A key task in Artificial Intelligence is learning effective policies for controlling agents in unknown environments to optimize performance measures. Off-policy learning methods, like Q-learning, allow learners to make optimal decisions based on past experiences. This paper studies off-policy learning from biased data in complex and high-dimensional domains where \emph{unobserved confounding} cannot be ruled out a priori. Building on the well-celebrated Deep Q-Network (DQN), we propose a novel deep reinforcement learning algorithm robust to confounding biases in observed data. Specifically, our algorithm attempts to find a safe policy for the worst-case environment compatible with the observations. We apply our method to twelve confounded Atari games, and find that it consistently dominates the standard DQN in all games where the observed input to the behavioral and target policies mismatch and unobserved confounders exist.

Abstract--Recently, large models have shown significant potential for smart healthcare. However, the deployment of Large Vision-Language Models (L VLMs) for clinical services is currently hindered by three critical challenges: a tendency to hallucinate answers not grounded in visual evidence, the inefficiency of fixed-depth reasoning, and the difficulty of multi-institutional collaboration. T o address these challenges, in this paper, we develop MedAlign, a novel framework to ensure visually accurate L VLM responses for Medical Visual Question Answering (Med-VQA). Specifically, we first propose a mul-timodal Direct Preference Optimization (mDPO) objective to explicitly align preference learning with visual context. T o achieve adaptive reasoning and facilitate multi-institutional collaboration, we propose a federated governance mechanism, where the selected expert, fine-tuned on clinical datasets based on mDPO, locally performs iterative Chain-of-Thought (CoT) reasoning via the local meta-cognitive uncertainty estimator . Extensive experiments on three representative Med-VQA datasets demonstrate that MedAlign achieves state-of-the-art performance, outperforming strong retrieval-augmented baselines by up to 11.85% in F1-score, and simultaneously reducing the average reasoning length by 51.60% compared with fixed-depth CoT approaches. Su, and S. Xie are with the School of Automation, Guangdong University of Technology, Guangzhou, China (e-mails: 3122000875@mail2.gdut.edu.cn, J. Wen is with the College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China (e-mail: jinbo1608@nuaa.edu.cn). H. Pan and Z. Zhong are with State Key Laboratory of Traditional Chinese Medicine Syndrome, The Second Affiliated Hospital of Guangzhou University of Chinese Medicine, Guangdong Provincial Hospital of Chinese Medicine, Guangdong Provincial Academy of Chinese Medical Sciences, Guangzhou, China, and Chinese Medicine Guangdong Laboratory, Zhuhai, China (e-mails: hdpan@gzucm.edu.cn,

Molecular property prediction (MPP) is a cornerstone of drug discovery and materials science, yet conventional deep learning approaches depend on large labeled datasets that are often unavailable. Few-shot Molecular property prediction (FSMPP) addresses this scarcity by incorporating relational inductive bias through a context graph that links molecule nodes to property nodes, but such molecule-property graphs offer limited structural guidance. We propose a comprehensive solution: Motif Driven Global-Local Context Graph for few-shot molecular property prediction, which enriches contextual information at both the global and local levels. At the global level, chemically meaningful motif nodes representing shared substructures, such as rings or functional groups, are introduced to form a global tri-partite heterogeneous graph, yielding motif-molecule-property connections that capture long-range compositional patterns and enable knowledge transfer among molecules with common motifs. At the local level, we build a subgraph for each node in the molecule-property pair and encode them separately to concentrate the model's attention on the most informative neighboring molecules and motifs. Experiments on five standard FSMPP benchmarks demonstrate that our framework consistently outperforms state-of-the-art methods. These results underscore the effectiveness of integrating global motif knowledge with fine-grained local context to advance robust few-shot molecular property prediction.

Large language models are influencing the education landscape, with students relying on them in their learning process. Often implemented using general-purpose models, these systems are likely to give away the answers, which could hinder conceptual understanding and critical thinking. We study the role of automatic hint generation as a pedagogical strategy to promote active engagement with the learning content, while guiding learners toward the answers. Focusing on scientific topics at the secondary education level, we explore the potential of large language models to generate chains of hints that scaffold learners without revealing answers. We compare two distinct hinting strategies: static hints, pre-generated for each problem, and dynamic hints, adapted to learners' progress. Through a quantitative study with 41 participants, we uncover different preferences among learners with respect to hinting strategies, and identify the limitations of automatic evaluation metrics to capture them. Our findings highlight key design considerations for future research on hint generation and intelligent tutoring systems that seek to develop learner-centered educational technologies.

Federated Learning (FL) enables collaborative model training across institutions without sharing raw data. However, gradient sharing still risks privacy leakage, such as gradient inversion attacks. Homomorphic Encryption (HE) can secure aggregation but often incurs prohibitive computational and communication overhead. Existing HE-based FL methods sit at two extremes: encrypting all gradients for full privacy at high cost, or partially encrypting gradients to save resources while exposing vulnerabilities. We present DictPFL, a practical framework that achieves full gradient protection with minimal overhead. DictPFL encrypts every transmitted gradient while keeping non-transmitted parameters local, preserving privacy without heavy computation. It introduces two key modules: Decompose-for-Partial-Encrypt (DePE), which decomposes model weights into a static dictionary and an updatable lookup table, only the latter is encrypted and aggregated, while the static dictionary remains local and requires neither sharing nor encryption; and Prune-for-Minimum-Encrypt (PrME), which applies encryption-aware pruning to minimize encrypted parameters via consistent, history-guided masks. Experiments show that DictPFL reduces communication cost by 402-748$\times$ and accelerates training by 28-65$\times$ compared to fully encrypted FL, while outperforming state-of-the-art selective encryption methods by 51-155$\times$ in overhead and 4-19$\times$ in speed. Remarkably, DictPFL's runtime is within 2$\times$ of plaintext FL, demonstrating for the first time, that HE-based private federated learning is practical for real-world deployment. The code is publicly available at https://github.com/UCF-ML-Research/DictPFL.

Deploying deep neural networks on mobile devices is increasingly important but remains challenging due to limited computing resources. On the other hand, their unified memory architecture and narrower gap between CPU and GPU performance provide an opportunity to reduce inference latency by assigning tasks to both CPU and GPU. The main obstacles for such collaborative execution are the significant synchronization overhead required to combine partial results, and the difficulty of predicting execution times of tasks assigned to CPU and GPU (due to the dynamic selection of implementations and parallelism level). To overcome these obstacles, we propose both a lightweight synchronization mechanism based on OpenCL fine-grained shared virtual memory (SVM) and machine learning models to accurately predict execution times. Notably, these models capture the performance characteristics of GPU kernels and account for their dispatch times. A comprehensive evaluation on four mobile platforms shows that our approach can quickly select CPU-GPU co-execution strategies achieving up to 1.89x speedup for linear layers and 1.75x speedup for convolutional layers (close to the achievable maximum values of 2.01x and 1.87x, respectively, found by exhaustive grid search on a Pixel~5 smartphone).

Among many mysteries behind the success of deep networks lies the exceptional discriminative power of their learned representations as manifested by the intriguing Neural Collapse (NC) phenomenon, where simple feature structures emerge at the last layer of a trained neural network. Prior works on the theoretical understandings of NC have focused on analyzing the optimization landscape of matrix-factorization-like problems by considering the last-layer features as unconstrained free optimization variables and showing that their global minima exhibit NC. In this paper, we show that gradient flow on a two-layer ReLU network for classifying orthogonally separable data provably exhibits NC, thereby advancing prior results in two ways: First, we relax the assumption of unconstrained features, showing the effect of data structure and nonlinear activations on NC characterizations. Second, we reveal the role of the implicit bias of the training dynamics in facilitating the emergence of NC.

Abstract--Question Answering (QA) has seen significant improvements with the advancement of machine learning models, further studies enhanced this question answering system by retrieving external information, called Retrieval-Augmented Generation (RAG) to produce more accurate and informative answers. However, these state-of-the-art-performance is predominantly in English language. T o address this gap we made an effort of bridging language gaps by incorporating Adaptive RAG system to Indonesian language. Adaptive RAG system integrates a classifier whose task is to distinguish the question complexity, which in turn determines the strategy for answering the question. T o overcome the limited availability of Indonesian language dataset, our study employs machine translation as data augmentation approach. Experiments show reliable question complexity classifier; however, we observed significant inconsistencies in multi-retrieval answering strategy which negatively impacted the overall evaluation when this strategy was applied. Recent Large Language Models (LLMs) have shown incredible performance for a lot of Natural Language tasks. However, despite the advancement of LLMs in all tasks in natural language processing, they still have problems answering questions that require a knowledge-intensive background, often resulting in hallucination answers [7]. LLMs often provide accurate answers when entities mentioned in the question are present in their training data. Furthermore, the performance of the models has a significant correlation with the entity popularity; less popular entities are often not answered accurately by LLMs [8]. Updating the LLM's knowledge frequently is not a good solution since the training of LLM with billions or even trillions of data from all over the internet takes too much time. In contrast, recent studies have demonstrated that augmenting non-parametric knowledge (information not contained in the model's training data) to the question-answering method commonly referred to as Retrieval Augmented Generation (RAG) [9], even smaller models outperform larger models in terms of parameters [10].

Timely assessment of integrity of structures after seismic events is crucial for public safety and emergency response. This study focuses on assessing the structural damage conditions using deep learning methods to detect exposed steel reinforcement in concrete buildings and bridges after large earthquakes. Steel bars are typically exposed after concrete spalling or large flexural or shear cracks. The amount and distribution of exposed steel reinforcement is an indication of structural damage and degradation. To automatically detect exposed steel bars, new datasets of images collected after the 2023 Turkey Earthquakes were labeled to represent a wide variety of damaged concrete structures. The proposed method builds upon a deep learning framework, enhanced with fine-tuning, data augmentation, and testing on public datasets. An automated classification framework is developed that can be used to identify inside/outside buildings and structural components. Then, a YOLOv11 (You Only Look Once) model is trained to detect cracking and spalling damage and exposed bars. Another YOLO model is finetuned to distinguish different categories of structural damage levels. All these trained models are used to create a hybrid framework to automatically and reliably determine the damage levels from input images. This research demonstrates that rapid and automated damage detection following disasters is achievable across diverse damage contexts by utilizing image data collection, annotation, and deep learning approaches.

In precision-sensitive classification tasks, false positives carry severe operational consequences. For example, when a text safety classifier incorrectly flags 10% of benign user queries as unsafe, it blocks legitimate queries from being processed, degrading the experience for millions of users and potentially driving them away from the service. Similarly, in hallucination detection within Retrieval-Augmented Generation (RAG) pipelines, when factually correct responses are incorrectly flagged as hallucinated, the system triggers regeneration or self-correction mechanisms, adding unnecessary computational overhead and latency that frustrates users waiting for responses. These deployment realities demand classifiers that operate at extremely low false positive rates--often below 1%--while maintaining acceptable recall. Large language models are increasingly deployed for such precision-critical classification tasks through specialized safety guardrails like Llama Guard (Inan et al., 2023) and ShieldGemma (Zeng et al., 2024), as well as hallucination detection systems (Huang et al., 2025). Recently, reasoning-augmented approaches have emerged as a promising direction: GuardReasoner (Liu et al., 2025) incorporates chain-of-thought reasoning for safety classification, while Lynx (Ravi et al., 2024) leverages reasoning for hallucination detection in RAG