Tung Kieu

Personalizing item selection during denoising boosts recommendation accuracy by leveraging user interactions more effectively than traditional methods.

Explicitly structuring contextual information in forecasting can lead to significantly improved accuracy in event-driven domains, outperforming traditional methods.

Historical patterns can dramatically improve time series imputation, with ALER-TI showing consistent performance gains over strong baselines.

Cultural values in LLMs can be subtly shifted through scenario-based dilemmas, revealing unexpected interdependencies that challenge traditional alignment approaches.

TimeBlocks outperforms traditional time-series models by dynamically constructing lightweight, adaptable models that excel in real-time processing.