检索结果-内蒙古大学图书馆

Rediscovering BCE Loss for Uniform Classification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Li, Qiufu Jia, Xi Zhou, Jiancan Shen, Linlin Duan, Jinming National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Computer Vision Institute Shenzhen University SZU Branch Shenzhen Institute of Artificial Intelligence and Robotics for Society China School of Computer Science University of Birmingham United Kingdom Aqara Lumi United Technology Co. Ltd China

This paper introduces the concept of uniform classification, which employs a unified threshold to classify all samples rather than adaptive threshold classifying each individual sample. We also propose the uniform classification accuracy as a metric to measure the model’s performance in uniform classification. Furthermore, begin with a naive loss, we mathematically derive a loss function suitable for the uniform classification, which is the BCE function integrated with a unified bias. We demonstrate the unified threshold could be learned via the bias. The extensive experiments on six classification datasets and three feature extraction models show that, compared to the SoftMax loss, the models trained with the BCE loss not only exhibit higher uniform classification accuracy but also higher sample-wise classification accuracy. In addition, the learned bias from BCE loss is very close to the unified threshold used in the uniform classification. The features extracted by the models trained with BCE loss not only possess uniformity but also demonstrate better intra-class compactness and inter-class distinctiveness, yielding superior performance on open-set tasks such as face recognition. © 2024, CC BY.

关键词： Classification (of information)

A Large-Scale Empirical Study on Semantic Versioning in Golang Ecosystem

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Wenke Wu, Feng Fu, Cai Zhou, Fan School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan China Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security China Co.Ltd Shenzhen China

Third-party libraries (TPLs) have become an essential component of software, accelerating development and reducing maintenance costs. However, breaking changes often occur during the upgrades of TPLs and prevent client programs from moving forward. Semantic versioning (SemVer) has been applied to standardize the versions of releases according to compatibility, but not all releases follow SemVer compliance. Lots of work focuses on SemVer compliance in ecosystems such as Java and JavaScript beyond Golang (Go for short). Due to the lack of tools to detect breaking changes and dataset for Go, developers of TPLs do not know if breaking changes occur and affect client programs, and developers of client programs may hesitate to upgrade dependencies in terms of breaking changes. To bridge this gap, we conduct the first large-scale empirical study in the Go ecosystem to study SemVer compliance in terms of breaking changes and their impact. In detail, we propose GoSVI (Go Semantic Versioning Insight) to detect breaking changes and analyze their impact by resolving identifiers in client programs and comparing their types with breaking changes. Moreover, we collect the first large-scale Go dataset with a dependency graph from GitHub, including 124K TPLs and 532K client programs. Based on the dataset, our results show that 86.3% of library upgrades follow SemVer compliance and 28.6% of no-major upgrades introduce breaking changes. Furthermore, the tendency to comply with SemVer has improved over time from 63.7% in 2018/09 to 92.2% in 2023/03. Finally, we find 33.3% of downstream client programs may be affected by breaking changes. These findings provide developers and users of TPLs with valuable insights to help make decisions related to SemVer. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

DINOv2-powered Few-Shot Semantic Segmentation: A Unified Framework via Cross-Model Distillation and 4D Correlation Mining

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Zhuo, Wei Tang, Zhiyue Xue, Wufeng Ding, Hao Shen, Linlin School of Artificial Intelligence Shenzhen University Shenzhen518060 China National Engineering Laboratory of Big Data System Computing Technology Shenzhen University China School of Biomedical Engineering Shenzhen University Medical School Shenzhen University China School of Computer Science and Software Engineering Shenzhen University China

Few-shot semantic segmentation has gained increasing interest due to its generalization capability, i.e., segmenting pixels of novel classes requiring only a few annotated images. Prior work has focused on meta-learning for support-query matching, with extensive development in both prototype-based and aggregation-based methods. To address data scarcity, recent approaches have turned to foundation models to enhance representation transferability for novel class segmentation. Among them, a hybrid dual-modal framework including both DINOv2 and SAM has garnered attention due to their complementary capabilities. We wonder "can we build a unified model with knowledge from both foundation models?" To this end, we propose FS-DINO, with only DINOv2’s encoder and a lightweight segmenter. The segmenter features a bottleneck adapter, a meta-visual prompt generator based on dense similarities and semantic embeddings, and a decoder. Through coarse-to-fine cross-model distillation, we effectively integrate SAM’s knowledge into our lightweight segmenter, which can be further enhanced by 4D correlation mining on support-query pairs. Extensive experiments on COCO-20i, PASCAL-5i, and FSS-1000 demonstrate the effectiveness and superiority of our method. Copyright © 2025, The Authors. All rights reserved.

关键词： Semantic Segmentation

APSeg: Auto-Prompt Network for Cross-Domain Few-Shot Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： He, Weizhao Zhang, Yang Zhuo, Wei Shen, Linlin Yang, Jiaqi Deng, Songhe Sun, Liang Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University China Shenzhen Institute of Artificial Intelligence and Robotics for Society China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University China School of Computer Science University of Nottingham China

Few-shot semantic segmentation (FSS) endeavors to segment unseen classes with only a few labeled samples. Current FSS methods are commonly built on the assumption that their training and application scenarios share similar domains, and their performances degrade significantly while applied to a distinct domain. To this end, we propose to leverage the cutting-edge foundation model, the Segment Anything Model (SAM), for generalization enhancement. The SAM however performs unsatisfactorily on domains that are distinct from its training data, which primarily comprise natural scene images, and it does not support automatic segmentation of specific semantics due to its interactive prompting mechanism. In our work, we introduce APSeg, a novel auto-prompt network for cross-domain few-shot semantic segmentation (CD-FSS), which is designed to be auto-prompted for guiding cross-domain segmentation. Specifically, we propose a Dual Prototype Anchor Transformation (DPAT) module that fuses pseudo query prototypes extracted based on cycle-consistency with support prototypes, allowing features to be transformed into a more stable domain-agnostic space. Additionally, a Meta Prompt Generator (MPG) module is introduced to automatically generate prompt embeddings, eliminating the need for manual visual prompts. We build an efficient model which can be applied directly to target domains without fine-tuning. Extensive experiments on four cross-domain datasets show that our model outperforms the state-of-the-art CD-FSS method by 5.24% and 3.10% in average accuracy on 1-shot and 5-shot settings, respectively. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

LDGCN: An Edge-End Lightweight Dual GCN Based on Single-Channel EEG for Driver Drowsiness Monitoring

学校读者我要写书评

暂无评论

IEEE Transactions on Consumer Electronics 2025年

作者： Huang, Jingwei Wang, Chuansheng Huang, Jiayan Fan, Haoyi Xue, Xingsi Zhang, Fuquan Fuzhou University College of Computer and Big Data Fuzhou China Polytechnic University of Catalonia Department of Automatic Control Barcelona Spain Putian University New Engineering Industry College Putian China Fujian University of Technology Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian Fuzhou350118 China Minjiang University Fujian Provincial Key Laboratory of Information Processing and Intelligent Control Fuzhou China Minjiang University College of Computer and Control Engineering Fuzhou China Sichuan Conservatory of Music Digital Media Art Key Laboratory of Sichuan Province Chengdu China Minjiang University Fuzhou Technology Innovation Center of Intelligent Manufacturing information System Fuzhou China Fuzhou China

Driver drowsiness electroencephalogram (EEG) signal monitoring can promptly alert drivers of their drowsiness in edge-end Consumer Electronics (CE), thereby reducing the probability of traffic accidents. Graph convolutional networks (GCNs) have shown significant advancements in processing the non-stationary, time-varying, and non-Euclidean nature of EEG signals. However, the existing single-channel EEG adjacency graph construction process lacks interpretability, which hinders the ability of GCNs to effectively extract adjacency graph features, thus affecting the performance of drowsiness monitoring. To address this issue, we propose an edge-end Lightweight Dual Graph Convolutional Network (LDGCN). Specifically, we are the first to incorporate neurophysiological knowledge to design a Baseline Drowsiness Status Adjacency Graph (BDSAG), which characterizes driver drowsiness status. Additionally, to express more features within limited EEG data, we introduce the Augmented Graph-level Module (AGM). This module captures global and local information at the graph level, ensuring that BDSAG features remain intact while enhancing effective feature expression capability. To further enhance the application of CE, we apply Adaptive Pruning Optimization (APO) to the channels and neurons of LDGCN, which nearly halves the inference latency. Experiments on benchmark datasets demonstrate that LDGCN offers the best trade-off between monitoring performance and hardware resource utilization, outperforming existing state-of-the-art algorithms. © 1975-2011 IEEE.

关键词： Neurons

Real-time stealth transmission via dither-based bias control

学校读者我要写书评

暂无评论

Real-time stealth transmission via dither-based bias control

Conference on Lasers and Electro-Optics (CLEO)

作者： Yuanxiang Wang Weidong Shao Linsheng Zhong Xiaoxiao Dai Qi Yang Lei Deng Deming Liu Mengfan Cheng Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology Wuhan China School of Optical and Electronic Information Huazhong University of Science and Technology Wuhan China

We report a real-time 1 kbps stealthy transmission in the 10 Gbps QPSK public communication. The stealth data is embedded in dither signals of bias control. The scheme is compatible with existing optical transmission ... 详细信息

ISBN: (纸本)9781957171258

关键词： Modulators Optical communications Optical networks Optical signals Quadrature phase shift keying Signal processing

Evolutionary Multiobjective Feature Selection Assisted by Unselected Features

学校读者我要写书评

暂无评论

Evolutionary Multiobjective Feature Selection Assisted by Un...

Congress on Evolutionary Computation

作者： Xuan Duan Songbai Liu Junkai Ji Lingjie Li Qiuzhen Lin Kay Chen Tan College of Computer Science and Software Engineering Shenzhen University Shenzhen China National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen China Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) Shenzhen University Shenzhen China Department of Computing The Hong Kong Polytechnic University Hong Kong SAR

ISBN: (数字)9798350308365

ISBN: (纸本)9798350308372

To enhance the generalization of multi-objective feature selection (MOFS) in classification, this paper proposes an evolutionary multitasking algorithm, diverging from previous approaches that exclusively target selected features. The algorithm integrates information from both selected and unselected features, introducing a novel objective to minimize the accuracy of unselected features. This objective, combined with the goal of minimizing classification errors for selected features, forms an auxiliary MOFS task. The paper presents a dual-population evolutionary multitasking framework that synergizes the main MOFS task with the auxiliary task. A knowledge transfer mechanism, based on accuracy preferences, seamlessly shares insights from the auxiliary to the main task, aiming to identify improved Pareto feature subsets. Empirical results demonstrate the superior performance of several state-of-the-art multi-objective algorithms within this framework, highlighting significant improvements across diverse datasets.

关键词： Accuracy Evolutionary computation Multitasking Feature extraction Classification algorithms Task analysis Knowledge transfer

UniFace: Unified Cross-Entropy Loss for Deep Face Recognition

学校读者我要写书评

暂无评论

UniFace: Unified Cross-Entropy Loss for Deep Face Recognitio...

International Conference on Computer Vision (ICCV)

作者： Jiancan Zhou Xi Jia Qiufu Li Linlin Shen Jinming Duan National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Computer Vision Institute Shenzhen University Aqara Lumi United Technology Co. Ltd. School of Computer Science University of Birmingham UK Alan Turing Institute UK

As a widely used loss function in deep face recognition, the softmax loss cannot guarantee that the minimum positive sample-to-class similarity is larger than the maximum negative sample-to-class similarity. As a result, no unified threshold is available to separate positive sample-to-class pairs from negative sample-to-class pairs. To bridge this gap, we design a UCE (Unified Cross-Entropy) loss for face recognition model training, which is built on the vital constraint that all the positive sample-to-class similarities shall be larger than the negative ones. Our UCE loss can be integrated with margins for a further performance boost. The face recognition model trained with the proposed UCE loss, UniFace, was intensively evaluated using a number of popular public datasets like MFR, IJB-C, LFW, CFP-FP, AgeDB, and MegaFace. Experimental results show that our approach outperforms SOTA methods like SphereFace, CosFace, ArcFace, Partial FC, etc. Especially, till the submission of this work (Mar. 8, 2023), the proposed UniFace achieves the highest TAR@MR-All on the academic track of the MFR-ongoing challenge. $\color{Blue}{\mathbf{Code}}$ is publicly available.

关键词：

Achieving High Open-Circuit Voltage in Efficient Kesterite Solar Cells Via Lanthanide Europium Ion Induced Carrier Lifetime Enhancement

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Chen, Xingye Zhao, Yunhai Ahmad, Nafees Zhao, Jun Zheng, Zhuanghao Su, Zhenghua Peng, Xiaogang Li, Xuejin Zhang, Xianghua Fan, Ping Liang, Guangxing Chen, Shuo Shenzhen Key Laboratory of Advanced Thin Films and Applications Key Laboratory of Optoelectronic Devices and Systems Ministry of Education and Guangdong Province College of Physics and Optoelectronic Engineering National Engineering Laboratory for Big Data System Computing Technology Shenzhen University Shenzhen518060 China UMR 6226 Université de Rennes Rennes F-35000 France Shenzhen518060 China

Short carrier lifetimes is a key challenge limiting the open-circuit voltage (VOC) and power conversion efficiency (PCE) of kesterite Cu2ZnSn(S,Se)4 (CZTSSe) solar cells. In this work, for the first time, lanthanide europium (Eu) ions have been introduced into Ag-incorporated CZTSSe absorber layer, which can modify the selenization reaction pathway during the initial selenization process, leading to high-quality (Ag, Eu)-CZTSSe absorber with large compact crystal grains and benign surface potential fluctuations. This innovative absorber engineering can also optimize defects dynamics with less detrimental defects and defects-assisted non-radiative recombination. Thus, significantly enhanced minority carrier lifetimes from 0.97 to 3.15 ns can be achieved, representing one of the top values among various bulk and/or interface engineering treated CZTSSe. The champion CZTSSe thin-film solar cell exhibits an impressive VOC of 545.6 mV, accompanied with a stimulating increase in PCE from 10.61% to 13.30%. These findings underscore the potential of lanthanide cation doping to drive significant advancements in CZTSSe photovoltaic technology, and therefore promoting its further development and future applications. © 2024, The Authors. All rights reserved.

关键词： Open circuit voltage