NSTL国家科技图书文献中心

1. Adaptive Multi-head Contrastive Learning NSTL国家科技图书文献中心

Lei Wang | Piotr Koniusz ... - 《Computer Vision - ECCV 2024,Part LXIX》 - European Conference on Computer Vision - 2025, - 404～421 - 共18页

摘要：In contrastive* learning, two views of an* | Multi-Head Contrastive* Learning (AMCL), can be applied* | contrastive learning methods such as SimCLR, MoCo, and | original image, generated by different augmentations, are | considered a positive pair, and their similarity is

关键词： Contrastive learning | Similarity | Adaptive temperature

2. Contrastive Learning with Synthetic Positives NSTL国家科技图书文献中心

Dewen Zeng | Yawen Wu ... - 《Computer Vision - ECCV 2024,Part XXXVII》 - European Conference on Computer Vision - 2025, - 430～447 - 共18页

摘要：Contrastive learning with the nearest neighbor | introduce a novel approach called Contrastive Learning | -supervised learning* (SSL) techniques by utilizing the* | included as supplementary positives in the contrastive | -art methods. On transfer learning* benchmarks, CLSP*

关键词： Self-supervised learning | Contrastive learning | Diffusion model

3. Contrastive Learning-Based Sequential Recommendation Model NSTL国家科技图书文献中心

Yuan Zhang | Minghua Nuo ... - 《Natural Language Processing and Chinese Computing,Part IV》 - CCF International Conference on Natural Language Processing and Chinese Computing - 2025, - 28～40 - 共13页

摘要：Contrastive learning has demonstrated | a novel contrastive* learning paradigm for* | contrastive learning framework effectively captures intra | instances, thereby facilitating the learning* of more* | sequential recommendation, termed CLSRec (Contrastive

关键词： Sequential recommendation | Contrastive learning | Graph neural networks | Collaborative filtering

4. Graph Contrastive Learning for Multi-behavior Recommendation NSTL国家科技图书文献中心

Haiying Li | Huihui Wang ... - 《Advanced Data Mining and Applications,Part VI》 - International Conference on Advanced Data Mining and Applications - 2025, - 34～48 - 共15页

摘要： graph. The graph contrastive* learning task is applied* | . In this paper, we propose a novel Graph Contrastive | Learning for Multi-Behavior Recommendation (GCLMBR | adopt a multi-task learning* strategy to jointly* | optimize the learning* objectives, and predict the user*

关键词： Multi-Behavior recommendation | Graph contrastive learning | Graph convolutional network

5. Dual-Mode Contrastive Learning-Enhanced Knowledge Tracing NSTL国家科技图书文献中心

Danni Huang | Jicheng Yu ... - 《PRICAI 2024,Part I》 - Pacific Rim International Conference on Artificial Intelligence - 2025, - 81～92 - 共12页

摘要： propose a Dual-mode Contrastive* Learning-Enhanced* | . Furthermore, we employ the dual-mode contrastive learning | Knowledge Tracing (KT) aims to model learners | ' dynamic knowledge states and predict their future | response performance. Most of the existing KT models

关键词： Knowledge tracing | Exercise representation enhancement | Heterogeneous contrastive learning

6. Credit-Based Negative Sample Denoising in Contrastive Learning NSTL国家科技图书文献中心

Xingyu Yang | Lidong Yao ... - 《Pattern Recognition and Computer Vision,Part XI》 - Chinese Conference on Pattern Recognition and Computer Vision - 2025, - 23～35 - 共13页

摘要：Exploiting contrastive* learning to learn a* | problem in contrastive* learning, where large batch sizes* | fundamental tasks in unsupervised learning. The notorious | good representation becomes one of the promising | and long training epochs are required to train the

关键词： Representation learning | Contrastive learning | Unsupervised learning

7. ConMix: Contrastive Learning with Mixup Augmentation for Dialogue Summarization NSTL国家科技图书文献中心

Zequan Chen | Jing Xiao - 《Advanced Data Mining and Applications,Part V》 - International Conference on Advanced Data Mining and Applications - 2025, - 258～273 - 共16页

摘要： problem still remains. Contrastive* learning has been* | contrastive learning methods have often been less | contrastive learning framework called ConMix, which | model's sensitivity to contrastive* sample pairs. Our* | Seq2seq models have achieved remarkable

关键词： Dialogue summarization | Contrastive learning | Data augmentation

8. Counterfactual Contrastive Learning: Robust Representations via Causal Image Synthesis NSTL国家科技图书文献中心

Melanie Roschewitz | Fabio de Sousa Ribei...... - 《Data Engineering in Medical Imaging》 - International Workshop on Data Engineering in Medical Imaging | International Conference on Medical ImageComputing and Computer Assisted Intervention - 2025, - 22～32 - 共11页

摘要： contrastive learning approach which leverages approximate | Contrastive pretraining is well-known to | improve downstream task performance and model | generalisation, especially in limited label settings. However | , it is sensitive to the choice of augmentation

关键词： Contrastive learning | Counterfactuals | Model robustness

9. Self-Augmented Contrastive Learning for Knowledge-aware Recommendation NSTL国家科技图书文献中心

Guixin Chu | Xinye Guan ... - 《Database Systems for Advanced Applications,Part III》 - International Conference on Database Systems for Advanced Applications | International Workshop on Big Data Management and Service | International Workshop on Graph Data Management and Analysis | International Workshop on Big Data Quality Management | Workshop on Emerging Results inData Science and Engineering - 2025, - 261～276 - 共16页

摘要： founded on contrastive* learning (CL). However, the* | contrastive learning (SACL) that addresses long-tail | enable model-agnostic contrastive* learning through the* | -aware studies is to develop multi-task learning* models* | problems by learning* to supplement missing information in*

关键词： Knowledge graphs | Long-tail distribution | Transfer learning | Contrastive learning

10. Contrasting Deepfakes Diffusion via Contrastive Learning and Global-Local Similarities NSTL国家科技图书文献中心

Lorenzo Baraldi | Federico Cocchi ... - 《Computer Vision - ECCV 2024,Part LXIII》 - European Conference on Computer Vision - 2025, - 199～216 - 共18页

摘要： via contrastive* learning by additionally enforcing* | . In this study, we propose CoDE (Contrastive | Discerning between authentic content and that | generated by advanced AI methods has become increasingly | challenging. While previous research primarily addresses the

摘要：In contrastive* learning, two views of an* | Multi-Head Contrastive* Learning (AMCL), can be applied* | contrastive learning methods such as SimCLR, MoCo, and | original image, generated by different augmentations, are | considered a positive pair, and their similarity is

关键词： Contrastive learning | Similarity | Adaptive temperature

摘要：Contrastive learning with the nearest neighbor | introduce a novel approach called Contrastive Learning | -supervised learning* (SSL) techniques by utilizing the* | included as supplementary positives in the contrastive | -art methods. On transfer learning* benchmarks, CLSP*

关键词： Self-supervised learning | Contrastive learning | Diffusion model

摘要：Contrastive learning has demonstrated | a novel contrastive* learning paradigm for* | contrastive learning framework effectively captures intra | instances, thereby facilitating the learning* of more* | sequential recommendation, termed CLSRec (Contrastive

关键词： Sequential recommendation | Contrastive learning | Graph neural networks | Collaborative filtering

摘要： graph. The graph contrastive* learning task is applied* | . In this paper, we propose a novel Graph Contrastive | Learning for Multi-Behavior Recommendation (GCLMBR | adopt a multi-task learning* strategy to jointly* | optimize the learning* objectives, and predict the user*

关键词： Multi-Behavior recommendation | Graph contrastive learning | Graph convolutional network

摘要： propose a Dual-mode Contrastive* Learning-Enhanced* | . Furthermore, we employ the dual-mode contrastive learning | Knowledge Tracing (KT) aims to model learners | ' dynamic knowledge states and predict their future | response performance. Most of the existing KT models

关键词： Knowledge tracing | Exercise representation enhancement | Heterogeneous contrastive learning

摘要：Exploiting contrastive* learning to learn a* | problem in contrastive* learning, where large batch sizes* | fundamental tasks in unsupervised learning. The notorious | good representation becomes one of the promising | and long training epochs are required to train the

关键词： Representation learning | Contrastive learning | Unsupervised learning

摘要： problem still remains. Contrastive* learning has been* | contrastive learning methods have often been less | contrastive learning framework called ConMix, which | model's sensitivity to contrastive* sample pairs. Our* | Seq2seq models have achieved remarkable

关键词： Dialogue summarization | Contrastive learning | Data augmentation

摘要： contrastive learning approach which leverages approximate | Contrastive pretraining is well-known to | improve downstream task performance and model | generalisation, especially in limited label settings. However | , it is sensitive to the choice of augmentation

关键词： Contrastive learning | Counterfactuals | Model robustness

摘要： founded on contrastive* learning (CL). However, the* | contrastive learning (SACL) that addresses long-tail | enable model-agnostic contrastive* learning through the* | -aware studies is to develop multi-task learning* models* | problems by learning* to supplement missing information in*

关键词： Knowledge graphs | Long-tail distribution | Transfer learning | Contrastive learning

摘要： via contrastive* learning by additionally enforcing* | . In this study, we propose CoDE (Contrastive | Discerning between authentic content and that | generated by advanced AI methods has become increasingly | challenging. While previous research primarily addresses the

关键词： Deepfake detection | Contrastive learning

4008-161-200 800-990-8900

国家科技图书文献中心

摘要：In contrastive learning, two views of an | Multi-Head Contrastive Learning (AMCL), can be applied | contrastive learning methods such as SimCLR, MoCo, and | original image, generated by different augmentations, are | considered a positive pair, and their similarity is

关键词： Contrastive learning | Similarity | Adaptive temperature

摘要：Contrastive learning with the nearest neighbor | introduce a novel approach called Contrastive Learning | -supervised learning (SSL) techniques by utilizing the | included as supplementary positives in the contrastive | -art methods. On transfer learning benchmarks, CLSP

关键词： Self-supervised learning | Contrastive learning | Diffusion model

摘要：Contrastive learning has demonstrated | a novel contrastive learning paradigm for | contrastive learning framework effectively captures intra | instances, thereby facilitating the learning of more | sequential recommendation, termed CLSRec (Contrastive

关键词： Sequential recommendation | Contrastive learning | Graph neural networks | Collaborative filtering

摘要： graph. The graph contrastive learning task is applied | . In this paper, we propose a novel Graph Contrastive | Learning for Multi-Behavior Recommendation (GCLMBR | adopt a multi-task learning strategy to jointly | optimize the learning objectives, and predict the user

关键词： Multi-Behavior recommendation | Graph contrastive learning | Graph convolutional network

摘要： propose a Dual-mode Contrastive Learning-Enhanced | . Furthermore, we employ the dual-mode contrastive learning | Knowledge Tracing (KT) aims to model learners | ' dynamic knowledge states and predict their future | response performance. Most of the existing KT models

关键词： Knowledge tracing | Exercise representation enhancement | Heterogeneous contrastive learning

摘要：Exploiting contrastive learning to learn a | problem in contrastive learning, where large batch sizes | fundamental tasks in unsupervised learning. The notorious | good representation becomes one of the promising | and long training epochs are required to train the

关键词： Representation learning | Contrastive learning | Unsupervised learning

摘要： problem still remains. Contrastive learning has been | contrastive learning methods have often been less | contrastive learning framework called ConMix, which | model's sensitivity to contrastive sample pairs. Our | Seq2seq models have achieved remarkable

关键词： Dialogue summarization | Contrastive learning | Data augmentation

摘要： contrastive learning approach which leverages approximate | Contrastive pretraining is well-known to | improve downstream task performance and model | generalisation, especially in limited label settings. However | , it is sensitive to the choice of augmentation

关键词： Contrastive learning | Counterfactuals | Model robustness

摘要： founded on contrastive learning (CL). However, the | contrastive learning (SACL) that addresses long-tail | enable model-agnostic contrastive learning through the | -aware studies is to develop multi-task learning models | problems by learning to supplement missing information in

关键词： Knowledge graphs | Long-tail distribution | Transfer learning | Contrastive learning

摘要： via contrastive learning by additionally enforcing | . In this study, we propose CoDE (Contrastive | Discerning between authentic content and that | generated by advanced AI methods has become increasingly | challenging. While previous research primarily addresses the

关键词： Deepfake detection | Contrastive learning

4008-161-200 800-990-8900

国家科技图书文献中心

摘要：In contrastive* learning, two views of an* | Multi-Head Contrastive* Learning (AMCL), can be applied* | contrastive learning methods such as SimCLR, MoCo, and | original image, generated by different augmentations, are | considered a positive pair, and their similarity is

摘要：Contrastive learning with the nearest neighbor | introduce a novel approach called Contrastive Learning | -supervised learning* (SSL) techniques by utilizing the* | included as supplementary positives in the contrastive | -art methods. On transfer learning* benchmarks, CLSP*

摘要：Contrastive learning has demonstrated | a novel contrastive* learning paradigm for* | contrastive learning framework effectively captures intra | instances, thereby facilitating the learning* of more* | sequential recommendation, termed CLSRec (Contrastive

摘要： graph. The graph contrastive* learning task is applied* | . In this paper, we propose a novel Graph Contrastive | Learning for Multi-Behavior Recommendation (GCLMBR | adopt a multi-task learning* strategy to jointly* | optimize the learning* objectives, and predict the user*

摘要： propose a Dual-mode Contrastive* Learning-Enhanced* | . Furthermore, we employ the dual-mode contrastive learning | Knowledge Tracing (KT) aims to model learners | ' dynamic knowledge states and predict their future | response performance. Most of the existing KT models

摘要：Exploiting contrastive* learning to learn a* | problem in contrastive* learning, where large batch sizes* | fundamental tasks in unsupervised learning. The notorious | good representation becomes one of the promising | and long training epochs are required to train the

摘要： problem still remains. Contrastive* learning has been* | contrastive learning methods have often been less | contrastive learning framework called ConMix, which | model's sensitivity to contrastive* sample pairs. Our* | Seq2seq models have achieved remarkable

摘要： founded on contrastive* learning (CL). However, the* | contrastive learning (SACL) that addresses long-tail | enable model-agnostic contrastive* learning through the* | -aware studies is to develop multi-task learning* models* | problems by learning* to supplement missing information in*

摘要： via contrastive* learning by additionally enforcing* | . In this study, we propose CoDE (Contrastive | Discerning between authentic content and that | generated by advanced AI methods has become increasingly | challenging. While previous research primarily addresses the