检索结果-内蒙古大学图书馆

A Survey on Federated Learning Poisoning Attacks and Defenses

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liang, Junchuan Wang, Rong Feng, Chaosheng Chang, Chin-Chen School of Computer Science Sichuan Normal University Chengdu China Visual Computing and Virtual Reality Key Laboratory of Sichuan Sichuan Normal University Chengdu China Department of Information Engineering and Computer Science Feng Chia University Taichung40724 Taiwan

As one kind of distributed machine learning technique, federated learning enables multiple clients to build a model across decentralized data collaboratively without explicitly aggregating the data. Due to its ability to break data silos, federated learning has received increasing attention in many fields, including finance, healthcare, and education. However, the invisibility of clients’ training data and the local training process result in some security issues. Recently, many works have been proposed to research the security attacks and defenses in federated learning, but there has been no special survey on poisoning attacks on federated learning and the corresponding defenses. In this paper, we investigate the most advanced schemes of federated learning poisoning attacks and defenses and point out the future directions in these areas. Copyright © 2023, The Authors. All rights reserved.

关键词： Learning systems

StegoFL: Using Steganography and Federated Learning to Transmit Malware

学校读者我要写书评

暂无评论

StegoFL: Using Steganography and Federated Learning to Trans...

IEEE International Conference on Trust, Security and Privacy in computing and Communications (TrustCom)

作者： Rong Wang Junchuan Liang Haiting Jiang Chaosheng Feng Chin-Chen Chang College of Computer Science Sichuan Normal University Chengdu China Visual Computing and Virtual Reality Key Laboratory of Sichuan Sichuan Normal University Chengdu China Graduate School of Science and Engineering Hosei University Tokyo Japan Department of Information Engineering and Computer Science Feng Chia University Taichung Taiwan

ISBN: (数字)9798331506209

ISBN: (纸本)9798331506216

Federated learning (FL) has gained increasing popularity in today’s privacy-focused world due to its ability to break data silos. Although federated learning can keep all the training data private at each site, it results in clients not being effectively monitored, which may pose security risks to the learning process. Previous studies mainly focus on backdoor and poisoning attacks against federated learning. In this paper, we propose a new stegomalware attack in FL, where an attacker, disguised as a benign client, hides malware into his local model and distributes the malware to other clients through the FL process. Conducting such an attack faces two challenges. First, embedding data into a local model may degrade its model performance, making it easier for a server in FL to detect this anomaly. Second, the local model with embedded malware is recalculated after model aggregation, which unavoidably alters the malware’s carrier and hinders the success of malware extraction. To address these challenges, we propose a method called StegoFL to incorporate steganography into federated learning to transmit malware. Specifically, we split the malware to be transmitted into segments and randomly select a few model parameters as carriers. Each malware segment is concealed within these carriers over several training rounds. We also propose a simple value-mapping method to extract the embedded data by comparing the aggregated carrier values with a threshold. Experimental results demonstrate that StegoFL can circumvent server-side detection mechanisms, i.e., accuracy tests and parametric distribution comparisons. In addition, it increases transmission capacity by at least 50 times compared to state-of-the-art covert communication methods in federated learning.

关键词： Training Steganography Federated learning Training data Malware Data models Security Servers Monitoring Faces

UNeLF: Unconstrained Neural Light Field for Self-Supervised Angular Super-Resolution

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： Zhao, Mingyuan Sheng, Hao Chen, Rongshan Cong, Ruixuan Cui, Zhenglong Yang, Da Beihang University State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beijing100191 China Beihang University Data Science and Intelligent Computing Laboratory Hangzhou International Innovation Institute Hangzhou311115 China Macao Polytechnic University Faculty of Applied Sciences 999078 China

Compared to supervised learning methods, self-supervised learning methods address the domain gap problem between light field (LF) datasets collected under varying acquisition conditions, which typically leads to decreased performance when differences exist in the distribution between the training and test sets. However, current self-supervised light field angular super-resolution (LFASR) techniques primarily focus on exploiting discrete spatial-angular features while neglecting continuous LF information. In contrast to previous work, we propose a self-supervised unconstrained neural light field (UNeLF) to continuously represent LF for LFASR. Specifically, any LF can be described as the camera pose for each sub-aperture image (SAI) and the two-plane that captures these SAIs. To describe the former, we introduce a SAIs-dependent pose optimization method to solve the issue that arises from the narrow baseline of most LF data, which hinders robust camera pose estimation. This mechanism reduces the number of trainable camera parameters from a quadratic to a constant scale, thereby alleviating the complexity of joint optimization. For the latter, we propose a novel adaptive two-plane parameterization strategy to determine the two-plane that captures these SAIs, facilitating refocusing. Finally, we jointly optimize the camera parameters, near-far planes and neural light field, efficiently mapping each adaptive two-plane parameterized ray to its correspondence color in a continuous manner. Comprehensive experiments demonstrate that UNeLF achieves faster training and inference with fewer computational resources while exhibiting superior performance on both synthetic and real-world datasets. © 1991-2012 IEEE.

关键词： Self-supervised learning

G-MAP: General Memory-Augmented Pre-trained Language Model for Domain Tasks

学校读者我要写书评

暂无评论

G-MAP: General Memory-Augmented Pre-trained Language Model f...

2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022

作者： Wan, Zhongwei Yin, Yichun Zhang, Wei Shi, Jiaxin Shang, Lifeng Chen, Guangyong Jiang, Xin Liu, Qun Guangdong Provincial Key Laboratory of Computer Vision and Virtual Reality Technology Shenzhen Institute of Advanced Technology Chinese Academy of Science China University of Chinese Academy of Sciences China Huawei Noah's Ark Lab Huawei Cloud Computing Zhejiang Lab China

Recently, domain-specific PLMs have been proposed to boost the task performance of specific domains (e.g., biomedical and computer science) by continuing to pre-train general PLMs with domain-specific corpora. However, this Domain-Adaptive Pre-Training (DAPT;Gururangan et al. (2020)) tends to forget the previous general knowledge acquired by general PLMs, which leads to a catastrophic forgetting phenomenon and sub-optimal performance. To alleviate this problem, we propose a new framework of General Memory-Augmented Pre-trained Language Model (G-MAP), which augments the domain-specific PLM by a memory representation built from the frozen general PLM without losing any general knowledge. Specifically, we propose a new memory-augmented layer, and based on it, different augmented strategies are explored to build the memory representation and then adaptively fuse it into the domain-specific PLM. We demonstrate the effectiveness of G-MAP on various domains (biomedical and computer science publications, news, and reviews) and different kinds (text classification, QA, NER) of tasks, and the extensive results show that the proposed G-MAP can achieve SOTA results on all tasks. © 2022 Association for Computational Linguistics.

关键词： Computational linguistics

Efficient Monaural Speech Enhancement using Spectrum Attention Fusion

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Long, Jinyu Gū, Jetic Bai, Binhao Yang, Zhibo Wei, Ping Li, Junli School of Computing Science Sichuan Normal University Chengdu610066 China School of Computing Science Simon Fraser University BurnabyBCV5A 1S6 Canada Visual Computing and Virtual Reality Key Laboratory of Sichuan Sichuan Normal University Chengdu610068 China Sichuan Key Laboratory of Translational Medicine of Traditional Chinese Medicine Chengdu610041 China

Speech enhancement is a demanding task in automated speech processing pipelines, focusing on separating clean speech from noisy channels. Transformer based models have recently bested RNN and CNN models in speech enhancement, however at the same time they are much more computationally expensive and require much more high quality training data, which is always hard to come by. In this paper, we present an improvement for speech enhancement models that maintains the expressiveness of self-attention while significantly reducing model complexity, which we have termed Spectrum Attention Fusion. We carefully construct a convolutional module to replace several self-attention layers in a speech Transformer, allowing the model to more efficiently fuse spectral features. Our proposed model is able to achieve comparable or better results against SOTA models but with significantly smaller parameters (0.58M) on the Voice Bank + DEMAND dataset. © 2023, CC BY.

关键词： Speech enhancement

An Optimization Method of Primer Design Based on Attention-BiLSTM

学校读者我要写书评

暂无评论

An Optimization Method of Primer Design Based on Attention-B...

Robotics, Artificial Intelligence and Intelligent Control (RAIIC), International Conference on

作者： Binhao Bai Jinyu Long Zhibo Yang Junli Li Ping Wei College of Computer Science Sichuan Normal University Chengdu China Visual Computing and Virtual Reality Key Laboratory of Sichuan Sichuan Normal University Chengdu China Sichuan Key Laboratory of Translational Medicine of Traditional Chinese Medicine Sichuan Academy of Traditional Chinese Medicine Sichuan Center of Translational Medicine Chengdu China

In this paper, we propose a method to predict the success of primer amplification based on the relationship existing between the sequence of primer and template, which can optimize the primer design and select the primer with better amplification from the candidate primer set. The double-stranded structure between primer and template nucleotide sequences is represented here by a number of words, each consisting of five characters that form sentences, as the dataset for the experiment, which is learned using an attention-based mechanism of bidirectional long short-term memory neural network model (Attention-BiLSTM), and then predicts primer amplification. The model predicted the results of polymerase chain reaction (PCR) involving specific primers and specific DNA templates with 82% accuracy, an improvement of about 2% over the performance of the LSTM with more stable value. These results show that the model can be used to effectively predict the results of PCR. This is the first paper to optimize primer design by screening the candidate primer set with a neural network model.

关键词：

S2WAT: Image Style Transfer via Hierarchical Vision Transformer Using Strips Window Attention

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Chiyu Xu, Xiaogang Wang, Lei Dai, Zaiyan Yang, Jun Sichuan Normal University China Nanjing University of Aeronautics and Astronautics China Zhejiang Lab China Zhejiang University China Visual Computing and Virtual Reality Key Laboratory of Sichuan Provience China

Transformer's recent integration into style transfer leverages its proficiency in establishing long-range dependencies, albeit at the expense of attenuated local modeling. This paper introduces Strips Window Attention Transformer (S2WAT), a novel hierarchical vision transformer designed for style transfer. S2WAT employs attention computation in diverse window shapes to capture both short- and long-range dependencies. The merged dependencies utilize the "Attn Merge" strategy, which adaptively determines spatial weights based on their relevance to the target. Extensive experiments on representative datasets show the proposed method's effectiveness compared to state-of-the-art (SOTA) transformer-based and other approaches. The code and pre-trained models are available at https://***/AlienZhang1996/S2WAT. Copyright © 2022, The Authors. All rights reserved.

关键词： Machine learning

An extended sequential quadratic method with extrapolation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Yongle Pong, Ting Kei Xu, Shiqi Department of Mathematics Visual Computing and Virtual Reality Key Laboratory of Sichuan Province Sichuan Normal University Chengdu China Department of Applied Mathematics the Hong Kong Polytechnic University Hong Kong Hong Kong Department of Mathematics Sichuan Normal University Chengdu China

We revisit and adapt the extended sequential quadratic method (ESQM) in [3] for solving a class of difference-of-convex optimization problems whose constraints are defined as the intersection of level sets of Lipschitz differentiable functions and a simple compact convex set. Particularly, for this class of problems, we develop a variant of ESQM, called ESQM with extrapolation (ESQMe), which incorporates Nesterov's extrapolation techniques for empirical acceleration. Under standard constraint qualifications, we show that the sequence generated by ESQMe clusters at a critical point if the extrapolation parameters are uniformly bounded above by a certain threshold. Convergence of the whole sequence and the convergence rate are established by assuming Kurdyka- Lojasiewicz (KL) property of a suitable potential function and imposing additional differentiability assumptions on the objective and constraint functions. In addition, when the objective and constraint functions are all convex, we show that linear convergence can be established if a certain exact penalty function is known to be a KL function with exponent 1/2;we also discuss how the KL exponent of such an exact penalty function can be deduced from that of the original extended objective (i.e., sum of the objective and the indicator function of the constraint set). Finally, we perform numerical experiments to demonstrate the empirical acceleration of ESQMe over a basic version of ESQM, and illustrate its effectiveness by comparing with the natural competing algorithm SCPls from [35]. Copyright © 2023, The Authors. All rights reserved.

关键词： Extrapolation

Unraveling the Neural Dynamics of Mathematical Interference in English Reading:A Novel Approach with Deep Learning and Fnirs Data

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Liang, Zhijie Wang, Ling Sun, Bo Wang, Daifa Yang, Juan School of Computer Science Sichuan Normal University Chengdu610066 China Visual Computing and Virtual Reality Key Laboratory of Sichuan Province Chengdu610066 China College of Foreign Languages Sichuan Normal University Chengdu610066 China School of Biological Science and Medical Engineering Beihang University Beijing100083 China

English has emerged as the predominant global language, making it a central focus in educational field seeking to integrate English with other disciplines for improved learning efficiency. Despite numerous studies demonstrating a connection between English learning performance and mathematical logic skills, current explanations and theoretical frameworks predominantly rely on behavioral analysis. This highlights the need for neuroscientific evidence to elucidate the neural mechanisms supporting this correlation, especially in terms of internal neural processes. To bridge this gap, the present study employs functional near-infrared spectroscopy (fNIRS) to construct a novel dataset on mathematical interference in English acquisition. Utilizing this dataset, a novel deep learning model named AC-LSTM is proposed, amalgamating Transformer and LSTM architectures to identify residual mathematical cognition during the English learning process. The AC-LSTM model achieves an exceptional accuracy rate of 99.8%, surpassing other machine learning and deep learning models. Moreover, a multi-class classification experiment is conducted to discern algebra, geometry, and quantitative reasoning interference, with the AC-LSTM model achieving the highest accuracy of 75.9% in this classification task. Furthermore, crucial brain channels for interference detection are pinpointed through grid search, and alterations in vital brain regions (R-Broca and L-Broca) are unveiled via association rule analysis. By integrating fNIRS, deep learning, and data mining techniques, this study delves into cognitive interference in English learning, providing valuable insights for educational neuroscience and data mining research. © 2024, The Authors. All rights reserved.

关键词： Data mining