NSTL国家科技图书文献中心

1. FEAttack: A Fast and Efficient Hard-Label Textual Attack Framework NSTL国家科技图书文献中心

Miaomiao Li | Jun Ma ... - 《Wireless Artificial Intelligent Computing Systems and Applications,Part II》 - International Conference on Wireless Artificial Intelligent Computing Systems and Applications - 2025, - 167～179 - 共13页

摘要：Hard-label textual attacks are realistic but | label information as a guide to generate adversarial | , we propose FEAttack, a Fast and Effective hard | -label* textual Attack framework for generating high* | challenging because they can only rely on the predicted

关键词： Hard-label | Textual attack | Two-stage optimization

2. Hard-Label Cryptanalytic Extraction of Neural Network Models NSTL国家科技图书文献中心

Yi Chen | Xiaoyang Dong ... - 《Advances in Cryptology - ASIACRYPT 2024,Part VIII》 - International Conference on the Theory and Application of Cryptology and Information Security - 2025, - 207～236 - 共30页

摘要： neural networks operate under a hard-label* setting* | hard-label setting, which applies to ReLU neural | The machine learning problem of extracting | neural network parameters has been proposed for nearly | three decades. Functionally equivalent extraction is a

关键词： Cryptanalysis | ReLu neural networks | Functionally equivalent extraction | Hard-Label

3. Feature-Level Knowledge Distillation for Place Recognition Based on Soft-Hard Labels Teaching Paradigm NSTL国家科技图书文献中心

Zhenyu Li | Pengjie Xu ... - 《IEEE transactions on intelligent transportation systems》 - 2025,26(2) - 2091～2101 - 共11页

摘要： soft-hard* label teaching feature-level knowledge* | -label* teaching and hard-label teaching, which* | The motivation of visual place recognition | (VPR) is to enable robots to identify and localize | specific places within an environment using visual cues

关键词： Adaptation models | Robots | Mobile robots | Feature extraction | Training | Knowledge engineering | Transformers | Navigation | Knowledge transfer | Target tracking

4. Meta-learning collaborative optimization for lifetime prediction of lithium-ion batteries considering label noise NSTL国家科技图书文献中心

Wang G. | Wang C.... - 《Journal of Energy Storage》 - 2025,107(Jan.) - 1.1～1.13 - 共13页

摘要：. Additionally, an information entropy-based hard label | model overfitting to label* noise. To address this* | noisy-labeled samples. Then, a soft label* generation* | , enhancing the accuracy and diversity of soft label | with label* noise, the proposed framework is proven*

关键词： Information entropy | Label noise | Lifetime prediction | Lithium-ion batteries | Meta-learning

5. A Quantum Unique Games Conjecture NSTL国家科技图书文献中心

Hamoon Mousavi | Taro Spirig - 《16th Innovations in Theoretical Computer Science Conference,Part 3 of 3》 - Innovations in Theoretical Computer Science Conference - 2025, - 共16页

摘要： remain hard* to approximate. While the quantum nonlocal* | to be hard* - indeed undecidable - their* | introduce definitions for the quantum extensions of Label | -Cover and Unique-Label-Cover. We show that these | After the NP-hardness of computational

关键词： Hardness of approximation | Quantum computing | Noncommutative constraint satisfaction problems | Fourier analysis

6. MFLSCI: Multi-granularity fusion and label semantic correlation information for multi-label legal text classification NSTL国家科技图书文献中心

Chunyun Meng | Yuki Todo ... - 《Engineering Applications of Artificial Intelligence》 - 2025,139(Jan. Pt.B) - 109604.1～109604.13 - 共13页

摘要：Multi-label* text classification tasks face* | need for effective utilization of label* correlations* | label semantic correlation information. Our model | leverages graph convolutional networks to extract label | label omission issues. Additionally, text

关键词： Multi-label classification | Legal text processing | Graph convolution neural network | Natural language processing

7. Improving Time Series Classification with Representation Soft Label Smoothing NSTL国家科技图书文献中心

Hengyi Ma | Weitong Chen - 《Advanced Data Mining and Applications,Part IV》 - International Conference on Advanced Data Mining and Applications - 2025, - 297～310 - 共14页

摘要： predictions, such as label* smoothing and confidence penalty* | . Building upon the concept of label* smoothing, we propose* | labels, which we refer to as representation soft label | smoothing. We apply label* smoothing, confidence penalty* | , and our method representation soft label* smoothing*

关键词： Time series classification | Overfitting | Label smoothing

8. On approximate reconfigurability of label cover NSTL国家科技图书文献中心

Naoto Ohsaka - 《Information processing letters》 - 2025,189(Mar.) - 106556.1～106556.9 - 共9页

摘要： satisfying labelings _(mi)i and _(tar), the Label* Cover* | into _(tar) by repeatedly changing the label* of a* | , referred to as Maxmin Label* Cover Reconfiguration: We are* | Label Cover within any constant factor, gives strong | inapproximability results for many NP-hard* problems, one may think*

关键词： Our results suggest that a reconfiguration analogue of the parallel repetition theorem is

9. Cellular spatial-semantic embedding for multi-label classification of cell clusters in thyroid fine needle aspiration biopsy whole slide images NSTL国家科技图书文献中心

Gao, Juntao | Zhang, Jing ... - 《Pattern recognition letters》 - 2025,188(Feb.) - 125～132 - 共8页

摘要：Multi-label* classification of cell clusters is* | obstacles to the precise multi-label* classification of* | structures and diverse label* semantics in FNAB-WSI, we* | propose a multi-label* classification method of cell* | multi-label* semantic information. To address the*

关键词： Computer-aided diagnosis | Whole-slide image | Cell cluster | Multi-label classification | Spatial-semantic embedding

10. Meta-Learning Guided Label Noise Distillation for Robust Signal Modulation Classification NSTL国家科技图书文献中心

Xiaoyang Hao | Zhixi Feng ... - 《IEEE internet of things journal》 - 2025,12(1) - 402～418 - 共17页

摘要：, label* mislabeling often occurs in practical scenarios* | meta-learning guided label* noise distillation method* | label noise or errors. Specifically, we propose a | discriminate and distill label* noise. Following the notion* | performance of hard-to-classify categories with few-shot

摘要：Hard-label textual attacks are realistic but | label information as a guide to generate adversarial | , we propose FEAttack, a Fast and Effective hard | -label* textual Attack framework for generating high* | challenging because they can only rely on the predicted

关键词： Hard-label | Textual attack | Two-stage optimization

摘要： neural networks operate under a hard-label* setting* | hard-label setting, which applies to ReLU neural | The machine learning problem of extracting | neural network parameters has been proposed for nearly | three decades. Functionally equivalent extraction is a

关键词： Cryptanalysis | ReLu neural networks | Functionally equivalent extraction | Hard-Label

摘要： soft-hard* label teaching feature-level knowledge* | -label* teaching and hard-label teaching, which* | The motivation of visual place recognition | (VPR) is to enable robots to identify and localize | specific places within an environment using visual cues

关键词： Adaptation models | Robots | Mobile robots | Feature extraction | Training | Knowledge engineering | Transformers | Navigation | Knowledge transfer | Target tracking

摘要：. Additionally, an information entropy-based hard label | model overfitting to label* noise. To address this* | noisy-labeled samples. Then, a soft label* generation* | , enhancing the accuracy and diversity of soft label | with label* noise, the proposed framework is proven*

关键词： Information entropy | Label noise | Lifetime prediction | Lithium-ion batteries | Meta-learning

摘要： remain hard* to approximate. While the quantum nonlocal* | to be hard* - indeed undecidable - their* | introduce definitions for the quantum extensions of Label | -Cover and Unique-Label-Cover. We show that these | After the NP-hardness of computational

关键词： Hardness of approximation | Quantum computing | Noncommutative constraint satisfaction problems | Fourier analysis

摘要：Multi-label* text classification tasks face* | need for effective utilization of label* correlations* | label semantic correlation information. Our model | leverages graph convolutional networks to extract label | label omission issues. Additionally, text

关键词： Multi-label classification | Legal text processing | Graph convolution neural network | Natural language processing

关键词： Time series classification | Overfitting | Label smoothing

关键词： Our results suggest that a reconfiguration analogue of the parallel repetition theorem is

摘要：Multi-label* classification of cell clusters is* | obstacles to the precise multi-label* classification of* | structures and diverse label* semantics in FNAB-WSI, we* | propose a multi-label* classification method of cell* | multi-label* semantic information. To address the*

关键词： Computer-aided diagnosis | Whole-slide image | Cell cluster | Multi-label classification | Spatial-semantic embedding

摘要：, label* mislabeling often occurs in practical scenarios* | meta-learning guided label* noise distillation method* | label noise or errors. Specifically, we propose a | discriminate and distill label* noise. Following the notion* | performance of hard-to-classify categories with few-shot

关键词： Noise | Robustness | Training | Noise measurement | Modulation | Industrial Internet of Things | Accuracy

4008-161-200 800-990-8900

国家科技图书文献中心

摘要：Hard-label textual attacks are realistic but | label information as a guide to generate adversarial | , we propose FEAttack, a Fast and Effective hard | -label textual Attack framework for generating high | challenging because they can only rely on the predicted

关键词： Hard-label | Textual attack | Two-stage optimization

摘要： neural networks operate under a hard-label setting | hard-label setting, which applies to ReLU neural | The machine learning problem of extracting | neural network parameters has been proposed for nearly | three decades. Functionally equivalent extraction is a

关键词： Cryptanalysis | ReLu neural networks | Functionally equivalent extraction | Hard-Label

摘要： soft-hard label teaching feature-level knowledge | -label teaching and hard-label teaching, which | The motivation of visual place recognition | (VPR) is to enable robots to identify and localize | specific places within an environment using visual cues

关键词： Adaptation models | Robots | Mobile robots | Feature extraction | Training | Knowledge engineering | Transformers | Navigation | Knowledge transfer | Target tracking

摘要：. Additionally, an information entropy-based hard label | model overfitting to label noise. To address this | noisy-labeled samples. Then, a soft label generation | , enhancing the accuracy and diversity of soft label | with label noise, the proposed framework is proven

关键词： Information entropy | Label noise | Lifetime prediction | Lithium-ion batteries | Meta-learning

摘要： remain hard to approximate. While the quantum nonlocal | to be hard - indeed undecidable - their | introduce definitions for the quantum extensions of Label | -Cover and Unique-Label-Cover. We show that these | After the NP-hardness of computational

关键词： Hardness of approximation | Quantum computing | Noncommutative constraint satisfaction problems | Fourier analysis

摘要：Multi-label text classification tasks face | need for effective utilization of label correlations | label semantic correlation information. Our model | leverages graph convolutional networks to extract label | label omission issues. Additionally, text

关键词： Multi-label classification | Legal text processing | Graph convolution neural network | Natural language processing

关键词： Time series classification | Overfitting | Label smoothing

关键词： Our results suggest that a reconfiguration analogue of the parallel repetition theorem is

摘要：Multi-label classification of cell clusters is | obstacles to the precise multi-label classification of | structures and diverse label semantics in FNAB-WSI, we | propose a multi-label classification method of cell | multi-label semantic information. To address the

关键词： Computer-aided diagnosis | Whole-slide image | Cell cluster | Multi-label classification | Spatial-semantic embedding

摘要：, label mislabeling often occurs in practical scenarios | meta-learning guided label noise distillation method | label noise or errors. Specifically, we propose a | discriminate and distill label noise. Following the notion | performance of hard-to-classify categories with few-shot

关键词： Noise | Robustness | Training | Noise measurement | Modulation | Industrial Internet of Things | Accuracy

4008-161-200 800-990-8900

国家科技图书文献中心

摘要：Hard-label textual attacks are realistic but | label information as a guide to generate adversarial | , we propose FEAttack, a Fast and Effective hard | -label* textual Attack framework for generating high* | challenging because they can only rely on the predicted

摘要： neural networks operate under a hard-label* setting* | hard-label setting, which applies to ReLU neural | The machine learning problem of extracting | neural network parameters has been proposed for nearly | three decades. Functionally equivalent extraction is a

摘要： soft-hard* label teaching feature-level knowledge* | -label* teaching and hard-label teaching, which* | The motivation of visual place recognition | (VPR) is to enable robots to identify and localize | specific places within an environment using visual cues

摘要：. Additionally, an information entropy-based hard label | model overfitting to label* noise. To address this* | noisy-labeled samples. Then, a soft label* generation* | , enhancing the accuracy and diversity of soft label | with label* noise, the proposed framework is proven*

摘要： remain hard* to approximate. While the quantum nonlocal* | to be hard* - indeed undecidable - their* | introduce definitions for the quantum extensions of Label | -Cover and Unique-Label-Cover. We show that these | After the NP-hardness of computational

摘要：Multi-label* text classification tasks face* | need for effective utilization of label* correlations* | label semantic correlation information. Our model | leverages graph convolutional networks to extract label | label omission issues. Additionally, text

摘要：Multi-label* classification of cell clusters is* | obstacles to the precise multi-label* classification of* | structures and diverse label* semantics in FNAB-WSI, we* | propose a multi-label* classification method of cell* | multi-label* semantic information. To address the*

摘要：, label* mislabeling often occurs in practical scenarios* | meta-learning guided label* noise distillation method* | label noise or errors. Specifically, we propose a | discriminate and distill label* noise. Following the notion* | performance of hard-to-classify categories with few-shot