Aligning large language models (LLMs) is a central objective of post-training, often achieved through reward modeling and reinforcement learning methods. Among these, direct preference optimization (DPO) has emerged as a widely adopted technique that fine-tunes LLMs on preferred completions over less favorable ones. While most frontier LLMs do not disclose their curated preference pairs, the broader LLM community has released several open-source DPO datasets, including TuluDPO, ORPO, UltraFeedback, HelpSteer, and Code-Preference-Pairs. However, systematic comparisons remain scarce, largely due to the high computational cost and the lack of rich quality annotations, making it difficult to understand how preferences were selected, which task types they span, and how well they reflect human judgment on a per-sample level. In this work, we present the first comprehensive, data-centric analysis of popular open-source DPO corpora. We leverage the Magpie framework to annotate each sample for task category, input quality, and preference reward, a reward-model-based signal that validates the preference order without relying on human annotations. This enables a scalable, fine-grained inspection of preference quality across datasets, revealing structural and qualitative discrepancies in reward margins. Building on these insights, we systematically curate a new DPO mixture, UltraMix, that draws selectively from all five corpora while removing noisy or redundant samples. UltraMix is 30% smaller than the best-performing individual dataset yet exceeds its performance across key benchmarks. We publicly release all annotations, metadata, and our curated mixture to facilitate future research in data-centric preference optimization.

Dynamic facial expression recognition (DFER) aims to identify emotional states by modeling the temporal changes in facial movements across video sequences. A key challenge in DFER is the many-to-one labeling problem, where a video composed of numerous frames is assigned a single emotion label. A common strategy to mitigate this issue is to formulate DFER as a Multiple Instance Learning (MIL) problem. However, MIL-based approaches inherently suffer from the visual diversity of emotional expressions and the complexity of temporal dynamics. To address this challenge, we propose TG-DFER, a text-guided weakly supervised framework that enhances MILbased DFER by incorporating semantic guidance and coherent temporal modeling. We incorporate a vision-language pre-trained (VLP) model is integrated to provide semantic guidance through fine-grained textual descriptions of emotional context. Furthermore, we introduce visual prompts, which align enriched textual emotion labels with visual instance features, enabling fine-grained reasoning and frame-level relevance estimation. In addition, a multi-grained temporal network is designed to jointly capture short-term facial dynamics and long-range emotional flow, ensuring coherent affective understanding across time. Extensive results demonstrate that TG-DFER achieves improved generalization, interpretability, and temporal sensitivity under weak supervision. Introduction Facial expressions serve as direct indicators of human emotions, playing a crucial role in interpreting feelings during human interactions [4, 5]. Recognizing these expressions is essential in various fields, including human-computer interaction (HCI) [1], health assessment [2], and driver assistance systems [3].

LLM-based agents are increasingly deployed in multi-agent systems (MAS). As these systems move toward real-world applications, their security becomes paramount. Existing research largely evaluates single-agent security, leaving a critical gap in understanding the vulnerabilities introduced by multi-agent design. However, existing systems fall short due to lack of unified frameworks and metrics focusing on unique rejection modes in MAS. We present SafeAgents, a unified and extensible framework for fine-grained security assessment of MAS. SafeAgents systematically exposes how design choices such as plan construction strategies, inter-agent context sharing, and fallback behaviors affect susceptibility to adversarial prompting. We introduce Dharma, a diagnostic measure that helps identify weak links within multi-agent pipelines. Using SafeAgents, we conduct a comprehensive study across five widely adopted multi-agent architectures (centralized, decentralized, and hybrid variants) on four datasets spanning web tasks, tool use, and code generation. Our findings reveal that common design patterns carry significant vulnerabilities. For example, centralized systems that delegate only atomic instructions to sub-agents obscure harmful objectives, reducing robustness. Our results highlight the need for security-aware design in MAS. Link to code is https://github.com/microsoft/SafeAgents

Diffusion model (DM)-based channel estimation, which generates channel samples via a posteriori sampling stepwise with denoising process, has shown potential in high-precision channel state information (CSI) acquisition. However, slow sampling speed is an essential challenge for recent developed DM-based schemes. To alleviate this problem, we propose a novel flow matching (FM)-based generative model for multiple-input multiple-output (MIMO) channel estimation. We first formulate the channel estimation problem within FM framework, where the conditional probability path is constructed from the noisy channel distribution to the true channel distribution. In this case, the path evolves along the straight-line trajectory at a constant speed. Then, guided by this, we derive the velocity field that depends solely on the noise statistics to guide generative models training. Furthermore, during the sampling phase, we utilize the trained velocity field as prior information for channel estimation, which allows for quick and reliable noise channel enhancement via ordinary differential equation (ODE) Euler solver. Finally, numerical results demonstrate that the proposed FM-based channel estimation scheme can significantly reduce the sampling overhead compared to other popular DM-based schemes, such as the score matching (SM)-based scheme. Meanwhile, it achieves superior channel estimation accuracy under different channel conditions.

Brain-computer interface (BCI) speech decoding has emerged as a promising tool for assisting individuals with speech impairments. In this context, the integration of electroencephalography (EEG) and electromyography (EMG) signals offers strong potential for enhancing decoding performance. Mandarin tone classification presents particular challenges, as tonal variations convey distinct meanings even when phonemes remain identical. In this study, we propose a novel cross-subject multimodal BCI decoding framework that fuses EEG and EMG signals to classify four Mandarin tones under both audible and silent speech conditions. Inspired by the cooperative mechanisms of neural and muscular systems in speech production, our neural decoding architecture combines spatial-temporal feature extraction branches with a cross-attention fusion mechanism, enabling informative interaction between modalities. We further incorporate domain-adversarial training to improve cross-subject generalization. We collected 4,800 EEG trials and 4,800 EMG trials from 10 participants using only twenty EEG and five EMG channels, demonstrating the feasibility of minimal-channel decoding. Despite employing lightweight modules, our model outperforms state-of-the-art baselines across all conditions, achieving average classification accuracies of 87.83% for audible speech and 88.08% for silent speech. In cross-subject evaluations, it still maintains strong performance with accuracies of 83.27% and 85.10% for audible and silent speech, respectively. We further conduct ablation studies to validate the effectiveness of each component. Our findings suggest that tone-level decoding with minimal EEG-EMG channels is feasible and potentially generalizable across subjects, contributing to the development of practical BCI applications.

Modern text-to-speech (TTS) systems, particularly those built on Large Audio-Language Models (LALMs), generate high-fidelity speech that faithfully reproduces input text and mimics specified speaker identities. While prior misuse studies have focused on speaker impersonation, this work explores a distinct content-centric threat: exploiting TTS systems to produce speech containing harmful content. Realizing such threats poses two core challenges: (1) LALM safety alignment frequently rejects harmful prompts, yet existing jailbreak attacks are ill-suited for TTS because these systems are designed to faithfully vocalize any input text, and (2) real-world deployment pipelines often employ input/output filters that block harmful text and audio. We present HARMGEN, a suite of five attacks organized into two families that address these challenges. The first family employs semantic obfuscation techniques (Concat, Shuffle) that conceal harmful content within text. The second leverages audio-modality exploits (Read, Spell, Phoneme) that inject harmful content through auxiliary audio channels while maintaining benign textual prompts. Through evaluation across five commercial LALMs-based TTS systems and three datasets spanning two languages, we demonstrate that our attacks substantially reduce refusal rates and increase the toxicity of generated speech. We further assess both reactive countermeasures deployed by audio-streaming platforms and proactive defenses implemented by TTS providers. Our analysis reveals critical vulnerabilities: deepfake detectors underperform on high-fidelity audio; reactive moderation can be circumvented by adversarial perturbations; while proactive moderation detects 57-93% of attacks. Our work highlights a previously underexplored content-centric misuse vector for TTS and underscore the need for robust cross-modal safeguards throughout training and deployment.

Large language models (LLMs) have demonstrated capabilities across diverse domains, yet their performance on rare disease diagnosis from narrative medical cases remains underexplored. We introduce a novel dataset of 176 symptom-diagnosis pairs extracted from House M.D., a medical television series validated for teaching rare disease recognition in medical education. We evaluate four state-of-the-art LLMs such as GPT 4o mini, GPT 5 mini, Gemini 2.5 Flash, and Gemini 2.5 Pro on narrative-based diagnostic reasoning tasks. Results show significant variation in performance, ranging from 16.48% to 38.64% accuracy, with newer model generations demonstrating a 2.3 times improvement. While all models face substantial challenges with rare disease diagnosis, the observed improvement across architectures suggests promising directions for future development. Our educationally validated benchmark establishes baseline performance metrics for narrative medical reasoning and provides a publicly accessible evaluation framework for advancing AI-assisted diagnosis research.

The rapidly growing computation demands of deep neural networks (DNNs) have driven hardware vendors to integrate matrix multiplication accelerators (MMAs), such as NVIDIA Tensor Cores and AMD Matrix Cores, into modern GPUs. However, due to distinct and undocumented arithmetic specifications for floating-point matrix multiplication, some MMAs can lead to numerical imprecision and inconsistency that can compromise the stability and reproducibility of DNN training and inference. This paper presents MMA-Sim, the first bit-accurate reference model that reveals the detailed arithmetic behaviors of the MMAs from ten GPU architectures (eight from NVIDIA and two from AMD). By dissecting the MMAs using a combination of targeted and randomized tests, our methodology derives nine arithmetic algorithms to simulate the floating-point matrix multiplication of the MMAs. Large-scale validation confirms bitwise equivalence between MMA-Sim and the real hardware. Using MMA-Sim, we investigate arithmetic behaviors that affect DNN training stability, and identify undocumented behaviors that could lead to significant errors.

This paper explores the automatic classification of exam questions and learning outcomes according to Bloom's Taxonomy. A small dataset of 600 sentences labeled with six cognitive categories - Knowledge, Comprehension, Application, Analysis, Synthesis, and Evaluation - was processed using traditional machine learning (ML) models (Naive Bayes, Logistic Regression, Support Vector Machines), recurrent neural network architectures (LSTM, BiLSTM, GRU, BiGRU), transformer-based models (BERT and RoBERTa), and large language models (OpenAI, Gemini, Ollama, Anthropic). Each model was evaluated under different preprocessing and augmentation strategies (for example, synonym replacement, word embeddings, etc.). Among traditional ML approaches, Support Vector Machines (SVM) with data augmentation achieved the best overall performance, reaching 94 percent accuracy, recall, and F1 scores with minimal overfitting. In contrast, the RNN models and BERT suffered from severe overfitting, while RoBERTa initially overcame it but began to show signs as training progressed. Finally, zero-shot evaluations of large language models (LLMs) indicated that OpenAI and Gemini performed best among the tested LLMs, achieving approximately 0.72-0.73 accuracy and comparable F1 scores. These findings highlight the challenges of training complex deep models on limited data and underscore the value of careful data augmentation and simpler algorithms (such as augmented SVM) for Bloom's Taxonomy classification.

While large language models (LLMs) offer promising capabilities for automating academic workflows, existing systems for academic peer review remain constrained by text-only inputs, limited contextual grounding, and a lack of actionable feedback. In this work, we present an interactive web-based system for multimodal, community-aware peer review simulation to enable effective manuscript revisions before paper submission. Our framework integrates textual and visual information through multimodal LLMs, enhances review quality via retrieval-augmented generation (RAG) grounded in web-scale OpenReview data, and converts generated reviews into actionable to-do lists using the proposed Action:Objective[\#] format, providing structured and traceable guidance. The system integrates seamlessly into existing academic writing platforms, providing interactive interfaces for real-time feedback and revision tracking. Experimental results highlight the effectiveness of the proposed system in generating more comprehensive and useful reviews aligned with expert standards, surpassing ablated baselines and advancing transparent, human-centered scholarly assistance.