殷荣，中国科学院信息工程研究所副研究员，专注于大模型，图表示学习，自监督表示学习，联邦学习，统计机器学习理论，高维数据分析/降维等。发表NeurIPS、ICML、AAAI、IJCAI、TIP、TKDE等高水平论文近30篇，其中CCF-A/中科院一区论文19篇 + 一作/通讯14篇。担任ICML、NeurIPS、AAAI、ICLR、MM、IJCAI、JMLR、NN、Cybersecurity等程序委员会委员/审稿人。担任《Cybersecurity》和《信息安全学报》学术期刊青年编委、中国计算机学会-人工智能与模式识别专委会委员、中国指挥与控制学会-情报与智能认知专委会委员。主持国家自然基金青年项目、 中国科学院 “特别研究助理资助项目”等。

课题组长期招收研究生，欢迎对大模型与图表示学习感兴趣的同学加入我们团队

   

081201-计算机系统结构
0812Z1-信息安全

人工智能，图神经网络，机器学习，无监督学习，统计学习理论，大数据处理
隐私保护

   

2023-12~2024-05,中国科学院信息工程研究所, 副研究员
2020-08~2023-12,中国科学院信息工程研究所, 预聘副研究员/博士后

2025-07-17-今,中国指挥与控制学会--情报与智能认知专委会, 委员
2025-06-04-今,中国计算机学会--人工智能与模式识别专委会, 委员
2024-03-01-今,《信息安全学报》学术期刊, 青年编委
2024-02-29-今,《Cybersecurity》学术期刊, 青年编委
2020-08-30-2024-06-29,中国科学院大学, 班主任

受邀担任人工智能、机器学习领域顶级会议/期刊 International Conference on Machine Learning (ICML)、Advances in Neural Information Processing Systems (NeurIPS)、AAAI Conference on Artificial Intelligence (AAAI)、International Conferenceon Learning Representations (ICLR)、ACM Multimedia (MM)、Journal of Machine Learning Research (JMLR)、Neural Networks (NN)、Cybersecurity 等程序委员会委员/审稿人。

(1) Multi-Modal Molecular Representation Learning via Structure Awareness. In IEEE Transactions on Image  Processing, 2025, 34: 3225-3238. (TIP 2025) (CCF-A，中科院一区，影响因子：10.8). 第 1 作者

(2) SSTAG: Structure-Aware Self-Supervised Learning Method for Text-Attributed Graphs. In Proceedings of Advances in Neural Information Processing Systems, 2025. (NeurIPS 2025) (CCF-A). 通讯作者

(3) AS-GCL: Asymmetric Spectral Augmentation on Graph Contrastive Learning. In IEEE Transaction on Multimedia, 2025. (TMM 2025) (中科院一区，影响因子：8.4). 通讯作者 

(4) MapFusion: A novel BEV feature fusion network for multi-modal map construction. In Information Fusion, 2025. (IF 2025) (中科院一区，影响因子：14.8). 通讯作者

(5) MSC-Bench: Benchmarking and Analyzing Multi-Sensor Corruption for Driving Perception. In IEEE International Conference on Multimedia & Expo. (ICME 2025) (CCF-B). 通讯作者

(6) SafeMap: Robust HD Map Construction from Incomplete Observations. In Proceedings of the 42th International Conference on Machine Learning. (ICML 2025) (CCF-A). 

(7) FedNK-RF: Federated Kernel Learning with Heterogeneous Data and Optimal Rates. In IEEE Transactions on Neural Networks and Learning Systems, 2025. (TNNLS 2025) (中科院一区，影响因子：8.9).

(8) Dual Prompt Clustering: Aligning and Adapting Multi-Views via Prompt Learning. Neurocomputing. (NC 2025) (中科院二区，影响因子：6.5)

(9) DADA++: Dual Alignment Domain Adaptation for Unsupervised Video-Text Retrieval. ACM Transactions on Multimedia Computing, Communications, and Applications, 2025. (TOMM 2025) (CCF-B，中科院三区，影响因子：5.1)

(10) What Really Matters for Robust Multi-Sensor HD Map Construction? In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems. (IROS 2025) (CORE-A).

(11) Improving Mathematical Reasoning Capabilities of Small Language Models via Feedback-Driven Distillation. In Proceedings of International Joint Conference on Neural Networks. (IJCNN 2025) (CORE-A).

(12) Improving Mathematical Reasoning Abilities of Small Language Models via Key-Point-Driven Distillation. In Proceedings of International Joint Conference on Neural Networks. (IJCNN 2025) (CORE-A).

(13) ASWT-SGNN: Adaptive Spectral Wavelet Transform-based Self-Supervised Graph Neural Network. In Proceedings of the 38th AAAI Conference on Artificial Intelligence, 2024, 38(12): 13990-13998. (AAAI 2024) (CCF-A). 通讯作者

(14) Unbiased and Augmentation-Free Self-Supervised Graph Representation Learning. In Pattern Recognition, 2024. (PR 2024) (中科院一区，影响因子：8). 通讯作者

(15) FTF-ER: Feature-Topology Fusion-Based Experience Replay Method for Continual Graph Learning. In Proceedings of the ACM Multimedia, 2024: 8336-8344. (MM 2024) (CCF-A).

(16) MapDistill: Boosting Efficient Camera-based HD Map Construction via Camera-LiDAR Fusion Model Distillation. In Proceedings of European Conference on Computer Vision, 2024: 166-183. (ECCV 2024) (CCF-B).

(17) Scalable Kernel k-Means with Randomized Sketching: From Theory to Algorithm. In IEEE Transactions on Knowledge and Data Engineering, 2023, 35(9): 9210-9224. (TKDE 2023) (CCF-A，中科院一区，影响因子：9.235). 第 1 作者

(18) Randomized Sketches for Clustering: Fast and Optimal Kernel k-Means. In Proceedings of Advances in Neural Information Processing Systems, 2022, 35: 6424-6436. (NeurIPS 2022) (CCF-A). 第 1 作者 

(19) Distributed Nystrom Kernel Learning with Communications. In Proceedings of the 28th International Conference on Machine Learning, PMLR, 2021: 12019-12028. (ICML 2021) (CCF-A). 第 1 作者 

(20) Distributed Randomized Sketching Kernel Learning. In Proceedings of the 36th AAAI Conference on Artificial Intelligence, 2022, 36(8): 8883-8891. (AAAI 2022) (CCF-A). 第 1 作者

(21) Divide-and-Conquer Learning with Nyström: Optimal Rate and Algorithm. In Proceedings of the 34th AAAI Conference on Artificial Intelligence, 2020, 34(04): 6696-6703. (AAAI 2020) (CCF-A). 第 1 作者

(22) Sketch Kernel Ridge Regression using Circulant Matrix: Algorithm and Theory. In IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(9): 3512-3524. (TNNLS 2020) (中科院一区，影响因子：11.683). 第 1 作者

(23) Extremely sparse Johnson-Lindenstrauss transform: From theory to algorithm. In Proceedings of IEEE International Conference on Data Mining, 2020: 1376-1381. (ICDM 2020) (CCF-B). 第 1 作者

(24) Triangle counting accelerations: From algorithm to in-memory computing architecture. IEEE Transactions on Computers, 2021, 71(10): 2462-2472. (TC 2021) (CCF-A，中科院一区，影响因子：3.7).

(25) Multi-Class Learning using Unlabeled Samples: Theory and Algorithm. In Proceedings of the 28th International Joint Conference on Artificial Intelligence, 2019: 2880-2886. (IJCAI 2019) (CCF-A). 

(26) Approximate Manifold Regularization: Scalable Algorithm and Generalization Analysis. In Proceedings of the 28th International Joint Conference on Artificial Intelligence, 2019: 2887-2893. (IJCAI 2019) (CCF-A).

(27) Multi-class learning: From theory to algorithm. In Proceedings of Advances in Neural Information Processing Systems, 2018. (NeurIPS 2018) (CCF-A). 

(28) Hashing Based Prediction for Large-Scale Kernel Machine. In Proceedings of the International Conference on Computational Science, 2020: 496-509. (ICCS 2020) (CORE-A).

   

（ 1 ） 跨媒体传播安全监管数据聚合与协同技术, 参与, 国家任务, 2022-12--2025-11
（ 2 ） 基于随机方法的大规模核学习理论和算法研究, 负责人, 国家任务, 2022-01--2024-12
（ 3 ） 大规模深度核学习的理论与算法研究, 参与, 国家任务, 2021-01--2024-12
（ 4 ） 基于显式假设空间的大规模核方法高效计算研究, 负责人, 中国科学院计划, 2020-09--2023-08
（ 5 ） 基于贝叶斯优化的 DNN 模型结构自动学习研究, 参与, 其他, 2019-10--2020-12
（ 6 ） 深度神经网络结构自动搜索理论与算法研究, 参与, 中国科学院计划, 2019-09--2024-09
（ 7 ） 大数据和人工智能科技发展现状研究与趋势预测算子谱分析的模型选择方法, 参与, 国家任务, 2019-02--2019-07

刘茹悦 博士研究生

张智铭  博士研究生 

郑磊 硕士研究生

张卓倚帆 硕士研究生