检索结果-内蒙古大学图书馆

Nerftap: Enhancing Transferability of Adversarial Patches on Face Recognition Using Neural Radiance Fields

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Liu, Xiaoliang Shen, Fu-Rao Han, Feng 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) technology plays a crucial role in various applications, but its vulnerability to adversarial attacks poses significant security concerns. Existing research primarily focuses on transferability to different FR models, overlooking the direct transferability to victim's face images, which is a practical threat in real-world scenarios. In this study, we propose a novel adversarial attack method that considers both the transferability to the FR model and the victim's face image, called NerfTAP. Leveraging Neural Radiance Fields (NeRF)-based 3D Generative Adversarial Network (GAN), we generate new view face images for the source and target subjects to enhance transferability of adversarial patches. We introduce a style consistency loss to ensure the visual similarity between the adversarial UV map and the target UV map under a 0-1 mask, enhancing the effectiveness and naturalness of the generated adversarial face images. Extensive experiments and evaluations on various FR models demonstrate the superiority of our approach over existing attack techniques. Our work provides valuable insights for enhancing the robustness of FR systems in practical adversarial settings. © 2024, The Authors. All rights reserved.

关键词： Face recognition

FuseFPS: Accelerating Farthest Point Sampling with Fusing KD-tree Construction for Point Clouds

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Han, Meng Wang, Liang Xiao, Limin Zhang, Hao Zhang, Chenhao Xie, Xilong Zheng, Shuai Dong, Jin State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beihang University Beijing100191 China Beijing Academy of Blockchain and Edge Computing China

Point cloud analytics has become a critical workload for embedded and mobile platforms across various applications. Farthest point sampling (FPS) is a fundamental and widely used kernel in point cloud processing. However, the heavy external memory access makes FPS a performance bottleneck for real-time point cloud processing. Although bucket-based farthest point sampling can significantly reduce unnecessary memory accesses during the point sampling stage, the KD-tree construction stage becomes the predominant contributor to execution time. In this paper, we present FuseFPS, an architecture and algorithm co-design for bucket-based farthest point sampling. We first propose a hardware-friendly sampling-driven KD-tree construction algorithm. The algorithm fuses the KD-tree construction stage into the point sampling stage, further reducing memory accesses. Then, we design an efficient accelerator for bucket-based point sampling. The accelerator can offload the entire bucket-based FPS kernel at a low hardware cost. Finally, we evaluate our approach on various point cloud datasets. The detailed experiments show that compared to the state-of-the-art accelerator QuickFPS, FuseFPS achieves about 4.3× and about 6.1× improvements on speed and power efficiency, respectively. Copyright © 2023, The Authors. All rights reserved.

关键词： Trees (mathematics)

EntAugment: Entropy-Driven Adaptive Data Augmentation Framework for Image Classification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yang, Suorong Shen, Furao Zhao, Jian 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 (DA) has been widely used to improve the generalization of deep neural networks. While existing DA methods have proven effective, they often rely on augmentation operations with random magnitudes to each sample. However, this approach can inadvertently introduce noise, induce distribution shifts, and increase the risk of overfitting. In this paper, we propose EntAugment, a tuning-free and adaptive DA framework. Unlike previous work, EntAugment dynamically assesses and adjusts the augmentation magnitudes for each sample during training, leveraging insights into both the inherent complexities of training samples and the evolving status of deep models. Specifically, in EntAugment, the magnitudes are determined by the information entropy derived from the probability distribution obtained by applying the softmax function to the model’s output. In addition, to further enhance the efficacy of EntAugment, we introduce a novel entropy regularization term, EntLoss, which complements the EntAugment approach. Theoretical analysis further demonstrates that EntLoss, compared to traditional cross-entropy loss, achieves closer alignment between the model distributions and underlying dataset distributions. Moreover, EntAugment and EntLoss can be utilized separately or jointly. We conduct extensive experiments across multiple image classification tasks and network architectures with thorough comparisons of existing DA methods. Importantly, the proposed methods outperform others without introducing any auxiliary models or noticeable extra computational costs, highlighting both effectiveness and efficiency. Code is available at https://***/Jackbrocp/EntAugment. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep neural networks

UncTrack: Reliable Visual Object Tracking with Uncertainty-Aware Prototype Memory Network

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yao, Siyuan Guo, Yang Yan, Yanyang Ren, Wenqi Cao, Xiaochun School of Computer Science National Pilot Software Engineering School Beijing University of Posts and Telecommunications Beijing100876 China State Key Lab of Processors Institute of Computing Technology Chinese Academy of Sciences Beijing100190 China School of Cyber Science and Technology Sun Yat-sen University Shenzhen Campus Shenzhen518107 China

Transformer-based trackers have achieved promising success and become the dominant tracking paradigm due to their accuracy and efficiency. Despite the substantial progress, most of the existing approaches tackle object tracking as a deterministic coordinate regression problem, while the target localization uncertainty has been greatly overlooked, which hampers trackers’ ability to maintain reliable target state prediction in challenging scenarios. To address this issue, we propose UncTrack, a novel uncertainty-aware transformer tracker that predicts the target localization uncertainty and incorporates this uncertainty information for accurate target state inference. Specifically, UncTrack utilizes a transformer encoder to perform feature interaction between template and search images. The output features are passed into an uncertainty-aware localization decoder (ULD) to coarsely predict the corner-based localization and the corresponding localization uncertainty. Then the localization uncertainty is sent into a prototype memory network (PMN) to excavate valuable historical information to identify whether the target state prediction is reliable or not. To enhance the template representation, the samples with high confidence are fed back into the prototype memory bank for memory updating, making the tracker more robust to challenging appearance variations. Extensive experiments demonstrate that our method outperforms other state-of-the-art methods. Our code is available at https://***/ManOfStory/UncTrack. Copyright © 2025, The Authors. All rights reserved.

关键词： Object tracking

Bridging Distinct Spaces in Graph-Based Machine Learning 1

学校读者我要写书评

暂无评论

7th Asian Conference on Pattern Recognition, ACPR 2023

作者： Jia, Linlin Ning, Xiao Gaüzère, Benoit Honeine, Paul Riesen, Kaspar Institute of Computer Science University of Bern Bern3012 Switzerland State Key Laboratory of Bioelectronics School of Biological Science and Medical Engineering Southeast University 2 Sipailou Nanjing210096 China The LITIS Lab INSA Rouen Normandie Rouen France The LITIS Lab Université de Rouen Normandie Rouen France

ISBN: (数字)9783031476372

ISBN: (纸本)9783031476365

Graph-based machine learning, encompassing Graph Edit Distances (GEDs), Graph Kernels, and Graph Neural Networks (GNNs), offers extensive capabilities and exciting potential. While each model possesses unique strengths for graph challenges, interrelations between their underlying spaces remain under-explored. In this paper, we introduce a novel framework for bridging these distinct spaces via GED cost learning. A supervised metric learning approach serves as an instance of this framework, enabling space alignment through pairwise distances and the optimization of edit costs. Experiments reveal the framework’s potential for enhancing varied tasks, including regression, classification, and graph generation, heralding new possibilities in these fields. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.

关键词： Machine learning

TH-SLP: Web Service Link Prediction Based on Topic-aware Heterogeneous Graph Neural Network

学校读者我要写书评

暂无评论

TH-SLP: Web Service Link Prediction Based on Topic-aware Het...

IEEE International Conference on Web Services (ICWS)

作者： Qian Peng Buqing Cao Xiang Xie Shanpeng Liu Guosheng Kang Jianxun Liu School of Computer Science and Engineering & Hunan Provincial Key Lab. For Services Computing and Novel Software Technology Hunan University of Science and Technology Xiangtan Hunan China

With the emergence of more and more Web services, finding suitable services becomes a difficult problem. Service link prediction is employed to disclose relationships among services, which facilitates the further development of service composition, selection, and recommendation. But the existing link prediction approaches simply utilize the structural features of the service network. In reality, the rich text content in service node description documents also carries latent but fine-grained semantics generated by multifaceted topic-aware factors, yet few efforts are committed to mining them. In this paper, we propose a Web service link prediction method based on a topic-aware heterogeneous graph neural network. Specifically, the method consists of two main layers, including the meta-path intra-decomposition and the meta-path inter-mergence. Meta-path intra-decomposition aims to mine the topic distribution of the meta-paths-based context while capturing fine-grained topic-aware semantics. Meta-path inter-mergence uniquely aggregates topic-aware factors according to the mined distribution and adopts a multifaceted attention mechanism to aggregate different meta-paths, enabling service nodes to generate multifaceted topic-aware embeddings that preserve not only the structure and but also the topic-aware semantics. In addition, a topic prior guidance regularization item is set up for quality assurance of multifaceted topic-aware embedding that depends on global knowledge of the unstructured text content in description documents. Experimental results on real datasets show that our proposed model outperforms other existing baselines methods in the link prediction task, successfully validating the effectiveness of our proposed method.

关键词：

Enhancing ASIC Technology Mapping via Parallel Supergate Computing

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Cai, Ye Yang, Zonglin Ni, Liwei Xie, Biwei Li, Xingquan College of Computer Science and Software Engineering Shenzhen University Shenzhen China Peng Cheng Laboratory Shenzhen China State Key Lab of Processors Institute of Computing Technology Chinese Academy of Sciences Beijing China

With the development of large-scale integrated circuits, electronic design automation (EDA) tools are increasingly emphasizing efficiency, with parallel algorithms becoming a trend. The optimization of delay reduction is a crucial factor for ASIC technology mapping, and supergate technology proves to be an effective method for achieving this in EDA tools flow. However, we have observed that increasing the number of generated supergates can reduce delay, but this comes at the cost of an exponential increase in computation time. In this paper, we propose a parallel supergate computing method that addresses the tradeoff between time-consuming and delay optimization. The proposed method utilizes the input-constrained supergate pattern to parallelly generate the supergate candidates, and then filter the valid supergates as the results. Experiment results show the efficiency of the proposed method, for example, it can attain the improvement of 4× speedup in computation time and 10.1 in delay reduction with 32 threads. Copyright © 2024, The Authors. All rights reserved.

关键词： Efficiency

Enhance Learning Efficiency of Oblique Decision Tree via Feature Concatenation

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Lyu, Shen-Huan He, Yi-Xiao Wang, Yanyan Qu, Zhihao Tang, Bin Ye, Baoliu Key Laboratory of Water Big Data Technology of Ministry of Water Resources College of Computer Science and Software Engineering Hohai University Nanjing China State Key Laboratory for Novel Software Technology Nanjing University Nanjing China School of Artificial Intelligence and Information Technology Nanjing University of Chinese Medicine Nanjing China

Oblique Decision Tree (ODT) separates the feature space by linear projections, as opposed to the conventional Decision Tree (DT) that forces axis-parallel splits. ODT has been proven to have a stronger representation ability than DT, as it provides a way to create shallower tree structures while still approximating complex decision boundaries. However, its learning efficiency is still insufficient, since the linear projections cannot be transmitted to the child nodes, resulting in a waste of model parameters. In this work, we propose an enhanced ODT method with Feature Concatenation (FC-ODT), which enables in-model feature transformation to transmit the projections along the decision paths. Theoretically, we prove that our method enjoys a faster consistency rate w.r.t. the tree depth, indicating that our method possesses a significant advantage in generalization performance, especially for shallow trees. Experiments show that FC-ODT can outperform the other state-of-the-art decision trees with a limited tree depth. Copyright © 2025, The Authors. All rights reserved.

关键词： Efficiency

Enhancing ASIC Technology Mapping via Parallel Supergate Computing

学校读者我要写书评

暂无评论

Enhancing ASIC Technology Mapping via Parallel Supergate Com...

Electronics Design Automation (ISEDA), International Symposium of

作者： Ye Cai Zonglin Yang Liwei Ni Biwei Xie Xingquan Li College of Computer Science and Software Engineering Shenzhen University Shenzhen China Peng Cheng Laboratory Shenzhen China State Key Lab of Processors Chinese Academy of Sciences Institute of Computing Technology Beijing China

ISBN: (数字)9798350352030

ISBN: (纸本)9798350352047

With the development of large-scale integrated circuits, electronic design automation (EDA) tools are increasingly emphasizing efficiency, with parallel algorithms becoming a trend. The optimization of delay reduction is a crucial factor for ASIC technology mapping, and supergate technology proves to be an effective method for achieving this in EDA tools flow. However, we have observed that increasing the number of generated supergates can reduce delay, but this comes at the cost of an exponential increase in computation time. In this paper, we propose a parallel supergate computing method that addresses the tradeoff between time-consuming and delay optimization. The proposed method utilizes the input-constrained supergate pattern to parallelly generate the supergate candidates, and then filter the valid supergates as the results. Experiment results show the efficiency of the proposed method, for example, it can attain the improvement of 4x speedup in computation time and 10.1 in delay reduction with 32 threads.

关键词： Integrated circuits Design automation Costs Instruction sets Aerospace electronics Market research Delays