Efficient Test-Time Scaling for LLM-based Time Series Forecasting
Xuan-May Le, Minh-Tuan Tran, Ling Luo, Uwe Aickelin, Dinh Phung, Trung Le
SIGKDD Conference on Knowledge Discovery and Data Mining, 2026
All research directions
Explainable & Efficient AI
Interpretable models and resource-efficient learning for deployable, trustworthy AI systems.
Goals
To make modern AI both understandable and practical, reducing compute cost while preserving transparency in how models reach their decisions.
Overview
This research pursues methods that improve the interpretability and efficiency of machine learning systems. It connects theoretical understanding of model behavior with algorithms that reduce training and inference cost for real-world deployment.
Key topics
- Model interpretability and transparency
- Efficient training and inference
- Test-time scaling and adaptation
- LLMs quantization
- Stable Diffusion Models quantization
News
Paper Efficient Test-Time Scaling for LLM-based Time Series Forecasting accepted at SIGKDD
Paper Layer-Wise High-Impact Parameter Ratio Optimization in Post-Training Quantization for Large Language Models accepted at at ACL
Paper Test-Time Instance-Specific Parameter Composition: A New Paradigm for Adaptive Generative Modeling accepted at CVPR
Paper Gradient-Aligned Calibration for Post-Training Quantization of Diffusion Models accepted at ICLR
Paper Mitigating Semantic Collapse in Generative Personalization with Test-Time Embedding Adjustment accepted at ICLR
Paper Geometry-Aware Collaborative Multi-Solutions Optimizer for Model Fine-Tuning with Parameter Efficiency accepted at NeurIPS
Paper Promoting Ensemble Diversity with Interactive Bayesian Distributional Robustness for Fine-tuning Foundation Models accepted at ICML
Paper Improving Generalization with Flat Hilbert Bayesian Inference accepted at ICML
Paper Preserving Clusters in Prompt Learning for Unsupervised Domain Adaptation accepted at CVPR
Paper NAYER: Noisy Layer Data Generation for Efficient and Effective Data-free Knowledge Distillation accepted at CVPR
Paper Text-Enhanced Data-free Approach for Federated Class-Incremental Learning accepted at CVPR
Paper MetaAug: Meta-Data Augmentation for Post-Training Quantization accepted at ICCV
(Best student paper award) Paper Feature-based Learning for Diverse and Privacy-Preserving Counterfactual Explanations accepted at SigKDD
Paper An Additive Instance-Wise Approach to Multi-class Model Interpretation accepted at ICLR
Papers in this direction
Layer-Wise High-Impact Parameter Ratio Optimization in Post-Training Quantization for Large Language Models
Cuong Pham, Anh Dung Hoang, Cuong C. Nguyen, Trung Le, Gustavo Carneiro, Thanh-Toan Do
Annual Meeting of the Association for Computational Linguistics (ACL 2026)Test-Time Instance-Specific Parameter Composition: A New Paradigm for Adaptive Generative Modeling
Minh-Tuan Tran, Xuan-May Le, Quan Hung Tran, Mehrtash Harandi, Dinh Phung, Trung Le
Conference on Computer Vision and Pattern Recognition 2026 (CVPR 2026)Gradient-Aligned Calibration for Post-Training Quantization of Diffusion Models
Dung Anh Hoang, Cuong Pham, Trung Le, Jianfei Cai, Thanh-Toan Do
International Conference on Representation Learning (ICLR 2026)Mitigating Semantic Collapse in Generative Personalization with Test-Time Embedding Adjustment
Anh Tuan Bui, Thuy-Trang Vu, Trung Le, Junae Kim, Tamas Abraham, Rollin Omari, Amardeep Kaur, Dinh Phung
International Conference on Representation Learning (ICLR 2026)Geometry-Aware Collaborative Multi-Solutions Optimizer for Model Fine-Tuning with Parameter Efficiency
Van-Anh Nguyen, Trung Le, Mehrtash Harandi, Ehsan Abbasnejad, Thanh-Toan Do, Dinh Phung
Conference on Neural Information Processing Systems (NeurIPS 2025)Promoting Ensemble Diversity with Interactive Bayesian Distributional Robustness for Fine-tuning Foundation Models
Tuan Truong, Ngoc-Quan Pham, Quyen Tran, Tan Minh Nguyen, Dinh Phung, Trung Le
International Conference on Machine Learning (ICML 2025)Improving Generalization with Flat Hilbert Bayesian Inference
Tuan Truong, Quyen Tran, Ngoc-Quan Pham, Dinh Phung, Nhat Ho, Trung Le
International Conference on Machine Learning (ICML 2025)Preserving Clusters in Prompt Learning for Unsupervised Domain Adaptation
Long Tung Vuong, Hoang Phan, Vy Vo, Anh Tuan Bui, Thanh-Toan Do, Dinh Phung, Trung Le
Conference on Computer Vision and Pattern Recognition 2025 (CVPR 2025)MetaAug: Meta-Data Augmentation for Post-Training Quantization
Cuong Pham, Dung Hoang, Cuong C. Nguyen, Trung Le, Dinh Phung, Gustavo Carneiro, Thanh-Toan Do
European Conference on Computer Vision 2024 (ECCV 2024)NAYER: Noisy Layer Data Generation for Efficient and Effective Data-free Knowledge Distillation
Tran Minh-Tuan, Trung Le, Xuan-May Le, Mehrtash Harandi, Quan Hung Tran, Dinh Phung
Conference on Computer Vision and Pattern Recognition 2024 (CVPR 2024)Text-Enhanced Data-free Approach for Federated Class-Incremental Learning
Minh-Tuan Tran, Trung Le, Xuan-May Le, Mehrtash Harandi, Dinh Phung
Conference on Computer Vision and Pattern Recognition 2024 (CVPR 2024)Feature-based Learning for Diverse and Privacy-Preserving Counterfactual Explanations
Vy Vo, Trung Le, Van Nguyen, He Zhao, Edwin Bonilla, Gholamreza Haffari, Dinh Phung
29th ACM SIGKDD Conference On Knowledge Discovery and Data Mining (KDD) 2023An Additive Instance-Wise Approach to Multi-class Model Interpretation
Vy Vo, Van Nguyen, Trung Le, Quan Hung Tran, Gholamreza Haffari, Seyit Camtepe, Dinh Phung
International Conference on Representation Learning (ICLR) 2023