Wei Yuan

LLM-aligned variable-length identifiers can revolutionize generative recommendation by providing a more nuanced and efficient representation of item semantics.

Self-Gating Attention achieves linear complexity in time series forecasting while maintaining competitive accuracy, revolutionizing the efficiency of attention mechanisms in this domain.

Now users can both selectively share *and* unshare their data in federated recommender systems, thanks to a novel contrastive unlearning mechanism that slashes storage costs.

Unleashing the full reasoning potential of VLMs, AgentM3D adaptively scales test-time reasoning paths to achieve state-of-the-art zero-shot multi-modal misinformation detection.