检索结果-内蒙古大学图书馆

Investigating the Effectiveness of Data Augmentation from Similarity and Diversity: An Empirical Study

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Yang, Suorong Guo, Suhan Zhao, Jian Shen, Furao State Key Laboratory for Novel Software Technology Nanjing University China Department of Computer Science and Technology Nanjing University China School of Artificial Intelligence Nanjing University China School of Electronic Science and Engineering Nanjing University China

Data augmentation has emerged as a widely adopted technique for improving the generalization capabilities of deep neural networks. However, evaluating the effectiveness of data augmentation methods solely based on model training is computationally demanding and lacks interpretability. Moreover, the absence of quantitative standards hinders our understanding of the underlying mechanisms of data augmentation approaches and the development of novel techniques. To this end, we propose interpretable quantitative measures that decompose the effectiveness of data augmentation methods into two key dimensions: similarity and diversity. The proposed similarity measure describes the overall similarity between the original and augmented datasets, while the diversity measure quantifies the divergence in inherent complexity between the original and augmented datasets in terms of categories. Importantly, our proposed measures are model training-agnostic, ensuring efficiency in their calculation. Through experiments on several benchmark datasets, including MNIST, CIFAR-10, CIFAR-100, and ImageNet, we demonstrate the efficacy of our measures in evaluating the effectiveness of various data augmentation methods. Furthermore, although the proposed measures are straightforward, they have the potential to guide the design and parameter tuning of data augmentation techniques and enable the validation of data augmentation methods' efficacy before embarking on large-scale model training. © 2023, The Authors. All rights reserved.

关键词： Image classification

RADAP: A Robust and Adaptive Defense Against Diverse Adversarial Patches on Face Recognition

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liu, Xiaoliang Shen, Furao Zhao, Jian Nie, Changhai State Key Laboratory for Novel Software Technology Nanjing University China Department of Computer Science and Technology Nanjing University China School of Artificial Intelligence Nanjing University China School of Electronic Science and Engineering Nanjing University China

Face recognition (FR) systems powered by deep learning have become widely used in various applications. However, they are vulnerable to adversarial attacks, especially those based on local adversarial patches that can be physically applied to real-world objects. In this paper, we propose RADAP, a robust and adaptive defense mechanism against diverse adversarial patches in both closed-set and open-set FR systems. RADAP employs innovative techniques, such as FCutout and F-patch, which use Fourier space sampling masks to improve the occlusion robustness of the FR model and the performance of the patch segmenter. Moreover, we introduce an edge-aware binary cross-entropy (EBCE) loss function to enhance the accuracy of patch detection. We also present the split and fill (SAF) strategy, which is designed to counter the vulnerability of the patch segmenter to complete white-box adaptive attacks. We conduct comprehensive experiments to validate the effectiveness of RADAP, which shows significant improvements in defense performance against various adversarial patches, while maintaining clean accuracy higher than that of the undefended Vanilla model. Copyright © 2023, The Authors. All rights reserved.

关键词： Face recognition

Advmask: A Sparse Adversarial Attack-Based Data Augmentation Method for Image Classification

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Yang, Suorong Li, Jinqiao Zhang, Tianyue Zhao, Jian Shen, Furao State Key Laboratory for Novel Software Technology Nanjing University China Department of Computer Science and Technology Nanjing University China School of Artificial Intelligence Nanjing University China School of Electronic Science and Engineering Nanjing University China

Data augmentation has been an essential technique for improving the generalization ability of deep neural networks in image classification ***, intensive changes in appearance and different degrees of occlusion in images are the key factors that severely affect the generalization ability of image classification models. Therefore, in order to enhance the generalization performance and robustness of deep models, data augmentation approaches by providing models with more diverse training data in various scenarios are widely applied. Although many existing data augmentation methods simulate occlusion in the augmented images to enhance the generalization of models, these methods randomly delete some areas in images without considering the semantic information of images. In this work, we propose a novel data augmentation method named AdvMask for image classification based on sparse adversarial attack techniques. AdvMask first identifies the key points that have the greatest influence on the classification results via a proposed end-to-end sparse adversarial attack module. During the data augmentation process, AdvMask efficiently generates diverse augmented data with structured occlusions based on the key points. By doing so, AdvMask can force deep models to seek other relevant content while the most discriminative content is hidden. Extensive experimental results on various benchmark datasets and deep models demonstrate that our proposed method can effectively improve the generalization performance of deep models and significantly outperforms previous data augmentation methods. © 2023, The Authors. All rights reserved.

关键词： Image classification

ON THE FINE-GRAINED QUERY COMPLEXITY OF SYMMETRIC FUNCTIONS

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Podder, Supartha Yao, Penghui Ye, Zekun Department of Computer Science Stony Brook University New York United States State Key Laboratory for Novel Software Technology Nanjing University Nanjing China Hefei National Laboratory Hefei China

Watrous conjectured that the randomized and quantum query complexities of symmetric functions are polynomially equivalent, which was resolved by Ambainis and Aaronson [1], and was later improved in [17, 13]. This paper explores a fine-grained version of the Watrous conjecture, including the randomized and quantum algorithms with success probabilities arbitrarily close to 1/2. Our contributions include the following: 1. An analysis of the optimal success probability of quantum and randomized query algorithms of two fundamental partial symmetric Boolean functions given a fixed number of queries. We prove that for any quantum algorithm computing these two functions using T queries, there exist randomized algorithms using poly(T) queries that achieve the same success probability as the quantum algorithm, even if the success probability is arbitrarily close to 1/2. These two classes of functions are instrumental in analyzing general symmetric functions. 2. We establish that for any total symmetric Boolean function f, if a quantum algorithm uses T queries to compute f with success probability 1/2 + β, then there exists a randomized algorithm using O(T2) queries to compute f with success probability 1/2 + Ω (δβ2) on a 1 − δ fraction of inputs, where β, δ can be arbitrarily small positive values. As a corollary, we prove a randomized version of Aaronson-Ambainis Conjecture [1] for total symmetric Boolean functions in the regime where the success probability of algorithms can be arbitrarily close to 1/2.3. We present tight polynomial equivalences for several fundamental complexity measures of partial symmetric Boolean functions. Specifically, we first prove that for certain partial symmetric Boolean functions, quantum query complexity is at most quadratic in approximate degree for any error arbitrarily close to 1/2. Next, we show exact quantum query complexity is at most quadratic in degree. Additionally, we give the tight bounds of several complexity measures, indicating th

关键词： Quantum theory

Dts-Adapstnet: An Adaptive Spatiotemporal Neural Networks for Traffic Prediction with Mulit-Graph Fusion

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Zhang, Jing Shi, Wenlong Ye, Xiucai Tsai, Peiwei School of Computer Science and Mathematics Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fuzhou China Department of Computer Science University of Tsukuba Tsukuba Japan Department of Computer Science and Software Engineering Swinburne University of Technology Hawthorn Australia

Traffic prediction is of vital importance in intelligent transportation systems. It can realize efficient route planning, avoid traffic congestion and reduce travel time, etc. However, it is difficult to make accurate road traffic prediction, due to the complex spatiotemporal dependencies of the traffic network. The construction and learning of spatial dependencies play a pivotal role in road traffic prediction, laying the foundation for accurate traffic prediction. Regrettably, most of the existing methods for capturing spatial dependencies only consider a single spatial relationship, ignoring other potential temporal and spatial correlations in the traffic network. Moreover, the end-to-end training methods are difficult to control the training direction during graph learning. Additionally, the result of traffic forecasting lacks of fusing multiple traffic data information. All of these have adverse effects on road traffic prediction. In other to capture the spatiotemporal dependencies of the traffic network accurately, a novel traffic prediction framework, Adaptive Spatio-Temporal Graph Neural Network based on Multi-graph Fusion (DTS-adapSTNet), is proposed. First of all, in order to better extract the hidden spatial dependencies, a method of fusing multiple factors is designed, which includs the distance relationship, transfer relationship and same-road segment relationship of traffic data, etc. Secondly, an adaptive learning method is proposed, which can control the learning direction of parameters better by the adaptive matrix generation module and traffic prediction module. Thirdly, an improved loss function is designed for training processes and a multi-matrix fusion module is designed to perform weighted fusion of the learned matrices, updating the spatial adjacency matrix continuously, which fuses as much traffic information as possible for more accurate traffic prediction. Finally, based on the experiments by two large real-world datasets, the DTS-adapSTNe

关键词： Travel time

Hardware Computation Graph for DNN Accelerator Design Automation without Inter-PU Templates

学校读者我要写书评

暂无评论

Hardware Computation Graph for DNN Accelerator Design Automa...

IEEE International Conference on computer-Aided Design

作者： Jun Li Wei Wang Wu-Jun Li Department of Computer Science and Technology National Key Laboratory for Novel Software Technology Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing University China

Existing deep neural network (DNN) accelerator design automation (ADA) methods adopt architecture templates to predetermine parts of design choices and then explore the left design choices beyond templates. These templates can be classified into intra-PU templates and inter-PU templates according to the architecture hierarchy. Since templates limit the flexibility of ADA, designing effective ADA methods without templates has become an important research topic. Although there have appeared some works to enhance the flexibility of ADA by removing intra-PU templates, to the best of our knowledge no existing works have studied ADA methods without inter-PU templates. ADA with predetermined inter-PU templates is typically inefficient in terms of resource utilization, especially for DNNs with complex topology. In this paper, we propose a novel method, called hardware computation graph (HCG), for ADA without inter-PU templates. Experiments show that HCG method can achieve competitive latency while using only 1.4x ~ 5x fewer on-chip memory, compared with existing state-of-the-art ADA methods.

关键词： Deep learning Design automation Neural networks computer architecture Hardware Topology System-on-chip

A Survey on Large Language Models for software Engineering

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhang, Quanjun Fang, Chunrong Xie, Yang Zhang, Yaxin Yang, Yun Sun, Weisong Yu, Shengcheng Chen, Zhenyu The State Key Laboratory for Novel Software Technology Nanjing University China The School of Computer Science and Engineering Nanyang Technological University Singapore The Department of Computing Technologies Swinburne University of Technology MelbourneVIC3122 Australia

software Engineering (SE) is the systematic design, development, maintenance, and management of software applications underpinning the digital infrastructure of our modern world. Very recently, the SE community has seen a rapidly increasing number of techniques employing Large Language Models (LLMs) to automate a broad range of SE tasks. Nevertheless, existing information of the applications, effects, and possible limitations of LLMs within SE is still not well-studied. In this paper, we provide a systematic survey to summarize the current state-of-the-art research in the LLM-based SE community. We summarize 62 representative LLMs of Code across three model architectures, 15 pre-training objectives across four categories, and 16 downstream tasks across five categories. We then present a detailed summarization of the recent SE studies for which LLMs are commonly utilized, including 947 studies for 112 specific code-related tasks across five crucial phases within the SE workflow. We also discuss several critical aspects during the integration of LLMs into SE, such as empirical evaluation, benchmarking, security and reliability, domain tuning, compressing and distillation. Finally, we highlight several challenges and potential opportunities on applying LLMs for future SE studies, such as exploring domain LLMs and constructing clean evaluation datasets. Overall, our work can help researchers gain a comprehensive understanding about the achievements of the existing LLM-based SE studies and promote the practical application of these techniques. Our artifacts are publicly available and will be continuously updated at the living repository: https://***/iSEngLab/AwesomeLLM4SE. © 2023, CC BY.

关键词： Application programs

On the Effectiveness of Distillation in Mitigating Backdoors in Pre-trained Encoder

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Han, Tingxu Huang, Shenghan Ding, Ziqi Sun, Weisong Feng, Yebo Fang, Chunrong Li, Jun Qian, Hanwei Wu, Cong Zhang, Quanjun Liu, Yang Chen, Zhenyu The State Key Laboratory for Novel Software Technology Nanjing University Jiangsu210093 China The School of Computer Science University of New South Wales NSW2052 Australia The School of Computer Science and Engineering Nanyang Technological University Nanyang639798 Singapore The Department of Computer Science The Director of the Network & Security Research Laboratory The University of Oregon United States

With the development of deep learning, self-supervised learning (SSL) has become particularly popular. SSL pre-trains an encoder (e.g., an image encoder) to learn generic features from a large amount of unlabeled data, releases it on a third-party platform, and allows any downstream task (e.g., an image classification task) to further fine-tune the encoder with task-specific data to derive the model for the task. However, despite the tremendous performance improvements it brings, SSL faces serious security risks. An adversary can poison pre-training datasets used to pre-train an encoder or publish a poisoned encoder directly to third-party platforms, which will also poison the model for any downstream task that uses these encoders. In this paper, we study a defense against poisoned encoders in SSL called distillation, which is a defense used in supervised learning originally. Distillation aims to distill knowledge from a given model (a.k.a the teacher net) and transfer it to another (a.k.a the student net). Now, we use it to distill benign knowledge from poisoned pre-trained encoders and transfer it to a new encoder, resulting in a clean pre-trained encoder. In particular, we conduct an empirical study on the effectiveness and performance of distillation against poisoned encoders. Using two state-of-the-art backdoor attacks against pre-trained image encoders and four commonly used image classification datasets, our experimental results show that distillation can reduce attack success rate from 80.87% to 27.51% while suffering a 6.35% loss in accuracy. Moreover, we investigate the impact of three core components of distillation on performance: Teacher net, student net, and distillation loss. By comparing 4 different teacher nets, 3 student nets, and 6 distillation losses, we find that fine-tuned teacher nets, warm-up-training-based student nets, and attention-based distillation loss perform best, respectively. Copyright © 2024, The Authors. All rights reserved.

关键词： Distillation

Semantic Communications with computer Vision Sensing for Edge Video Transmission

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Peng, Yubo Xiang, Luping Yang, Kun Wang, Kezhi Debbah, Mérouane State Key Laboratory of Novel Software Technology Nanjing University Nanjing China Suzhou China Department of Computer Science Brunel University London United Kingdom Department of Electrical Engineering and Computer Science The KU 6G Center Khalifa University Abu Dhabi127788 United Arab Emirates

Despite the widespread adoption of vision sensors in edge applications, such as surveillance, the transmission of video data consumes substantial spectrum resources. Semantic communication (SC) offers a solution by extracting and compressing information at the semantic level, preserving the accuracy and relevance of transmitted data while significantly reducing the volume of transmitted information. However, traditional SC methods face inefficiencies due to the repeated transmission of static frames in edge videos, exacerbated by the absence of sensing capabilities, which results in spectrum inefficiency. To address this challenge, we propose a SC with computer vision sensing (SCCVS) framework for edge video transmission. The framework first introduces a compression ratio (CR) adaptive SC (CRSC) model, capable of adjusting CR based on whether the frames are static or dynamic, effectively conserving spectrum resources. Additionally, we implement an object detection and semantic segmentation models-enabled sensing (OSMS) scheme, which intelligently senses the changes in the scene and assesses the significance of each frame through in-context analysis. Hence, The OSMS scheme provides CR prompts to the CRSC model based on real-time sensing results. Moreover, both CRSC and OSMS are designed as lightweight models, ensuring compatibility with resource-constrained sensors commonly used in practical edge applications. Experimental simulations validate the effectiveness of the proposed SCCVS framework, demonstrating its ability to enhance transmission efficiency without sacrificing critical semantic information. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantic Segmentation