OccamVTS: Distilling Vision Models to 1% Parameters for Time Series Forecasting
Sisuo Lyu, Siru Zhong, Weilin Ruan, Qingxiang Liu, Qingsong Wen, Hui Xiong, Yuxuan Liang
[Paper]

• This work proposes OccamVTS, a novel knowledge distillation framework that extracts only the essential 1% of predictive information from large vision models (LVMs) to address the severe parameter redundancy and semantic misalignment challenges in time series forecasting by employing pre-trained LVMs as privileged teachers to guide lightweight student networks through pyramid-style feature alignment and selective distillation, effectively filtering out irrelevant high-level semantic noise while preserving crucial low-level temporal patterns.

Multi-view Hypergraph-based Contrastive Learning Model for Cold-Start Micro-video Recommendation (Oral)
Sisuo Lyu, Xiuze Zhou, Xuming Hu
[Paper]  [Code]  [PPT]  [Poster]

• This work proposes MHCR, the first model to leverage hypergraphs and contrastive learning for cold-start micro-video recommendation, implementing multi-view multimodal feature extraction with user-item graph, item-item affinity graph, and hypergraph layers combined with cross-modal and graph-hypergraph contrastive learning to address sparse interaction signals and over-smoothing challenges in platforms like TikTok and Kwai.

Improving Foundation Model Group Robustness with Auxiliary Sentence Embeddings (Under Review)
Hong Liu*, Sisuo Lyu*, Jie Li, Yan Teng, Yingchun Wang (Equal Contribution)

• This work proposes DoubleCCA, a novel framework that leverages auxiliary sentence embedding models to enhance group robustness of vision-language foundation models against spurious correlations, addressing critical bias mitigation challenges in trustworthy AI deployment through a two-stage canonical correlation analysis technique that first aligns augmented and original text embeddings in a shared semantic space, then reconstructs invariant features to merge with visual representations, effectively reducing sensitivity to group-based biases.

Image Matching Challenge 2022
Gold medal | 8/642
[Code]

• This work fused matching results from four models (LoFTR, SuperGlue, DKM, and QuadTreeAttention), and employed Thread PoolExecutor for efficient, multi-threaded inference on key points and F-matrix calculation within the competition’s time constraints.

Sorghum -100 Cultivar Identification - FGVC 9
3/252
[Code]
[Plog]

• This work utilized histogram equalization to improve image quality while employing IBN-Net integrates instance normalization (IN) and batch normalization (BN), bolstering model learning capacity and generalization to address competition challenges. Implementing ArcFace enhances the model’s discriminative ability for superior performance.

System Certification Risk Prediction
National First Prize | 1/1143
[Code]

• This work leverages LightGBM to enhance prediction performance, extracting crucial time features and improving the model through statistical crossover and categorical feature word embeddings. Employed hierarchical multi-fold cross-validation for superior generalization and applied normalization and fusion methods post-ranking to boost complementary capabilities.