dong Xiaochen Yang Dongsheng Li Yujiang Wang Robert Dick Qin Lv Yingying Zhao Fan Yang Tun Lu Ning Gu Li Shang China and Shanghai Key Laboratory of Data Science School of Computer Science Fudan University School of Mathematics Statistics University of Glasgow Microsoft Research Asia Shanghai China Department of Engineering Science University of Oxford Department of Electrical Engineering and Computer Science University of Michigan Department of Computer Science University of Colorado Boulder School of Microelectronics Fudan University

Despite their prevalence in deep-learning communities, over-parameterized models convey high demands of computational costs for proper training. This work studies the fine-grained, modular-level learning dynamics of over-parameterized models to attain a more efficient and fruitful training strategy. Empirical evidence reveals that when scaling down into network modules, such as heads in self-attention models, we can observe varying learning patterns implicitly associated with each module's trainability. To describe such modular-level learning capabilities, we introduce a novel concept dubbed modular neural tangent kernel (mNTK), and we demonstrate that the quality of a module's learning is tightly associated with its mNTK's principal eigenvalue λmax. A large λmax indicates that the module learns features with better convergence, while those miniature ones may impact generalization negatively. Inspired by the discovery, we propose a novel training strategy termed Modular Adaptive Training (MAT) to update those modules with their λmax exceeding a dynamic threshold selectively, concentrating the model on learning common features and ignoring those superfluous ones. Unlike most existing training schemes with a complete BP cycle across all network modules, MAT can significantly save computations by its partially-updating strategy and can further improve performance. Experiments show that MAT nearly halves the computational cost of model training and outperforms the accuracy of baselines.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Research on APSO-WNN and its Application in Vibration Fault Diagnosis of Hydroelectric Generating Units

引用

Journal of the Chinese Society of Mechanical Engineers, Transactions of the Chinese Institute of Engineers, Series C/Chung-Kuo Chi Hsueh Kung Ch'eng Hsuebo Pao 2021年第2期42卷 163-172页

作者： Xu, Yan-Chun Xia, Hai-Ting Fang, Shao-Chen Lu, Mi Hubei Provincial Key Laboratory for Operation and Control of Cascaded Hydropower Station China Three Gorges University Yichang443001 China Department of Electrical & Computer Engineering Texas A& M University College Station91207 United States

Based on the research of wavelet neural network (WNN), an adaptive particle swarm optimization (APSO) is proposed to solve the complex nonlinear relationship between the vibration characteristics and fault types of hydropower systems. This algorithm combines characteristics of evolutionary compute and swarm intelligences. It can change inertia weight according to the states of the particle adaptively. APSO rises the training speed of wavelet neural network, and improves network training accuracy. Experiments indicate that wavelet neural network based on APSO contains a higher precision and faster speed of diagnosis, compared with back propagation (BP) neural network and wavelet neural network. The algorithm is a new method for fault diagnosis of hydroelectric generating units (HGU), and it can be effectively applied to practical engineering. © 2021, Chinese Mechanical engineering Society. All right reserved.

关键词： Failure analysis

来源：评论

学校读者我要写书评

暂无评论

Deep Neural Network Enhanced Chip-on-board Hybrid-Packaged Photonic-electrical Silicon Transceiver for High-speed Optical Interconnects

Deep Neural Network Enhanced Chip-on-board Hybrid-Packaged P...

引用

2024 Conference on Lasers and Electro-Optics, CLEO 2024

作者： Qin, Jun Yang, Qipeng Han, Changhao Zhou, Yan Zhang, Yunhao Wu, Yichen Sun, Yu Lu, Junde Liu, Siming Li, Yueqin Sun, Jian Min, Miao Hu, Weiwei He, Zhixue Wang, Lei Yu, Shaohua Shu, Haowen Wang, Xingjun Key Laboratory of the Ministry of Education for Optoelectronic Measurement Technology and Instrument Beijing Information Science and Technology University Beijing100192 China State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Peking University Beijing100871 China Department of Electrical and Computer Engineering University of California Santa BarbaraCA93106 United States Peking University Yangtze Delta Institute of Optoelectronics Nantong226010 China Peng Cheng Laboratory Shenzhen518055 China State Key Laboratory of Information Photonics and Optical Communications BUPT Beijing100876 China

We explore implementing a multilevel deep neural network to enhance the performance of a 4-channel photonic-electrical hybrid-packaged silicon transceiver. Stable transmission and reception of 150 Gbps/λ PAM4 signals... 详细信息

ISBN: (纸本)9781957171395

关键词： Optical interconnects

来源：评论

学校读者我要写书评

暂无评论

UHVDC Islanded Operation System Ultralow-Frequency Oscillation and Its Countermeasures

IEEE Canadian Journal of Electrical and Computer Engineering

引用

IEEE Canadian Journal of electrical and computer engineering 2021年第2期44卷 110-117页

作者： Li, Wei Gole, Aniruddha M. Kaffashan, Iman Xiao, Xiangning The Department of Electrical and Computer Engineering University of Manitoba ManitobaMBR3T 5V6 Canada The State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources North China Electric Power University Beijing102206 China

There exists serious frequency instability in the UHVDC islanded operation system consisting of a large capacity UHVDC and its corresponding power source with hydraulic power generators. Ultralowfrequency oscillation (ULFO) is one of the frequency instable problems, which will happen because of the inconsistent frequency control features between UHVDC and hydraulic power generators. Through the frequency-domain analysis, there exists a frequency band with negative damping in the hydraulic turbine control system, and UHVDC is negative damping for frequency deviation with constant power control mode. There are four negative damping contributing factors as hydrowater starting time constant, hydrogenerator governor forward gain, islanded system total inertia, and UHVDC power. These four factors are analyzed in this article in the frequency domain. Based on the detailed ULFO damping analysis, effective frequency control strategies for UHVDC islanded system are proposed in this article by withdrawing a part of primary frequency control functions of generators or coordinating the frequency controls between hydraulic generator governors and UHVDC frequency limitation controller (FLC). Through the proposed frequency strategies, islanded system frequency is mainly controlled by UHVDC FLC, which can successfully suppress the ULFO. Validation simulations are performed on the RTDS platform, and the simulation results prove the proposed strategies. © 2021 IEEE.

关键词： Frequency domain analysis

来源：评论

学校读者我要写书评

暂无评论

FaceBench: A Multi-View Multi-Level Facial Attribute VQA Dataset for Benchmarking Face Perception MLLMs

arXiv

引用

arXiv 2025年

作者： Wang, Xiaoqin Ma, Xusen Hou, Xianxu Ding, Meidan Li, Yudong Chen, Junliang Chen, Wenting Peng, Xiaoyang Shen, Linlin Computer Vision Institute College of Computer Science and Software Engineering Shenzhen University China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China Guangdong Provincial Key Laboratory of Intelligent Information Processing China AIAC Xi’an Jiaotong-Liverpool University China Tsinghua University China The Hong Kong Polytechnic University Hong Kong City University of Hong Kong Hong Kong Sun Yat-sen University China

Multimodal large language models (MLLMs) have demonstrated remarkable capabilities in various tasks. However, effectively evaluating these MLLMs on face perception remains largely unexplored. To address this gap, we introduce FaceBench, a dataset featuring hierarchical multi-view and multi-level attributes specifically designed to assess the comprehensive face perception abilities of MLLMs. Initially, we construct a hierarchical facial attribute structure, which encompasses five views with up to three levels of attributes, totaling over 210 attributes and 700 attribute values. Based on the structure, the proposed FaceBench consists of 49,919 visual question-answering (VQA) pairs for evaluation and 23,841 pairs for fine-tuning. Moreover, we further develop a robust face perception MLLM baseline, Face-LLaVA, by training with our proposed face VQA data. Extensive experiments on various mainstream MLLMs and Face-LLaVA are conducted to test their face perception ability, with results also compared against human performance. The results reveal that, the existing MLLMs are far from satisfactory in understanding the fine-grained facial attributes, while our Face-LLaVA significantly outperforms existing open-source models with a small amount of training data and is comparable to commercial ones like GPT-4o and Gemini. The dataset will be released at https://***/CVI-SZU/FaceBench. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Tunnel Junction-Enabled Monolithically Integrated GaN Micro-Light Emitting Transistor

arXiv

引用

arXiv 2024年

作者： Rahman, Sheikh Ifatur Awwad, Mohammad Joishi, Chandan Jamal-Eddine, Zane Gunning, Brendan Armstrong, Andrew Rajan, Siddharth Department of Electrical and Computer Engineering The Ohio State University ColumbusOH United States Sandia National Laboratory AlbuquerqueNM United States Department of Material Science and Engineering The Ohio State University ColumbusOH United States

GaN/InGaN microLEDs are a very promising technology for next generation displays. Switching control transistors and their integration are key components in achieving high-performance, efficient displays. Monolithic integration of microLEDs with GaN switching devices provides an opportunity to control microLED output power with capacitive (voltage) control rather than current controlled schemes. This approach can greatly reduce system complexity for the driver circuit arrays while maintaining device opto-electronic performance. In this work, we demonstrate a 3-terminal GaN micro-light emitting transistor that combines a GaN/InGaN blue tunneling-based microLED with a GaN n-channel FET. The integrated device exhibits excellent gate control, drain current control and optical emission control. This work provides a promising pathway for future monolithic integration of GaN FETs with microLED to enable fast switching high efficiency microLED display and communication systems. Copyright © 2024, The Authors. All rights reserved.

关键词： Transistors

来源：评论

学校读者我要写书评

暂无评论

Coherence resonance in fractional van der Pol oscillators

引用

The European Physical Journal B 2024年第4期97卷 1-7页

作者： Li, Shangyuan Wang, Zhongqiu Hao, Chenhang Yang, Jianhua Jiangsu Key Laboratory of Mine Mechanical and Electrical Equipment School of Mechatronic Engineering China University of Mining and Technology Xuzhou People’s Republic of China School of Computer Science and Technology China University of Mining and Technology Xuzhou People’s Republic of China

The phenomenon of coherence resonance (CR) in two typical kinds of fractional van del Pol oscillator is investigated. For the first model, the damping term of the ordinary van der Pol oscillator is replaced by a fractional-order damping. While in the second model, the fractional-order damping is instead of the inertia term of the ordinary van der Pol oscillator. In the first model, there is obvious CR by adjusting the noise intensity. The resonance frequency mainly depends on the value of the fractional-order but it is almost independent of the noise intensity when the noise intensity is small, but change with the noise intensity when the noise intensity lies in a slightly large range. However, the resonance frequency of the ordinary van der Pol oscillator does not influence by the noise. In addition, with the increase of the fractional-order, the resonance frequency decreases monotonicity. Whereas, the resonance amplitude is a nonmonotonic function of the fractional-order. When the fractional oscillator deviates from the ordinary one, the resonance amplitude is much greater. In the second model, CR occurs only when the fractional oscillator is very close to the ordinary van der Pol oscillator. Comparing CR in the first model with that in the second model, we find that the value of the fractional-order can be taken over a much larger range to induce CR for the first one. The study shows that the inertia term is an indispensable factor in causing CR phenomenon while the fractional-order damping only changes the resonance frequency in different kinds of van der Pol oscillators. Coherence resonance phenomenon in the fractional van der Pol oscillator under different values of the fractional-order and noise intensity.

关键词：

来源：评论

学校读者我要写书评

暂无评论

SEE-MCAM: Scalable Multi-Bit FeFET Content Addressable Memories for Energy Efficient Associative Search

SEE-MCAM: Scalable Multi-Bit FeFET Content Addressable Memor...

引用

IEEE International Conference on computer-Aided Design

作者： Shengxi Shou Che-Kai Liu Sanggeon Yun Zishen Wan Kai Ni Mohsen Imani X. Sharon Hu Jianyi Yang Cheng Zhuo Xunzhao Yin College of Information Science and Electronic Engineering Zhejiang University China Department of Information and Computer Science University of California Irvine USA School of Electrical and Computer Engineering Georgia Institute of Technology USA Department of Electrical and Microelectronic Engineering Rochester Institute of Technology USA Department of Computer Science and Engineering University of Notre Dame USA School of Micro-Nano Electronics. Zhejiang University China Key Laboratory of Collaborative Sensing and Autonomous Unmaned Systems of Zhejiang Province China

Artificial intelligence has made remarkable advancements in recent years, leading to the development of algorithms and models capable of handling ever-increasing amounts of data. The computational demands of these algorithms necessitate circuit and architecture designs that go beyond the von-Neumann paradigm. Content addressable memories (CAMs), which implement parallel associative search functionality within memory blocks to overcome the memory wall bottleneck, have proven to be effective for data-intensive tasks. While current CAM designs have achieved higher storage density and energy efficiency than their CMOS-based counterparts by leveraging emerging non-volatile memories (NVM), most of these implementations are limited to binary storage cells. In this work, we propose SEE-MCAM, scalable and compact multi-bit CAM (MCAM) designs that utilize the three-terminal ferroelectric FET (FeFET) as the proxy. By exploiting the multi-level-cell characteristics of FeFETs, our proposed SEE-MCAM designs enable multi-bit associative search functions and achieve better energy efficiency and performance than existing FeFET-based CAM designs. We validated the functionality of our proposed designs by achieving 3 bits per cell CAM functionality, resulting in 3x improvement in storage density. The area per bit of the proposed SEE-MCAM cell is 8% of the conventional CMOS CAM. We thoroughly investigated the scalability and robustness of the proposed design. Evaluation results suggest that the proposed 2FeFET-1 T SEE-MCAM achieves 9.8× more energy efficiency and 1.6× less search latency compared to the CMOS CAM, respectively. When compared to existing MCAM designs, the proposed SEE-MCAM can achieve 8.7× and 4.9× more energy efficiency than ReRAM-based and FeFET-based MCAMs, respectively. Benchmarking results show that our approach provides up to 3 orders of magnitude improvement in speedup and energy efficiency over a GPU implementation in accelerating a novel quantized hyperdimensiona

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancement of quantum excitation transport by photonic nonreciprocity

引用

Physical Review A 2022年第3期106卷 033501-033501页

作者： S. Ali Hassani Gangaraj Lei Ying Francesco Monticone Zongfu Yu Department of Electrical and Computer Engineering University of Wisconsin-Madison Madison Wisconsin 53706 USA Department of Physics Interdisciplinary Center for Quantum Information State Key Laboratory of Modern Optical Instrumentation and Zhejiang Province Key Laboratory of Quantum Technology and Devices Zhejiang University Hangzhou 310027 China School of Electrical and Computer Engineering Cornell University Ithaca New York 14853 USA

Enhanced interaction between two two-level emitters (e.g., atoms) by nonreciprocal photonic media can be of benefit to broad areas, from quantum information science to biological detection. Here we provide a detailed analysis on why nonreciprocal photon-mediated interaction enhances interatomic excitation transport efficiency. We investigate a system consisting of two two-level emitters embedded in a generic photonic environment. By comparing symmetric and asymmetric photon exchange, we analytically show that breaking electromagnetic reciprocity makes it possible for the cooperative decay rate to exceed the spontaneous decay rate, even in a translation-invariant homogeneous system. This means that the excitation of an emitter must decay mostly into the other emitter rather than leaking and dissipating into the reservoir photonic modes. We also provide an example where a chain of two-level emitters dominantly interacts via the reciprocal modes of a plasmonic waveguide. We then show that breaking reciprocity in such a system via driving a dc current through the plasmonic material can drastically increase the probability of photon emission from one emitter to another, leading to an order-of-magnitude enhancement in quantum energy-transport efficiency.

关键词： Photonics Quantum optics Surface plasmons

来源：评论

学校读者我要写书评

暂无评论

Learning a Better Control Barrier Function

Learning a Better Control Barrier Function

引用

IEEE Conference on Decision and Control

作者： Bolun Dai Prashanth Krishnamurthy Farshad Khorrami Electrical & Computer Engineering Department Control/Robotics Research Laboratory Tandon School of Engineering New York University Brooklyn NY

ISBN: (数字)9781665467612

ISBN: (纸本)9781665467629

Control barrier functions (CBFs) are widely used in safety-critical controllers. However, constructing a valid CBF is challenging, especially under nonlinear or non-convex constraints and for high relative degree systems. Meanwhile, finding a conservative CBF that only recovers a portion of the true safe set is usually possible. In this work, starting from a "conservative" handcrafted CBF (HCBF), we develop a method to find a CBF that recovers a reasonably larger portion of the safe set. Since the learned CBF controller is not guaranteed to be safe during training iterations, we use a model predictive controller (MPC) to ensure safety during training. Using the collected trajectory data containing safe and unsafe interactions, we train a neural network to estimate the difference between the HCBF and a CBF that recovers a closer solution to the true safe set. With our proposed approach, we can generate safe controllers that are less conservative and computationally more efficient. We validate our approach on two systems: a second-order integrator and a ball-on-beam.

关键词： Training Heuristic algorithms Neural networks Predictive models Data collection Data models Safety

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：