arXiv subjects

Usman Naseem

Publications and source records attributed to Usman Naseem.

At least 73 records · Page 4Linked to original sources

arXiv2025

Kardia-R1: Unleashing LLMs to Reason toward Understanding and Empathy for Emotional Support via Rubric-as-Judge Reinforcement Learning

As web platforms evolve towards greater personalization and emotional complexity, conversational agents must transcend superficial empathy to demonstrate identity-aware emotional reasoning. However, existing systems face two limitations: (1) reliance on situation-centric datasets lacking persistent user identity, which hampers the capture of personalized affective nuances; and (2) dependence on opaque, coarse reward signals that hinder development of verifiable empathetic reasoning. To address these gaps, we introduce KardiaBench, a large-scale user-grounded benchmark comprising 178,080 QA pairs across 22,080 multi-turn conversations anchored to 671 real-world profiles. The dataset is constructed via a model-in-the-loop pipeline with iterative rubric-guided refinement to ensure psychological plausibility and persona consistency. This progressive empathy pipeline that integrates user comprehension, contextual reasoning, and emotion perception into conversations, followed by iterative critique and rubric-based refinement to ensure psychological plausibility, emotional fidelity, and persona consistency. Building on this, we propose Kardia-R1, a framework that trains models for interpretable, stepwise empathetic cognition. Kardia-R1 leverages Rubric-as-Judge Empathetic Reinforcement Learning (Rubric-ERL), a GRPO-based method that uses explainable, human-aligned rubric rewards to tightly couple user understanding, emotional inference, and supportive response generation. Extensive experiments across four LLM backbones demonstrate that Kardia-R1 consistently outperforms othet methods in emotion accuracy, empathy, relevance, persona consistency, and safety. Our dataset and model will be released at https://github.com/JhCircle/Kardia-R1.

cs.CL↗

arXiv2025

TurnBench-MS: A Benchmark for Evaluating Multi-Turn, Multi-Step Reasoning in Large Language Models

Despite impressive advances in large language models (LLMs), existing benchmarks often focus on single-turn or single-step tasks, failing to capture the kind of iterative reasoning required in real-world settings. To address this limitation, we introduce TurnBench, a novel benchmark that evaluates multi-turn, multi-step reasoning through an interactive code-breaking task inspired by the "Turing Machine Board Game." In each episode, a model must uncover hidden logical or arithmetic rules by making sequential guesses, receiving structured feedback, and integrating clues across multiple rounds. This dynamic setup requires models to reason over time, adapt based on past information, and maintain consistency across steps-capabilities underexplored in current benchmarks. TurnBench includes two modes: Classic, which tests standard reasoning, and Nightmare, which introduces increased complexity and requires robust inferential chains. To support fine-grained analysis, we provide ground-truth annotations for intermediate reasoning steps. Our evaluation of state-of-the-art LLMs reveals significant gaps: the best model achieves 84% accuracy in Classic mode, but performance drops to 18% in Nightmare mode. In contrast, human participants achieve 100% in both, underscoring the challenge TurnBench poses to current models. By incorporating feedback loops and hiding task rules, TurnBench reduces contamination risks and provides a rigorous testbed for diagnosing and advancing multi-step, multi-turn reasoning in LLMs.

cs.CL↗

arXiv2025

From Generation to Detection: A Multimodal Multi-Task Dataset for Benchmarking Health Misinformation

Infodemics and health misinformation have significant negative impact on individuals and society, exacerbating confusion and increasing hesitancy in adopting recommended health measures. Recent advancements in generative AI, capable of producing realistic, human like text and images, have significantly accelerated the spread and expanded the reach of health misinformation, resulting in an alarming surge in its dissemination. To combat the infodemics, most existing work has focused on developing misinformation datasets from social media and fact checking platforms, but has faced limitations in topical coverage, inclusion of AI generation, and accessibility of raw content. To address these issues, we present MM Health, a large scale multimodal misinformation dataset in the health domain consisting of 34,746 news article encompassing both textual and visual information. MM Health includes human-generated multimodal information (5,776 articles) and AI generated multimodal information (28,880 articles) from various SOTA generative AI models. Additionally, We benchmarked our dataset against three tasks (reliability checks, originality checks, and fine-grained AI detection) demonstrating that existing SOTA models struggle to accurately distinguish the reliability and origin of information. Our dataset aims to support the development of misinformation detection across various health scenarios, facilitating the detection of human and machine generated content at multimodal levels.

cs.CL↗

arXiv2025

Beyond the Black Box: Demystifying Multi-Turn LLM Reasoning with VISTA

Recent research has increasingly focused on the reasoning capabilities of Large Language Models (LLMs) in multi-turn interactions, as these scenarios more closely mirror real-world problem-solving. However, analyzing the intricate reasoning processes within these interactions presents a significant challenge due to complex contextual dependencies and a lack of specialized visualization tools, leading to a high cognitive load for researchers. To address this gap, we present VISTA, an web-based Visual Interactive System for Textual Analytics in multi-turn reasoning tasks. VISTA allows users to visualize the influence of context on model decisions and interactively modify conversation histories to conduct "what-if" analyses across different models. Furthermore, the platform can automatically parse a session and generate a reasoning dependency tree, offering a transparent view of the model's step-by-step logical path. By providing a unified and interactive framework, VISTA significantly reduces the complexity of analyzing reasoning chains, thereby facilitating a deeper understanding of the capabilities and limitations of current LLMs. The platform is open-source and supports easy integration of custom benchmarks and local models.

cs.CL↗

arXiv2025

Dynamic Traceback Learning for Medical Report Generation

Automated medical report generation has demonstrated the potential to significantly reduce the workload associated with time-consuming medical reporting. Recent generative representation learning methods have shown promise in integrating vision and language modalities for medical report generation. However, when trained end-to-end and applied directly to medical image-to-text generation, they face two significant challenges: i) difficulty in accurately capturing subtle yet crucial pathological details, and ii) reliance on both visual and textual inputs during inference, leading to performance degradation in zero-shot inference when only images are available. To address these challenges, this study proposes a novel multimodal dynamic traceback learning framework (DTrace). Specifically, we introduce a traceback mechanism to supervise the semantic validity of generated content and a dynamic learning strategy to adapt to various proportions of image and text input, enabling text generation without strong reliance on the input from both modalities during inference. The learning of cross-modal knowledge is enhanced by supervising the model to recover masked semantic information from a complementary counterpart. Extensive experiments conducted on two benchmark datasets, IU-Xray and MIMIC-CXR, demonstrate that the proposed DTrace framework outperforms state-of-the-art methods for medical report generation.

cs.CV↗

arXiv2025

Agentic Moderation: Multi-Agent Design for Safer Vision-Language Models

Agentic methods have emerged as a powerful and autonomous paradigm that enhances reasoning, collaboration, and adaptive control, enabling systems to coordinate and independently solve complex tasks. We extend this paradigm to safety alignment by introducing Agentic Moderation, a model-agnostic framework that leverages specialised agents to defend multimodal systems against jailbreak attacks. Unlike prior approaches that apply as a static layer over inputs or outputs and provide only binary classifications (safe or unsafe), our method integrates dynamic, cooperative agents, including Shield, Responder, Evaluator, and Reflector, to achieve context-aware and interpretable moderation. Extensive experiments across five datasets and four representative Large Vision-Language Models (LVLMs) demonstrate that our approach reduces the Attack Success Rate (ASR) by 7-19%, maintains a stable Non-Following Rate (NF), and improves the Refusal Rate (RR) by 4-20%, achieving robust, interpretable, and well-balanced safety performance. By harnessing the flexibility and reasoning capacity of agentic architectures, Agentic Moderation provides modular, scalable, and fine-grained safety enforcement, highlighting the broader potential of agentic systems as a foundation for automated safety governance.

cs.AI↗

arXiv2025

SHIELD: Classifier-Guided Prompting for Robust and Safer LVLMs

Large Vision-Language Models (LVLMs) unlock powerful multimodal reasoning but also expand the attack surface, particularly through adversarial inputs that conceal harmful goals in benign prompts. We propose SHIELD, a lightweight, model-agnostic preprocessing framework that couples fine-grained safety classification with category-specific guidance and explicit actions (Block, Reframe, Forward). Unlike binary moderators, SHIELD composes tailored safety prompts that enforce nuanced refusals or safe redirection without retraining. Across five benchmarks and five representative LVLMs, SHIELD consistently lowers jailbreak and non-following rates while preserving utility. Our method is plug-and-play, incurs negligible overhead, and is easily extendable to new attack types -- serving as a practical safety patch for both weakly and strongly aligned LVLMs.

cs.CL↗

arXiv2025

Steering Over-refusals Towards Safety in Retrieval Augmented Generation

Safety alignment in large language models (LLMs) induces over-refusals -- where LLMs decline benign requests due to aggressive safety filters. We analyze this phenomenon in retrieval-augmented generation (RAG), where both the query intent and retrieved context properties influence refusal behavior. We construct RagRefuse, a domain-stratified benchmark spanning medical, chemical, and open domains, pairing benign and harmful queries with controlled context contamination patterns and sizes. Our analysis shows that context arrangement / contamination, domain of query and context, and harmful-text density trigger refusals even on benign queries, with effects depending on model-specific alignment choices. To mitigate over-refusals, we introduce \textsc{SafeRAG-Steering}, a model-centric embedding intervention that steers the embedding regions towards the confirmed safe, non-refusing output regions at inference time. This reduces over-refusals in contaminated RAG pipelines while preserving legitimate refusals.

cs.CL↗

arXiv2025

LLMs on a Budget? Say HOLA

Running Large Language Models (LLMs) on edge devices is constrained by high compute and memory demands posing a barrier for real-time applications in sectors like healthcare, education, and embedded systems. Current solutions such as quantization, pruning, and retrieval-augmented generation (RAG) offer only partial optimizations and often compromise on speed or accuracy. We introduce HOLA, an end-to-end optimization framework for efficient LLM deployment. Internally, it leverages Hierarchical Speculative Decoding (HSD) for faster inference without quality loss. Externally, AdaComp-RAG adjusts retrieval complexity based on context needs. Together with LoBi, which blends structured pruning (LoRA) and quantization, HOLA delivers significant gains: 17.6% EMA on GSM8K, 10.5% MCA on ARC, and reduced latency and memory on edge devices like Jetson Nano--proving both scalable and production-ready.

cs.LG↗

arXiv2025

LLM-Based Multi-Task Bangla Hate Speech Detection: Type, Severity, and Target

Online social media platforms are central to everyday communication and information seeking. While these platforms serve positive purposes, they also provide fertile ground for the spread of hate speech, offensive language, and bullying content targeting individuals, organizations, and communities. Such content undermines safety, participation, and equity online. Reliable detection systems are therefore needed, especially for low-resource languages where moderation tools are limited. In Bangla, prior work has contributed resources and models, but most are single-task (e.g., binary hate/offense) with limited coverage of multi-facet signals (type, severity, target). We address these gaps by introducing the first multi-task Bangla hate-speech dataset, BanglaMultiHate, one of the largest manually annotated corpus to date. Building on this resource, we conduct a comprehensive, controlled comparison spanning classical baselines, monolingual pretrained models, and LLMs under zero-shot prompting and LoRA fine-tuning. Our experiments assess LLM adaptability in a low-resource setting and reveal a consistent trend: although LoRA-tuned LLMs are competitive with BanglaBERT, culturally and linguistically grounded pretraining remains critical for robust performance. Together, our dataset and findings establish a stronger benchmark for developing culturally aligned moderation tools in low-resource contexts. For reproducibility, we will release the dataset and all related scripts.

cs.CL↗

arXiv2025

Medical Question Summarization with Entity-driven Contrastive Learning

By summarizing longer consumer health questions into shorter and essential ones, medical question-answering systems can more accurately understand consumer intentions and retrieve suitable answers. However, medical question summarization is very challenging due to obvious distinctions in health trouble descriptions from patients and doctors. Although deep learning has been applied to successfully address the medical question summarization (MQS) task, two challenges remain: how to correctly capture question focus to model its semantic intention, and how to obtain reliable datasets to fairly evaluate performance. To address these challenges, this paper proposes a novel medical question summarization framework based on entity-driven contrastive learning (ECL). ECL employs medical entities present in frequently asked questions (FAQs) as focuses and devises an effective mechanism to generate hard negative samples. This approach compels models to focus on essential information and consequently generate more accurate question summaries. Furthermore, we have discovered that some MQS datasets, such as the iCliniq dataset with a 33% duplicate rate, have significant data leakage issues. To ensure an impartial evaluation of the related methods, this paper carefully examines leaked samples to reorganize more reasonable datasets. Extensive experiments demonstrate that our ECL method outperforms the existing methods and achieves new state-of-the-art performance, i.e., 52.85, 43.16, 41.31, 43.52 in terms of ROUGE-1 metric on MeQSum, CHQ-Summ, iCliniq, HealthCareMagic dataset, respectively. The code and datasets are available at https://github.com/yrbobo/MQS-ECL.

cs.CL↗

arXiv2025

Steering Towards Fairness: Mitigating Political Bias in LLMs

Recent advancements in large language models (LLMs) have enabled their widespread use across diverse real-world applications. However, concerns remain about their tendency to encode and reproduce ideological biases along political and economic dimensions. In this paper, we employ a framework for probing and mitigating such biases in decoder-based LLMs through analysis of internal model representations. Grounded in the Political Compass Test (PCT), this method uses contrastive pairs to extract and compare hidden layer activations from models like Mistral and DeepSeek. We introduce a comprehensive activation extraction pipeline capable of layer-wise analysis across multiple ideological axes, revealing meaningful disparities linked to political framing. Our results show that decoder LLMs systematically encode representational bias across layers, which can be leveraged for effective steering vector-based mitigation. This work provides new insights into how political bias is encoded in LLMs and offers a principled approach to debiasing beyond surface-level output interventions.

cs.CL↗

arXiv2025

Pluralistic Alignment for Healthcare: A Role-Driven Framework

As large language models are increasingly deployed in sensitive domains such as healthcare, ensuring their outputs reflect the diverse values and perspectives held across populations is critical. However, existing alignment approaches, including pluralistic paradigms like Modular Pluralism, often fall short in the health domain, where personal, cultural, and situational factors shape pluralism. Motivated by the aforementioned healthcare challenges, we propose a first lightweight, generalizable, pluralistic alignment approach, EthosAgents, designed to simulate diverse perspectives and values. We empirically show that it advances the pluralistic alignment for all three modes across seven varying-sized open and closed models. Our findings reveal that health-related pluralism demands adaptable and normatively aware approaches, offering insights into how these models can better respect diversity in other high-stakes domains.

cs.CL↗

arXiv2025

CUFG: Curriculum Unlearning Guided by the Forgetting Gradient

As privacy and security take center stage in AI, machine unlearning, the ability to erase specific knowledge from models, has garnered increasing attention. However, existing methods overly prioritize efficiency and aggressive forgetting, which introduces notable limitations. In particular, radical interventions like gradient ascent, influence functions, and random label noise can destabilize model weights, leading to collapse and reduced reliability. To address this, we propose CUFG (Curriculum Unlearning via Forgetting Gradients), a novel framework that enhances the stability of approximate unlearning through innovations in both forgetting mechanisms and data scheduling strategies. Specifically, CUFG integrates a new gradient corrector guided by forgetting gradients for fine-tuning-based unlearning and a curriculum unlearning paradigm that progressively forgets from easy to hard. These innovations narrow the gap with the gold-standard Retrain method by enabling more stable and progressive unlearning, thereby improving both effectiveness and reliability. Furthermore, we believe that the concept of curriculum unlearning has substantial research potential and offers forward-looking insights for the development of the MU field. Extensive experiments across various forgetting scenarios validate the rationale and effectiveness of our approach and CUFG. Codes are available at https://anonymous.4open.science/r/CUFG-6375.

cs.LG↗

arXiv2025

Too Helpful, Too Harmless, Too Honest or Just Right?

Large Language Models (LLMs) exhibit strong performance across a wide range of NLP tasks, yet aligning their outputs with the principles of Helpfulness, Harmlessness, and Honesty (HHH) remains a persistent challenge. Existing methods often optimize for individual alignment dimensions in isolation, leading to trade-offs and inconsistent behavior. While Mixture-of-Experts (MoE) architectures offer modularity, they suffer from poorly calibrated routing, limiting their effectiveness in alignment tasks. We propose TrinityX, a modular alignment framework that incorporates a Mixture of Calibrated Experts (MoCaE) within the Transformer architecture. TrinityX leverages separately trained experts for each HHH dimension, integrating their outputs through a calibrated, task-adaptive routing mechanism that combines expert signals into a unified, alignment-aware representation. Extensive experiments on three standard alignment benchmarks-Alpaca (Helpfulness), BeaverTails (Harmlessness), and TruthfulQA (Honesty)-demonstrate that TrinityX outperforms strong baselines, achieving relative improvements of 32.5% in win rate, 33.9% in safety score, and 28.4% in truthfulness. In addition, TrinityX reduces memory usage and inference latency by over 40% compared to prior MoE-based approaches. Ablation studies highlight the importance of calibrated routing, and cross-model evaluations confirm TrinityX's generalization across diverse LLM backbones.

cs.CL↗

arXiv2025

Competing LLM Agents in a Non-Cooperative Game of Opinion Polarisation

We introduce a novel non-cooperative game to analyse opinion formation and resistance, incorporating principles from social psychology such as confirmation bias, resource constraints, and influence penalties. Our simulation features Large Language Model (LLM) agents competing to influence a population, with penalties imposed for generating messages that propagate or counter misinformation. This framework integrates resource optimisation into the agents' decision-making process. Our findings demonstrate that while higher confirmation bias strengthens opinion alignment within groups, it also exacerbates overall polarisation. Conversely, lower confirmation bias leads to fragmented opinions and limited shifts in individual beliefs. Investing heavily in a high-resource debunking strategy can initially align the population with the debunking agent, but risks rapid resource depletion and diminished long-term influence

cs.AI↗

arXiv2025

A Survey on Progress in LLM Alignment from the Perspective of Reward Design

Reward design plays a pivotal role in aligning large language models (LLMs) with human values, serving as the bridge between feedback signals and model optimization. This survey provides a structured organization of reward modeling and addresses three key aspects: mathematical formulation, construction practices, and interaction with optimization paradigms. Building on this, it develops a macro-level taxonomy that characterizes reward mechanisms along complementary dimensions, thereby offering both conceptual clarity and practical guidance for alignment research. The progression of LLM alignment can be understood as a continuous refinement of reward design strategies, with recent developments highlighting paradigm shifts from reinforcement learning (RL)-based to RL-free optimization and from single-task to multi-objective and complex settings.

cs.CL↗

arXiv2025

ExBigBang: A Dynamic Approach for Explainable Persona Classification through Contextualized Hybrid Transformer Analysis

In user-centric design, persona development plays a vital role in understanding user behaviour, capturing needs, segmenting audiences, and guiding design decisions. However, the growing complexity of user interactions calls for a more contextualized approach to ensure designs align with real user needs. While earlier studies have advanced persona classification by modelling user behaviour, capturing contextual information, especially by integrating textual and tabular data, remains a key challenge. These models also often lack explainability, leaving their predictions difficult to interpret or justify. To address these limitations, we present ExBigBang (Explainable BigBang), a hybrid text-tabular approach that uses transformer-based architectures to model rich contextual features for persona classification. ExBigBang incorporates metadata, domain knowledge, and user profiling to embed deeper context into predictions. Through a cyclical process of user profiling and classification, our approach dynamically updates to reflect evolving user behaviours. Experiments on a benchmark persona classification dataset demonstrate the robustness of our model. An ablation study confirms the benefits of combining text and tabular data, while Explainable AI techniques shed light on the rationale behind the model's predictions.

cs.LG↗