Dung Duy Le

Pareto Set Learning (PSL) is an efficient approach for approximating the complete set of optimal solutions in Multi-objective Learning (MOL). By learning a set of solutions that map to a dense Pareto front in objective space, PSL aims to recover the underlying Pareto set. However, most existing methods focus heavily on convergence toward optimality, often neglecting solution diversity. To explicitly balance convergence and diversity, we propose Stein Variational Hypernetwork for MOL (SVH-MOL), a novel framework that incorporates Stein variational updates into PSL. SVH-MOL updates solutions via two complementary components: (i) a driving term that guides particles toward the Pareto set, and (ii) a repulsive term that encourages diversity among solutions. These two terms inherently compete, making stable and effective learning challenging. To address this issue, we introduce an annealing schedule that adaptively controls the relative strength of each term during training. Extensive experiments on synthetic multi-objective benchmarks and real-world multi-task learning problems demonstrate that SVH-MOL achieves a better tradeoff between convergence and diversity than existing methods. Our implementation is released at https://github.com/nguyenduc810/SVH-MOL-official .

Code-switching (CS), which is when Vietnamese speech uses English words like drug names or procedures, is a common phenomenon in Vietnamese medical communication. This creates challenges for Automatic Speech Recognition (ASR) systems, especially in low-resource languages like Vietnamese. Current most ASR systems struggle to recognize correctly English medical terms within Vietnamese sentences, and no benchmark addresses this challenge. In this paper, we construct a 34-hour \textbf{Vi}etnamese \textbf{Med}ical \textbf{C}ode-\textbf{S}witching \textbf{S}peech dataset (ViMedCSS) containing 16,576 utterances. Each utterance includes at least one English medical term drawn from a curated bilingual lexicon covering five medical topics. Using this dataset, we evaluate several state-of-the-art ASR models and examine different specific fine-tuning strategies for improving medical term recognition to investigate the best approach to solve in the dataset. Experimental results show that Vietnamese-optimized models perform better on general segments, while multilingual pretraining helps capture English insertions. The combination of both approaches yields the best balance between overall and code-switched accuracy. This work provides the first benchmark for Vietnamese medical code-switching and offers insights into effective domain adaptation for low-resource, multilingual ASR systems.

Agentic systems powered by Large Language Models (LLMs) have shown strong potential in recommender systems but remain hindered by several challenges. Fine-tuning LLMs is parameter-inefficient, and prompt-based agentic reasoning is limited by context length and hallucination risk. Moreover, existing agentic recommendation systems predominantly leverages semantic knowledge while neglecting the collaborative filtering (CF) signals essential for implicit preference modeling. To address these limitations, we propose AMEM4Rec, an agentic LLM-based recommender that learns collaborative signals in an end-to-end manner through cross-user memory evolution. AMEM4Rec stores abstract user behavior patterns from user histories in a global memory pool. Within this pool, memories are linked to similar existing ones and iteratively evolved to reinforce shared cross-user patterns, enabling the system to become aware of CF signals without relying on a pre-trained CF model. Extensive experiments on Amazon and MIND datasets show that AMEM4Rec consistently outperforms state-of-the-art LLM-based recommenders, demonstrating the effectiveness of evolving memory-guided collaborative filtering.

The discovery of crystalline reticular materials remains largely trial-and-error despite their societal importance. We introduce our algorithmic iterative reticular synthesis (AIRES) cycle, which integrates automated synthesis, image recognition, single-crystal X-ray diffraction and, crucially, customized algorithmic decision-making, to maximize distinct crystal discoveries rather than optimizing single targets. Demonstrated on zeolitic imidazolate frameworks (ZIFs), AIRES achieves twice the discovery rate of random exploration, crystallizing 10 new linkers into diverse ZIF topologies and expanding the single-linker Zn-ZIF library by one-third. By transforming reticular synthesis from an empirical process to a systematic exploration, AIRES provides a scalable and efficient blueprint for accelerating materials discovery.

The rapid growth of electric vehicles (EVs) necessitates the strategic placement of charging stations to optimize resource utilization and minimize user inconvenience. Reinforcement learning (RL) offers an innovative approach to identifying optimal charging station locations; however, existing methods face challenges due to their deterministic reward systems, which limit efficiency. Because real-world conditions are dynamic and uncertain, a deterministic reward structure cannot fully capture the complexities of charging station placement. As a result, evaluation becomes costly and time-consuming, and less reflective of real-world scenarios. To address this challenge, we propose a novel framework that integrates deep RL with agent-based simulations to model EV movement and estimate charging demand in real time. Our approach employs a hybrid RL agent with dual Q-networks to select optimal locations and configure charging ports, guided by a hybrid reward function that combines deterministic factors with simulation-derived feedback. Case studies in Hanoi, Vietnam, show that our method reduces average waiting times by 53.28% compared to the initial state, outperforming static baseline methods. This scalable and adaptive solution enhances EV infrastructure planning, effectively addressing real-world complexities and improving user experience.