检索结果-内蒙古大学图书馆

arXiv 2021年

作者： Duan, Yuanzhi Hu, Xiaofang Zhou, Yue Liu, Qiang Duan, Shukai College of Artificial Intelligence Southwest University China College of Electronics and Information Engineering Southwest University China Brain-inspired Computing & Intelligent Control of Chongqing Key Lab China Harbin Institute of Technology China

Neural network pruning has remarkable performance for reducing the complexity of deep network models. Recent network pruning methods usually focused on removing unimportant or redundant filters in the network. In this paper, by exploring the similarities between feature maps, we propose a novel filter pruning method-Central Filter (CF), which suggests that a filter is approximately equal to a set of other filters after appropriate adjustments. Our method is based on the discovery that the average similarity between feature maps changes very little, regardless of the number of input images. Based on this finding, we establish similarity graphs on feature maps and calculate the closeness centrality of each node to select the Central Filter. Moreover, we design a method to directly adjust weights in the next layer corresponding to the Central Filter, effectively minimizing the error caused by pruning. Through experiments on various benchmark networks and datasets, CF yields state-of-the-art performance. For example, with ResNet-56, CF reduces approximately 39.7% of FLOPs by removing 47.1% of the parameters, with even 0.33% accuracy improvement on CIFAR-10. With GoogLeNet, CF reduces approximately 63.2% of FLOPs by removing 55.6% of the parameters, with only a small loss of 0.35% in top-1 accuracy on CIFAR-10. With ResNet-50, CF reduces approximately 47.9% of FLOPs by removing 36.9% of the parameters, with only a small loss of 1.07% in top-1 accuracy on ImageNet. The codes can be availab.e at https://***/8ubpshLR23/Central-Filter. Copyright © 2021, The Authors. All rights reserved.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Cavity-Quantum Electrodynamics with Moiré Flatband Photonic Crystals

arXiv

引用

arXiv 2024年

作者： Wang, Yu-Tong Ye, Qi-Hang Yan, Jun-Yong Qiao, Yufei Chen, Chen Cheng, Xiao-Tian Li, Chen-Hui Zhang, Zi-Jian Huang, Cheng-Nian Meng, Yun Zou, Kai Zhan, Wen-Kang Zhao, Chao Hu, Xiaolong Tee, Clarence Augustine T.H. Sha, Wei E.I. Huang, Zhixiang Liu, Huiyun Jin, Chao-Yuan Ying, Lei Liu, Feng State Key Laboratory of Extreme Photonics and Instrumentation College of Information Science and Electronic Engineering Zhejiang University Hangzhou310027 China School of Physics Zhejiang Key Laboratory of Micro-nano Quantum Chips and Quantum Control Zhejiang University Hangzhou310027 China School of Precision Instrument and Optoelectronic Engineering Tianjin University Tianjin300072 China Key Laboratory of Optoelectronic Information Science and Technology Ministry of Education Tianjin300072 China Laboratory of Solid State Optoelectronics Information Technology Institute of Semiconductors Chinese Academy of Sciences Beijing100083 China College of Materials Science and Opto-Electronic Technology University of Chinese Academy of Science Beijing101804 China College of Physics and Electrical Information Engineering Zhejiang Normal University Hangzhou310058 China Key Laboratory of Intelligent Computing and Signal Processing Ministry of Education Anhui University Hefei230039 China Department of Electronic and Electrical Engineering University College London LondonWC1E 7JE United Kingdom International Joint Innovation Center Zhejiang University Haining314400 China

Quantum emitters are a key component in photonic quantum technologies. Enhancing their single-photon emission by engineering the photonic environment using cavities can improve the overall efficiency in quantum information processing. However, this enhancement is often limited by the need for precise nanoscale control over emitter’s position within micro/nano-cavities. Inspired by the fascinated physics of moiré pattern, we present an approach to strongly modify the spontaneous emission rate of a quantum emitter using fine-designed multilayer moiré photonic crystal with a robust isolate-flatband dispersion. Theoretical analysis reveals that due to its nearly infinite photonic density of states, the moiré cavity can simultaneously possess high Purcell factor and large tolerance over emitter’s position. We experimentally demonstrate the coupling between this moiré cavity and a quantum dot via the cavity-determined polarization of the dot’s emission. The radiative lifetime of the quantum dot can be tuned by a factor of 40 from 42 ps to 1692 ps, attributed to strong Purcell enhancement and Purcell inhibition effects. Our findings pave the way for moiré flatband cavity-enhanced quantum light sources, quantum optical switches and quantum nodes for quantum internet. Copyright © 2024, The Authors. All rights reserved.

关键词： Particle beams

来源：评论

学校读者我要写书评

暂无评论

Customized Conversational Recommender Systems

arXiv

引用

arXiv 2022年

作者： Li, Shuokai Zhu, Yongchun Xie, Ruobing Tang, Zhenwei Zhang, Zhao Zhuang, Fuzhen He, Qing Xiong, Hui Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China University of Chinese Academy of Sciences Beijing100049 China WeChat Search Application Department Tencent China King Abdullah University of Science and Technology Thuwal Saudi Arabia Institute of Artificial Intelligence Beihang University Beijing100191 China Sklsde School of Computer Science Beihang University Beijing100191 China Artificial Intelligence Thrust The Hong Kong University of Science and Technology Guangzhou China The Key Lab of Intelligent Information Processing of Chinese Academy of Sciences China

Conversational recommender systems (CRS) aim to capture user's current intentions and provide recommendations through real-time multi-turn conversational interactions. As a human-machine interactive system, it is essential for CRS to improve the user experience. However, most CRS methods neglect the importance of user experience. In this paper, we propose two key points for CRS to improve the user experience: (1) Speaking like a human, human can speak with different styles according to the current dialogue context. (2) Identifying fine-grained intentions, even for the same utterance, different users have diverse finegrained intentions, which are related to users' inherent preference. Based on the observations, we propose a novel CRS model, coined Customized Conversational Recommender System (CCRS), which customizes CRS model for users from three perspectives. For human-like dialogue services, we propose multi-style dialogue response generator which selects context-aware speaking style for utterance generation. To provide personalized recommendations, we extract user's current fine-grained intentions from dialogue context with the guidance of user's inherent preferences. Finally, to customize the model parameters for each user, we train the model from the meta-learning perspective. Extensive experiments and a series of analyses have shown the superiority of our CCRS on both the recommendation and dialogue service. © 2022, CC BY-SA.

关键词： Recommender systems

来源：评论

学校读者我要写书评

暂无评论

A Tale of HodgeRank and Spectral Method: Target Attack Against Rank Aggregation Is the Fixed Point of Adversarial Game

arXiv

引用

arXiv 2022年

作者： Ma, Ke Xu, Qianqian Zeng, Jinshan Li, Guorong Cao, Xiaochun Huang, Qingming The School of Computer Science and Technology University of Chinese Academy of Sciences Beijing100049 China The Key Laboratory of Intelligent Information Processing Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China The School of Computer and Information Engineering Jiangxi Normal University Jiangxi Nanchang330022 China School of Cyber Science and Technology Shenzhen Campus Sun Yat-sen University Shenzhen518107 China University of Chinese Academy of Sciences Beijing100049 China Peng Cheng Laboratory Shenzhen518055 China

Rank aggregation with pairwise comparisons has shown promising results in elections, sports competitions, recommendations, and information retrieval. However, little attention has been paid to the security issue of such algorithms, in contrast to numerous research work on the computational and statistical characteristics. Driven by huge profit, the potential adversary has strong motivation and incentives to manipulate the ranking list. Meanwhile, the intrinsic vulnerability of the rank aggregation methods is not well studied in the literature. To fully understand the possible risks, we focus on the purposeful adversary who desires to designate the aggregated results by modifying the pairwise data in this paper. From the perspective of the dynamical system, the attack behavior with a target ranking list is a fixed point belonging to the composition of the adversary and the victim. To perform the targeted attack, we formulate the interaction between the adversary and the victim as a game-theoretic framework consisting of two continuous operators while Nash equilibrium is established. Then two procedures against HodgeRank and RankCentrality are constructed to produce the modification of the original data. Furthermore, we prove that the victims will produce the target ranking list once the adversary masters the complete information. It is noteworthy that the proposed methods allow the adversary only to hold incomplete information or imperfect feedback and perform the purposeful attack. The effectiveness of the suggested target attack strategies is demonstrated by a series of toy simulations and several real-world data experiments. These experimental results show that the proposed methods could achieve the attacker’s goal in the sense that the leading candidate of the perturbed ranking list is the designated one by the adversary. Copyright © 2022, The Authors. All rights reserved.

关键词： Dynamical systems

来源：评论

学校读者我要写书评

暂无评论

Closed-Loop Safe Correction for Reinforcement Learning Policy 4th

Closed-Loop Safe Correction for Reinforcement Learning Poli...

引用

4th International Conference on Ubiquitous Security, UbiSec 2024

作者： Yi, Zhi Lv, Qi Chen, Shuhong Liang, Ying Dai, Yinglong Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing College of Information Science and Engineering Hunan Normal University Changsha410081 China School of Computer Science and Cyber Engineering Guangzhou University Guangzhou510006 China Blockchain Innovation Lab Swinburne University of Technology MelbourneVIC3122 Australia Molecular Nutrition Branch National Engineering Research Center of Rice and By-product Deep Processing College of Food Science and Engineering Central South University of Forestry and Technology Hunan Changsha410004 China

ISBN: (纸本)9789819648351

Trial and error learning is an approach with uncertain consequences. How to maintain policy security, stability, and efficiency under controlled circumstances, posing a significant academic challenge. Such as Reinforcement Learning (RL) leverages the neural network parameter space to identify the optimal solution for achieving the optimal mapping of the policy space. However, even a slight fluctuation in the parameter space can lead to policy collapse. To solve it, we propose a method called Closed-loop Safe Correction (CLSC). This method employs twin agents to establish a closed-loop framework for policy and sampling data distribution, gradually improving policy efficiency while ensuring policy safety. Base on theoretical and experimental analysis, CLSC elucidates the performance, influencing factors, and advantages of this approach in policy safety scenarios. These findings provide novel research insights and practical directions for advancing safe RL. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

The Existence Theory of the Nonlinear Discrete System and the Convergence Theory of the Gummel Iteration for the PNP Equations

引用

Journal of Scientific computing 2025年第1期104卷

作者： Liu, Yang Shu, Shi Yang, Ying Hunan Key Laboratory for Computation and Simulation in Science and Engineering Key Laboratory of Intelligent Computing and Information Processing of Ministry of Education Xiangtan University Hunan Xiangtan411105 China School of Mathematics Hunan Institute of Science and Technology Hunan Yueyang414006 China Guilin University of Electronic Technology Guangxi Guilin541004 China

The Poisson-Nernst-Planck equations play a crucial role in modeling ion mass conservation and electrostatic diffusion in various applications, such as biological ion channels, semiconductor devices, and nanopore systems. Due to the inherent strong coupling and nonlinearity, developing robust theoretical frameworks for discrete schemes and associated fast algorithms is challenging. While the finite element method and some improved methods are widely used to solve these equations, existing theoretical analyses are insufficient, particularly regarding the existence of solutions for discrete methods or nonlinear discrete systems, and the convergence of fast algorithms. This study aims to contribute to this area by proposing a framework that includes establishing the existence of solutions for an abstract discrete scheme, which encompasses schemes such as standard finite element and edge-averaged finite element methods. We establish the existence theory for the nonlinear discrete system by carefully constructing a suitable compact convex set for the fixed-point operator and then using the Brouwer’s fixed-point theorem. Additionally, we present a convergence theory for the Gummel iteration, a commonly used iteration for the Poisson-Nernst-Planck equations, associated with our abstract discrete scheme. We further explore the theoretical results for both standard and edge-averaged finite element schemes as applications of the framework. Numerical experiments are conducted to verify the theoretical results for the Gummel iteration of these two schemes. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.

关键词： Fixed point arithmetic

来源：评论

学校读者我要写书评

暂无评论

Group-wise Inhibition based Feature Regularization for Robust Classification

Group-wise Inhibition based Feature Regularization for Robus...

引用

International Conference on Computer Vision (ICCV)

作者： Haozhe Liu Haoqian Wu Weicheng Xie Feng Liu Linlin Shen Computer Vision Institute College of Computer Science and Software Engineering SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society National Engineering Laboratory for Big Data System Computing Technology Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China

The convolutional neural network (CNN) is vulnerable to degraded images with even very small variations (e.g. corrupted and adversarial samples). One of the possible reasons is that CNN pays more attention to the most discriminative regions, but ignores the auxiliary features when learning, leading to the lack of feature diversity for final judgment. In our method, we propose to dynamically suppress significant activation values of CNN by group-wise inhibition, but not fixedly or randomly handle them when training. The feature maps with different activation distribution are then processed separately to take the feature independence into account. CNN is finally guided to learn richer discriminative features hierarchically for robust classification according to the proposed regularization. Our method is comprehensively evaluated under multiple settings, including classification against corruptions, adversarial attacks and low data regime. Extensive experimental results show that the proposed method can achieve significant improvements in terms of both robustness and generalization performances, when compared with the state-of-the-art methods. Code is availab.e at https://***/LinusWu/TENET_Training.

关键词： Training Computer vision Codes Heuristic algorithms Robustness Classification algorithms Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

An Image Denoising Method Based on Deep Residual GAN 4

An Image Denoising Method Based on Deep Residual GAN

引用

2020 4th International Workshop on Advanced Algorithms and Control Engineering, IWAACE 2020

作者： Wang, Ziyuan Wang, Lidan Duan, Shukai Li, Yunfei Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing Chongqing400715 China Schoolof Electronic and Information Engineering Southwest University Chongqing400715 China College of Artificiall Intelligence Southwest University Chongqing400715 China Braininspired Computing and Intelligent Control of Chongqing Key Lab Chongqing400715 China National and Local Joint Engineering Laboratory of Intelligent Transmission and Control Technology Chongqing400715 China Chongqing Brain Science Collaborative Innovation Center Chongqing400715 China

As people come into contact with image data more often, high quality and clear images attract more attention. Many methods have been proposed to deal with image noise problem including deep learning (DL). However most of them is lack of capability when customers want more perceptual details of the image without information loss. In this paper, a deep residual network based on generative adversarial (GAN) network was proposed to complete the image denoising mission. Firstly, a generative-adversarial network structure based on residual blocks was designed. Secondly, a refined loss function was given to train the GAN network. The well designed loss function can help the generated image to be very close to the clear counterpart (ground truth) while enhancing more details in colours and brightness. Finally, extensive experiments show that our network is not only convincing for images denoising, but also effective for other image process tasks, such as image defogging, medical CT denoising etc., presenting impressive and competitive effects. © Published under licence by IOP Publishing Ltd.

关键词： Image denoising

来源：评论

学校读者我要写书评

暂无评论

A Novel Economical Design of Multistage Decimators for Low Latency Audio Applications 22

A Novel Economical Design of Multistage Decimators for Low L...

引用

22nd IEEE International Conference on High Performance computing and Communications, 18th IEEE International Conference on Smart City and 6th IEEE International Conference on Data Science and Systems, HPCC-SmartCity-DSS 2020

作者： Hu, Wei Zheng, Chongwei Wang, Yonghao Liu, Jing Chen, Cen Zeng, Zeng Wuhan University of Science and Technology College of Computer Science and Technology China Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System Wuhan China Birmingham City University Digital Media Technology Lab Birmingham United Kingdom Institute for Infocomm Research Singapore

ISBN: (纸本)9781728176499

The oversampling sigma-delta analog-To-digital converter (ADC) is widely used in digital audio systems for its low hardware cost and high precision. The digital filter part of the ADC is the primary source of the silicon and power cost. In this work, we proposed an economical multistage decimation filter architecture to meet the requirements of commercial ADC specifications with a further novel multi-objective optimization strategy. We took group delay and computational cost as objective functions and used annealing algorithm to readjust the stopband attenuation, stopband edge and passband ripple of each filter. Our multistage architecture combines cascaded integrator-comb (CIC) filters, compensating filters and equiripple finite impulse response (FIR) filters at the different stages to balance the overall performance. The experimental results showed that our design approach worked with multi-objective optimization strategy could achieve low group delay with lower cost compared to traditional designs. © 2020 IEEE.

关键词： Group delay

来源：评论

学校读者我要写书评

暂无评论

Constructing Sample-to-Class Graph for Few-Shot Class-Incremental Learning

SSRN

引用

SSRN 2024年

作者： Hu, Fuyuan Zhang, Jian Lyu, Fan Zhao, Linglan Liu, Chengyan Zhang, Tao Yu, Baoqing School of Electronics and Information Engineering Suzhou University of Science and Technology No.99 Xuefu Road Jiangsu Suzhou215000 China Jiangsu Industrial Intelligent Low Carbon Technology Engineering Center Jiangsu Suzhou215009 China Suzhou Key Laboratory of Intelligent Low Carbon Technology Application Jiangsu Suzhou215009 China New Laboratory of Pattemn Recognition Institute of Automation Chinese Academy of Sciences College of Intelligence and Computing No.95 Zhongguancun East Road Beijing100190 China Tencent Youtu Lab Shanghai201600 China Seventh People’s Hospital Shanghai University of Traditional Chinese Medicine No.358 Datong Road Pudong New District Shanghai200137 China

Few-shot class-incremental learning (FSCIL) aims to build machine learning models that can continually learn novel classes from a few training samples, without forgetting knowledge of old classes. The challenges of FSCIL lie in the limited data of novel classes, which not only lead to significant overfitting issues but also exacerbate the notorious catastrophic forgetting problems. As proved in early studies, building sample relationships is beneficial for learning from few-shot classes. In this paper, we advance the idea to the incremental scenario and propose a Sample-to-Class (S2C) graph learning method for FSCIL. Concretely, we propose a Sample-level Graph Network (SGN) that focuses on analyzing sample relationships within a single session. This network helps aggregate similar samples, ultimately leading to the extraction of more refined class-level features. Then, we present a Class-level Graph Network (CGN) that establishes connections across class-level features of both novel and old classes. This network plays a crucial role in linking the knowledge between different sessions and helps improve overall learning performance in the FSCIL scenario. Moreover, we design a multi-stage strategy for training the S2C model, which mitigates the training challenges posed by limited data in the incremental learning process. The multi-stage training strategy is designed to build an S2C graph from base to incremental sessions and improve the capacity via an extra pseudo-incremental stage. Extensive experiments on three popular benchmark datasets show that our method outperforms the baselines and sets new state-of-the-art results in FSCIL. Our codes will be availab.e upon acceptance. © 2024, The Authors. All rights reserved.

关键词： Contrastive Learning

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：