TLTrung Le

Trung Le

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Trung Le has a total of 109 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Perceptual Alignment Reveals Vulnerability in Unlearned Diffusion Models

    by Jing Wu, Trung Le, Mingyi Zhou, Dinh Phung, Jianfei Cai, Mehrtash Harandi

  • Co-steer: Cross-Modal Collaborative Steering for Jailbreaking MLLMs

    by Jingmin Zhu, Rollin Omari, Tamas Abraham, Junae Kim, Amardeep Kaur, Trung Le, Dinh Phung, Qiuhong Ke

  • Open Access

    Efficient Test-Time Scaling for LLM-based Time Series Forecasting

    by Xuan-May Le, Minh-Tuan Tran, Ling Luo, Uwe Aickelin, Dinh Phung, Trung Le

    Long-term time series forecasting benefits from preserving global structure such as trends and seasonality. Recent LLM-based forecasters often improve accuracy through test-time scaling (e.g., iterative refinement), but these methods are computationally expensive and increasingly prone to global-shape mismatch as the prediction horizon extends. We propose SCALER, a coarse-to-fine forecasting framework that first employs a lightweight Transformer tailored to long-term shape modeling to predict a coarse representation of future dynamics. This predicted shape then serves as a compact guide for an LLM to perform test-time scaling via iterative coarse-to-fine residual token refinement, while processing substantially fewer tokens at each step. By guiding refinement with an explicit future-shape prediction, SCALER reduces reliance on long description prompts, and its fixed-step refinement avoids costly reward-model-based selection, further lowering computational overhead. Experimental results demonstrate that SCALER outperforms strong forecasting baselines in long-term, short-term and zero-shot forecasting while significantly reducing the inference cost associated with scaled LLM for time series forecasting. Code: https://github.com/xuanmay2701/SCALER.

  • Open Access

    Ordering-based Causal Discovery via Generalized Score Matching

    by Vy Vo, Trung Le, He Zhao, Edwin Bonilla, Dinh Phung

  • Open Access

    CTPD: Cross Tokenizer Preference Distillation

    by Truong Nguyen, Phi Van Dat, Ngan Nguyen, Linh Ngo Van, Trung Le, Thanh Hong Nguyen

    While knowledge distillation has seen widespread use in pre-training and instruction tuning, its application to aligning language models with human preferences remains underexplored, particularly in the more realistic cross-tokenizer setting. The incompatibility of tokenization schemes between teacher and student models has largely prevented fine-grained, white-box distillation of preference information. To address this gap, we propose Cross-Tokenizer Preference Distillation (CTPD), the first unified framework for transferring human-aligned behavior between models with heterogeneous tokenizers. CTPD introduces three key innovations: (1) Aligned Span Projection, which maps teacher and student tokens to shared character-level spans for precise supervision transfer; (2) a cross-tokenizer adaptation of Token-level Importance Sampling (TIS-DPO) for improved credit assignment; and (3) a Teacher-Anchored Reference, allowing the student to directly leverage the teacher's preferences in a DPO-style objective. Our theoretical analysis grounds CTPD in importance sampling, and experiments across multiple benchmarks confirm its effectiveness, with significant performance gains over existing methods. These results establish CTPD as a practical and general solution for preference distillation across diverse tokenization schemes, opening the door to more accessible and efficient alignment of language models.

  • Open Access

    Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models

    by Dung Anh Hoang, Cuong Pham, Cuong Nguyen, Trung le, Jianfei Cai, Thanh-Toan Do

    Large Language Models (LLMs) deliver strong performance across a wide range of NLP tasks, but their massive sizes hinder deployment on resource-constrained devices. To reduce their computational and memory burden, various compression techniques have been proposed, including quantization, pruning, and knowledge distillation. Among these, post-training quantization (PTQ) is widely adopted for its efficiency, as it requires no retraining and only a small dataset for calibration, enabling low-cost deployment. Recent advances for post-training quantization have demonstrated that even sub-4-bit methods can maintain most of the original model performance. However, 1-bit quantization that converts floating-point weights to \(\pm\)1, remains particularly challenging, as existing 1-bit PTQ methods often suffer from significant performance degradation compared to the full-precision models. Specifically, most of existing 1-bit PTQ approaches focus on weight alignment, aligning the full-precision model weights with those of the quantized models, rather than directly aligning their outputs. Although the output-matching approach objective is more intuitive and aligns with the quantization goal, naively applying it in 1-bit LLMs often leads to notable performance degradation. In this paper, we investigate why and under what conditions output-matching fails, in the context of 1-bit LLM quantization. Based on our findings, we propose a novel data-aware PTQ approach for 1-bit LLMs that explicitly accounts for activation error accumulation while keeping optimization efficient. Empirical experiments demonstrate that our solution consistently outperforms existing 1-bit PTQ methods with minimal overhead.

  • DeepVulMatch: Learning and Matching Latent Vulnerability Representations for Dual-Granularity Vulnerability Detection

    by Michael Fu, Trung Le, Van Nguyen, Chakkrit Tantithamthavorn, Dinh Phung

    Deep learning (DL) models are widely used to detect software vulnerabilities, but identifying vulnerabilities at the line level remains challenging due to varied coding styles and the spread of vulnerabilities across multiple lines. We observe that vulnerable line embeddings tend to form clusters in the feature space, which can help models capture hidden patterns more effectively. In this article, we propose a novel approach that leverages vector quantization (VQ) and optimal transport (OT) to exploit the clustering characteristics of vulnerable line embeddings and enhance detection performance. Specifically, we extract vulnerable line embeddings from the training data to form a vulnerability collection, which we condense into a compact vulnerability codebook using VQ and OT. Inspired by static analysis tools that rely on pattern matching, our model uses this codebook to match latent vulnerability representations during inference. Our approach also introduces dual-granularity detection, predicting both vulnerable functions and, when a function is predicted vulnerable, identifying the specific vulnerable lines within it. We evaluate our approach against 12 baselines on two large-scale datasets of real-world open-source vulnerabilities. Our method achieves the highest F1 scores at both the function and line levels.


  • Open Access

    Preserving Clusters in Prompt Learning for Unsupervised Domain Adaptation

    by Tung-Long Vuong, Hoang Phan, Vy Vo, Anh Bui, Thanh-Toan Do, Trung Le, Dinh Phung

    Recent approaches leveraging multi-modal pre-trained models like CLIP for Unsupervised Domain Adaptation (UDA) have shown significant promise in bridging domain gaps and improving generalization by utilizing rich semantic knowledge and robust visual representations learned through extensive pre-training on diverse image-text datasets. While these methods achieve state-of-the-art performance across benchmarks, much of the improvement stems from base pseudo-labels (CLIP zero-shot predictions) and self-training mechanisms. Thus, the training mechanism exhibits a key limitation wherein the visual embedding distribution in target domains can deviate from the visual embedding distribution in the pre-trained model, leading to misguided signals from class descriptions. This work introduces a fresh solution to reinforce these pseudo-labels and facilitate target-prompt learning, by exploiting the geometry of visual and text embeddings - an aspect that is overlooked by existing methods. We first propose to directly leverage the reference predictions (from source prompts) based on the relationship between source and target visual embeddings. We later show that there is a strong clustering behavior observed between visual and text embeddings in pre-trained multi-modal models. Building on optimal transport theory, we transform this insight into a novel strategy to enforce the clustering property in text embeddings, further enhancing the alignment in the target domain. Our experiments and ablation studies validate the effectiveness of the proposed approach, demonstrating superior performance and improved quality of target prompts in terms of representation.

  • Open Access

    DAViT: A Domain-Adapted Vision Transformer for Automated Pneumonia Detection and Explanation Using Chest X-Ray Images

    by Michael Fu, Chakkrit Tantithamthavorn, Trung Le

    Pneumonia, a leading cause of mortality worldwide, especially among children under five, is typically diagnosed via chest X-rays. However, detecting it is challenging as expert radiologists must discern subtle patterns. Artificial intelligence (AI) offers a scalable alternative by automating diagnosis through deep learning (DL) models. Despite progress, current methods face two key limitations: (1) reliance on CNNs that capture local but may overlook global features, and (2) the use of pre-trained models from natural image datasets like ImageNet, which lack the contextual relevance of medical imaging, leading to suboptimal performance. To address these challenges, we propose DAViT (Domain-Adapted Vision Transformer), a hybrid architecture that combines Vision Transformers (ViTs) and shallow CNNs with domain adaptation. The ViT leverages self-attention to capture global features, while the CNN extracts local ones. To mitigate domain differences, we adapt the model using a diverse chest X-ray dataset. We evaluate DAViT on a real-world dataset of 5,856 chest X-rays. The results demonstrate that DAViT achieves state-of-the-art performance with a 97% F1-score and 96% AUC for pneumonia detection, outperforming twelve baseline methods. For pneumonia type classification, DAViT achieves an 81% F1-score and 84% AUC, outperforming baselines by 25% to 74%. An ablation study highlights the critical contributions of domain adaptation, ViT, and CNN components, collectively enhancing performance by 21%. Finally, we apply Grad-CAM on top of DAViT to generate interpretable heatmaps that highlight relevant areas for bacterial and viral pneumonia cases, providing insights to assist medical practitioners in decisionmaking. These findings indicate the potential of DAViT to assist clinicians in pneumonia diagnosis through improved model accuracy and interpretability. The training code and pre-trained models are available at https://github.com/awsm-research/DAViT.


  • Open Access

    Promoting Ensemble Diversity with Interactive Bayesian Distributional Robustness for Fine-tuning Foundation Models

    by Ngoc-Quan Pham, Tuan Truong, Quyen Tran, Tan Nguyen, Dinh Phung, Trung Le

    We introduce Interactive Bayesian Distributional Robustness (IBDR), a novel Bayesian inference framework that allows modeling the interactions between particles, thereby enhancing ensemble quality through increased particle diversity. IBDR is grounded in a generalized theoretical framework that connects the distributional population loss with the approximate posterior, motivating a practical dual optimization procedure that enforces distributional robustness while fostering particle diversity. We evaluate IBDR's performance against various baseline methods using the VTAB-1K benchmark and the common reasoning language task. The results consistently show that IBDR outperforms these baselines, underscoring its effectiveness in real-world applications.