检索结果-内蒙古大学图书馆

A Quantitative Approach to Understand Self-Supervised Models as Cross-lingual Feature Extractors

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Shuyue Stella Xu, Beining Zhang, Xiangyu Liu, Hexin Chao, Wenhan Garcia, Leibny Paola Center for Language and Speech Processing Johns Hopkins University United States School of Computer Science and Engineering Beihang University China School of Electrical and Electronic Engineering Nanyang Technological University Singapore

In this work, we study the features extracted by English self-supervised learning (SSL) models in cross-lingual contexts and propose a new metric to predict the quality of feature representations. Using automatic speech recognition (ASR) as a downstream task, we analyze the effect of model size, training objectives, and model architecture on the models’ performance as a feature extractor for a set of topologically diverse corpora. We develop a novel metric, the Phonetic-Syntax Ratio (PSR), to measure the phonetic and synthetic information in the extracted representations using deep generalized canonical correlation analysis. Results show the contrastive loss in the wav2vec2.0 objective facilitates more effective cross-lingual feature extraction. There is a positive correlation between PSR scores and ASR performance, suggesting that phonetic information extracted by monolingual SSL models can be used for downstream tasks in cross-lingual settings. The proposed metric is an effective indicator of the quality of the representations and can be useful for model selection.1 © 2023, CC BY.

关键词： Topology

A Multi-Token Coordinate Descent Method for Semi-Decentralized Vertical Federated Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Valdeira, Pedro Chi, Yuejie Soares, Cláudia Xavier, João CMU IST United States CMU United States NOVA IST Department of Electrical and Computer Engineering Carnegie Mellon University United States Institute for Systems and Robotics Instituto Superior Técnico Portugal Department of Computer Science NOVA School of Science and Technology Portugal

Communication efficiency is a major challenge in federated learning (FL). In client-server schemes, the server constitutes a bottleneck, and while decentralized setups spread communications, they do not necessarily reduce them due to slower convergence. We propose Multi-Token Coordinate Descent (MTCD), a communication-efficient algorithm for semi-decentralized vertical federated learning, exploiting both client-server and client-client communications when each client holds a small subset of features. Our multi-token method can be seen as a parallel Markov chain (block) coordinate descent algorithm and it subsumes the client-server and decentralized setups as special cases. We obtain a convergence rate of O(1/T) for nonconvex objectives when tokens roam over disjoint subsets of clients and for convex objectives when they roam over possibly overlapping subsets. Numerical results show that MTCD improves the state-of-the-art communication efficiency and allows for a tunable amount of parallel communications. © 2023, CC BY.

关键词： Efficiency

Exploring the Robustness of Decentralized Training for Large Language Models

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Lu, Lin Dai, Chenxi Tao, Wangcheng Yuan, Binhang Sun, Yanan Zhou, Pan Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science of Technology Wuhan China School of Cyber Science and Engineering Huazhong University of Science of Technology Wuhan China Hong Kong University of Science of Technology Hong Kong College of Computer Science Sichuan University Sichuan China

Decentralized training of large language models has emerged as an effective way to democratize this technology. However, the potential threats associated with this approach have not been carefully discussed, which would hinder the development of decentralized training infrastructures. This paper aims to initiate discussion towards this end by exploring the robustness of decentralized training from three main perspectives. First, we demonstrate the vulnerabilities inherent in decentralized training frameworks in terms of hardware, data, and models. Second, we highlight the fundamental difference between decentralized foundation model training and vanilla federated learning, where the security techniques employed in federated learning cannot be applied directly. Third, we discuss the essential components required for a robust and efficient decentralized training framework and present a case study by modeling a concrete threat model. Our objective in this vision paper is to emphasize the importance of addressing security concerns in the context of decentralized training for large language models. Copyright © 2023, The Authors. All rights reserved.

关键词： Computational linguistics

Improving Sharpness-Aware Minimization with Fisher Mask for Better Generalization on Language Models

学校读者我要写书评

暂无评论

Improving Sharpness-Aware Minimization with Fisher Mask for ...

2022 Findings of the Association for Computational Linguistics: EMNLP 2022

作者： Zhong, Qihuang Ding, Liang Shen, Li Mi, Peng Liu, Juhua Du, Bo Tao, Dacheng National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University China JD Explore Academy China School of Informatics Xiamen University China Research Center for Graphic Communication Printing and Packaging Institute of Artificial Intelligence Wuhan University China

Fine-tuning large pretrained language models on a limited training corpus usually suffers from poor generalization. Prior works show that the recently-proposed sharpness-aware minimization (SAM) optimization method can improve the model generalization. However, SAM adds a perturbation to each model parameter equally (but not all parameters contribute equally to the optimization of training), which we argue is sub-optimal and will lead to excessive computation. In this paper, we propose a novel optimization procedure, namely FSAM, which introduces a Fisher mask to improve the efficiency and performance of SAM. In short, instead of adding perturbation to all parameters, FSAM uses the Fisher information to identity the important parameters and formulates a Fisher mask to obtain the sparse perturbation, i.e., making the optimizer focus on these important parameters. Experiments on various tasks in GLUE and SuperGLUE benchmarks show that FSAM consistently outperforms the vanilla SAM by 0.67∼1.98 average score among four different pretrained models. We also empirically show that FSAM works well in other complex scenarios, e.g., fine-tuning on generation tasks or limited training data. Encouragingly, when training data is limited, FSAM improves the SAM by a large margin, i.e., up to 15.1. © 2022 Association for Computational Linguistics.

关键词： Fisher information matrix

Evolving temporal convolutional neural networks for modeling turntable servo systems

学校读者我要写书评

暂无评论

Applied Soft Computing 2025年 172卷

作者： Xie, Cheng Xue, Bing Zhang, Mengjie Chen, Songlin Liu, Yang Control and Simulation Center School of Astronautics Harbin Institute of Technology Heilongjiang Harbin150001 China Centre for Data Science and Artificial Intelligence & School of Engineering and Computer Science Victoria University of Wellington Wellington6140 New Zealand Shenyang Institute of Automation Chinese Academy of Sciences Liaoning Shenyang110169 China

Turntable servo systems are important experimental equipment used for semi-physical simulation and testing of aircraft, which have very strict requirements for tracking performance. To achieve high-precision servo performance, some advanced model-based control methods have been used recently, but the control performance is greatly influenced by the accuracy of the model. Therefore, this paper proposes evolving temporal convolutional networks (TCNs) for accurately modeling turntable servo systems. Considering the advantages of TCNs in capturing local dependencies in sequence data and parallel computing, a TCN is adopted to compensate the unmodeled dynamics to significantly improve the accuracy of the existing dynamics model. Then, an evolutionary algorithm (EA) based architecture optimization algorithm for the above TCN model is proposed, and the network performance is continuously improved by simulating the population evolution in nature. Finally, the weight inheritance method is utilized to evaluate the performance of individuals during the evolution process to effectively improve the execution efficiency of the whole algorithm. In addition, the subnets inherited from the supernet meet the real-time requirements of the actual tasks by setting real-time constraints on the supernet. The experimental results demonstrate the superiority of the proposed model against the peer competitors in terms of the prediction performance. In addition, the proposed optimization algorithm also shows higher execution efficiency as verified by ablation experiments. © 2025 Elsevier B.V.

关键词： Evolutionary algorithms (EAs) System identification and modeling Temporal convolutional networks (TCNs) Turntable servo systems

TPE for JPEG images with Dynamic M-ary Decomposition and Adaptive Threshold Constraints

学校读者我要写书评

暂无评论

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Ma, Yakun Chai, Xiuli Long, Guoqiang Gan, Zhihua Zhang, Yushu Henan University School of Artificial Intelligence Henan Engineering Research Center for Industrial Internet of Things Zhengzhou450046 China Henan University School of Software Intelligent Data Processing Engineering Research Center of Henan Province Institute of Intelligent Network System Kaifeng475004 China Nanjing University of Aeronautics and Astronautics College of Computer Science and Technology Nanjing211106 China

Traditional JPEG image encryption that prioritizes solely confidentiality fails to account for the pressing usability requirements of cloud-based environments, thus boosting the boom in thumbnail-preserving encryption (TPE) to balance image privacy and usability. However, existing TPE schemes for JPEG images face numerous challenges, such as insufficient security, inability to achieve lossless decryption, and high file extension. To address these challenges, we propose a TPE scheme based on dynamic M-ary decomposition and adaptive threshold constraints (TPE-MDTC). First, the valid ranges of quantized DC coefficients for JPEG images are determined. Then, a sum-preserving encryption method for quantized DC coefficients with compliance threshold constraints is designed using the bit-plane permutation to preserve thumbnails with high accuracy. Next, the introduction of dynamic M-ary decomposition effectively changes bit statistical characteristics preserved by bit-plane permutation, enhancing the ciphertext security. Finally, a quantized AC encryption method with RV (Run/Value) pair global permutation is proposed, effectively modifying the unit block features, thereby significantly improving the security and attack resistance of encrypted images. Experimental results show that the proposed TPE-MDTC scheme can reconstruct the original JPEG images without loss, and the generated ciphertext images exhibit significant advantages over previous schemes regarding file extension and security. © 1991-2012 IEEE.

关键词： Ciphertext

The Monte Carlo Computational Summit - October 25 & 26, 2023 - Notre Dame, Indiana, USA

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Morgan, Joanna Piper Mote, Alexander Pasmann, Samuel Lee Ridley, Gavin Palmer, Todd Niemeyer, Kyle E. McClarren, Ryan Center for Exascale Monte Carlo Neutron Transport School of Mechanical Industrial and Manufacturing Engineering Oregon State University United States School of Electrical Engineering and Computer Science Oregon State University United States Department of Aerospace and Mechanical Engineering University of Notre Dame United States Department of Nuclear Science and Engineering Massachusetts Institute of Technology United States School of Nuclear Science and Engineering Oregon State University United States

The Monte Carlo Computational Summit was held on the campus of the University of Notre Dame in South Bend, Indiana, USA on 25-26 October 2023. The goals of the summit were to discuss algorithmic and software alterations required for successfully porting respective code bases to exascale-class computing hardware, compare software engineering techniques used by various code teams, and consider the adoption of industry-standard benchmark problems to better facilitate code-to-code performance comparisons. A large portion of the meeting included candid discussions of direct experiences with approaches that have and have not worked. Participants reported that identifying and implementing suitable Monte Carlo algorithms for GPUs continues to be a sticking point. They also report significant difficulty porting existing algorithms between GPU APIs (specifically Nvidia CUDA to AMD ROCm). To better compare code-to-code performance, participants decided to design a C5G7-like benchmark problem with a defined figure of merit, with the expectation of adding more benchmarks in the future. Problem specifications and results will eventually be hosted in a public repository and will be open to submissions by all Monte Carlo transport codes capable of running the benchmark problem. The participants also identified the need to explore the intermediate and long-term future of the Monte Carlo neutron transport community and how best to modernize and contextualize Monte Carlo as a useful tool in modern industry. Overall the summit was considered to be a success by the organizers and participants, and the group shared a strong desire for future, potentially larger, Monte Carlo summits. © 2024, CC BY.

关键词： Graphics processing unit

A Method for Removing Complex Visible Watermarks of Electronic Map Based on Conditional Generative Adversarial Nets 29

学校读者我要写书评

暂无评论

A Method for Removing Complex Visible Watermarks of Electron...

29th International Conference on Geoinformatics, Geoinformatics 2022

作者： Baode, Jiang Shaofen, Xu Liufeng, Tao Zhong, Xie National Engineering Research Center of Geographic Information System China University of Geosciences Wuhan430074 China School of Computer Science China University of Geosciences Wuhan430074 China State Key Laboratory of Geo-information Engineering Xi'an710054 China

ISBN: (数字)9798350309881

ISBN: (纸本)9798350309881

With the rapid development and extensive application of the electronic map technology, complex visible watermarks have been widely used in e-maps and map tiles delivered by map service providers to protect copyright. To evaluate the robustness and effectiveness of visible watermarks, many researches on watermark removal have attracted much attention in recent years. However, most of the existing watermark removal methods have the problems of watermark residue or repaired area being indistinct and unreal. To address these difficulties, this paper presented a deep neural network model based on conditional generative adversarial networks (cGANs) to remove complex visible watermark for electronic map tiles. The model is mainly consisted of a generator and a discriminator. The generator adopts a U-Net structure with jump connections to generate watermark-free maps based on watermarked maps, and the discriminator uses a full convolutional network based on region discrimination model to distinguish the differences between the generated watermark-free maps and real watermark-free ones. Specifically, a perceptual loss function was introduced to enhance the structural details of the generated maps. The experiments show that the proposed method can remove not only simple textual watermarks, but also colorful graphic watermarks with complex structures as well, and it does not need manual intervention and can achieve fast batch removal of a wide variety of watermarks for electronic maps. In addition, the detailed information contained in map tiles can be effectively preserved while acquiring high-quality watermark-free electronic map tiles. © 2022 IEEE.

关键词： Deep neural networks

LL-ICM: Image Compression for Low-level Machine Vision via Large Vision-Language Model

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xue, Yuan Zhang, Qi Jia, Chuanmin Wang, Shiqi School of Computer Science City University of Hong Kong Hong Kong National Engineering Research Center of Visual Technology Peking University Beijing China Wangxuan Institute of Computer Technology Peking University Beijing China

Image Compression for Machines (ICM) aims to compress images for machine vision tasks rather than human viewing. Current works predominantly concentrate on high-level tasks like object detection and semantic segmentation. However, the quality of original images is usually not guaranteed in the real world, leading to even worse perceptual quality or downstream task performance after compression. Low-level (LL) machine vision models, like image restoration models, can help improve such quality, and thereby their compression requirements should also be considered. In this paper, we propose a pioneered ICM framework for LL machine vision tasks, namely LL-ICM. By jointly optimizing compression and LL tasks, the proposed LL-ICM not only enriches its encoding ability in generalizing to versatile LL tasks but also optimizes the processing ability of down-stream LL task models, achieving mutual adaptation for image codecs and LL task models. Furthermore, we integrate large-scale vision-language models into the LL-ICM framework to generate more universal and distortion-robust feature embeddings for LL vision tasks. Therefore, one LL-ICM codec can generalize to multiple tasks. We establish a solid benchmark to evaluate LL-ICM, which includes extensive objective experiments by using both full and no-reference image quality assessments. Experimental results show that LL-ICM can achieve 22.65% BD-rate reductions over the state-of-the-art methods. © 2024, CC BY.

关键词： Semantic Segmentation