Peer Revenue Fully funded

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science Computation and Language

cs.CL Watch Computation and Language

Sunday 7 December

Computation and Language

Progress Ratio Embeddings: An Impatience Signal for Robust Length Control in Neural Text Generation

Ivanhoé Botcazou, Tassadit Amghar, Sylvain Lamprier et al. (+1 more)

Modern neural language models achieve high accuracy in text generation, yet precise control over generation length remains underdeveloped. In this paper, we first investigate a recent length control method based on Reverse Positional Embeddings (RPE) and show its limits when control is requested beyond the training distribution. In particular, using a discrete countdown signal tied to the absolute remaining token count leads to instability. To provide robust length control, we introduce Progress Ratio Embeddings (PRE), as continuous embeddings tied to a trigonometric impatience signal. PRE integrates seamlessly into standard Transformer architectures, providing stable length fidelity without degrading text accuracy under standard evaluation metrics. We further show that PRE generalizes well to unseen target lengths. Experiments on two widely used news-summarization benchmarks validate these findings.
Read the full abstract

All authors Ivanhoé Botcazou, Tassadit Amghar, Sylvain Lamprier, Frédéric Saubion

Computation and Language · Cross-listed from cs.CE

TxSum: User-Centered Ethereum Transaction Understanding with Micro-Level Semantic Grounding

Zifan Peng, Jingyi Zheng, Yule Liu et al. (+8 more)

Understanding the economic intent of Ethereum transactions is critical for user safety, yet current tools expose only raw on-chain data or surface-level intent, leading to widespread “blind signing” (approving transactions without understanding them). Through interviews with 16 Web3 users, we find that effective explanations should be structured, risk-aware, and grounded at the token-flow level. Motivated by these findings, we formulate TxSum, a new domain-grounded NLP task for DeFi transaction explanation, and construct a dataset of 187 complex Ethereum transactions with 2,375 token-flow annotations and transaction-level summaries. We further introduce MATEX, a grounded multi-agent framework for high-stakes transaction explanation. It selectively retrieves external knowledge under uncertainty and audits explanations against raw traces to improve token-flow-level factual consistency. MATEX achieves the strongest overall explanation quality, especially on micro-level factuality and intent quality. It improves user comprehension on complex transactions from 52.9% to 76.5% over the strongest baseline and raises malicious-transaction rejection from 36.0% to 88.0%, while maintaining a low false-rejection rate on benign transactions.
Read the full abstract

All authors Zifan Peng, Jingyi Zheng, Yule Liu, Huaiyu Jia, Qiming Ye, Jingyu Liu, Xufeng Yang, Mingchen Li, Qingyuan Gong, Xuechao Wang, Xinlei He

Computation and Language

XAM: Interactive Explainability for Authorship Attribution Models

Milad Alshomary, Anisha Bhatnagar, Peter Zeng et al. (+3 more)

We present IXAM, an Interactive eXplainability framework for Authorship Attribution Models. Given an authorship attribution (AA) task and an embedding-based AA model, our tool enables users to interactively explore the model's embedding space and construct an explanation of the model's prediction as a set of writing style features at different levels of granularity. Through a user evaluation, we demonstrate the value of our framework compared to predefined stylistic explanations.
Read the full abstract

All authors Milad Alshomary, Anisha Bhatnagar, Peter Zeng, Smaranda Muresan, Owen Rambow, Kathleen McKeown

Computation and Language

Large Language Models and Forensic Linguistics: Navigating Opportunities and Threats in the Age of Generative AI

George Mikros

Large language models (LLMs) present a dual challenge for forensic linguistics. They serve as powerful analytical tools enabling scalable corpus analysis and embedding-based authorship attribution, while simultaneously destabilising foundational assumptions about idiolect through style mimicry, authorship obfuscation, and the proliferation of synthetic texts. Recent stylometric research indicates that LLMs can approximate surface stylistic features yet exhibit detectable differences from human writers, a tension with significant forensic implications. However, current AI-text detection techniques, whether classifier-based, stylometric, or watermarking approaches, face substantial limitations: high false positive rates for non-native English writers and vulnerability to adversarial strategies such as homoglyph substitution. These uncertainties raise concerns under legal admissibility standards, particularly the Daubert and Kumho Tire frameworks. The article concludes that forensic linguistics requires methodological reconfiguration to remain scientifically credible and legally admissible. Proposed adaptations include hybrid human-AI workflows, explainable detection paradigms beyond binary classification, and validation regimes measuring error and bias across diverse populations. The discipline's core insight, i.e., that language reveals information about its producer, remains valid but must accommodate increasingly complex chains of human and machine authorship.
Read the full abstract

Computation and Language · Cross-listed from cs.CV

NeuroABench: A Multimodal Evaluation Benchmark for Neurosurgical Anatomy Identification

Ziyang Song, Zelin Zang, Xiaofan Ye et al. (+7 more)

Multimodal Large Language Models (MLLMs) have shown significant potential in surgical video understanding. With improved zero-shot performance and more effective human-machine interaction, they provide a strong foundation for advancing surgical education and assistance. However, existing research and datasets primarily focus on understanding surgical procedures and workflows, while paying limited attention to the critical role of anatomical comprehension. In clinical practice, surgeons rely heavily on precise anatomical understanding to interpret, review, and learn from surgical videos. To fill this gap, we introduce the Neurosurgical Anatomy Benchmark (NeuroABench), the first multimodal benchmark explicitly created to evaluate anatomical understanding in the neurosurgical domain. NeuroABench consists of 9 hours of annotated neurosurgical videos covering 89 distinct procedures and is developed using a novel multimodal annotation pipeline with multiple review cycles. The benchmark evaluates the identification of 68 clinical anatomical structures, providing a rigorous and standardized framework for assessing model performance. Experiments on over 10 state-of-the-art MLLMs reveal significant limitations, with the best-performing model achieving only 40.87% accuracy in anatomical identification tasks. To further evaluate the benchmark, we extract a subset of the dataset and conduct an informative test with four neurosurgical trainees. The results show that the best-performing student achieves 56% accuracy, with the lowest scores of 28% and an average score of 46.5%. While the best MLLM performs comparably to the lowest-scoring student, it still lags significantly behind the group's average performance. This comparison underscores both the progress of MLLMs in anatomical understanding and the substantial gap that remains in achieving human-level performance.
Read the full abstract

All authors Ziyang Song, Zelin Zang, Xiaofan Ye, Boqiang Xu, Long Bai, Jinlin Wu, Hongliang Ren, Hongbin Liu, Jiebo Luo, Zhen Lei

Computation and Language

Automated PRO-CTCAE Symptom Selection based on Prior Adverse Event Profiles

Francois Vandenhende, Anna Georgiou, Michalis Georgiou et al. (+2 more)

The PRO-CTCAE is an NCI-developed patient-reported outcome system for capturing symptomatic adverse events in oncology trials. It comprises a large library drawn from the CTCAE vocabulary, and item selection for a given trial is typically guided by expected toxicity profiles from prior data. Selecting too many PRO-CTCAE items can burden patients and reduce compliance, while too few may miss important safety signals. We present an automated method to select a minimal yet comprehensive PRO-CTCAE subset based on historical safety data. Each candidate PRO-CTCAE symptom term is first mapped to its corresponding MedDRA Preferred Terms (PTs), which are then encoded into Safeterm, a high-dimensional semantic space capturing clinical and contextual diversity in MedDRA terminology. We score each candidate PRO item for relevance to the historical list of adverse event PTs and combine relevance and incidence into a utility function. Spectral analysis is then applied to the combined utility and diversity matrix to identify an orthogonal set of medical concepts that balances relevance and diversity. Symptoms are rank-ordered by importance, and a cut-off is suggested based on the explained information. The tool is implemented as part of the Safeterm trial-safety app. We evaluate its performance using simulations and oncology case studies in which PRO-CTCAE was employed. This automated approach can streamline PRO-CTCAE design by leveraging MedDRA semantics and historical data, providing an objective and reproducible method to balance signal coverage against patient burden.
Read the full abstract

All authors Francois Vandenhende, Anna Georgiou, Michalis Georgiou, Theodoros Psaras, Ellie Karekla

Computation and Language

An Analysis of Large Language Models for Simulating User Responses in Surveys

Ziyun Yu, Yiru Zhou, Chen Zhao et al. (+1 more)

Using Large Language Models (LLMs) to simulate user opinions has received growing attention. Yet LLMs, especially trained with reinforcement learning from human feedback (RLHF), are known to exhibit biases toward dominant viewpoints, raising concerns about their ability to represent users from diverse demographic and cultural backgrounds. In this work, we examine the extent to which LLMs can simulate human responses to cross-domain survey questions through direct prompting and chain-of-thought prompting. We further propose a claim diversification method CLAIMSIM, which elicits viewpoints from LLM parametric knowledge as contextual input. Experiments on the survey question answering task indicate that, while CLAIMSIM produces more diverse responses, both approaches struggle to accurately simulate users. Further analysis reveals two key limitations: (1) LLMs tend to maintain fixed viewpoints across varying demographic features, and generate single-perspective claims; and (2) when presented with conflicting claims, LLMs struggle to reason over nuanced differences among demographic features, limiting their ability to adapt responses to specific user profiles.
Read the full abstract

All authors Ziyun Yu, Yiru Zhou, Chen Zhao, Hongyi Wen

Computation and Language

Role-aware Heuristic Episodic Attention for Conversational LLMs

Wanyang Hong, Zhaoning Zhang, Yi Chen et al. (+5 more)

Large language models often lose track of persistent instructions and relevant information as multi-turn conversations grow. We study this cumulative contextual decay through three related failure modes: attention pollution, dilution, and drift. We propose REA(Role-aware Heuristic Episodic Attention), a context-management framework that assigns different persistence and representation policies to instructions and episodic interactions. Instructional Memory retains identified global constraints in a dedicated prefix. Episodic Memory preserves user inputs and compresses model replies, while heuristic retrieval selects raw text, compressed representations, or omission for each historical turn. On Long-MT-Bench+, REA improves the judge score from 6.32 to 7.36 on a 10-point scale, a 16.5% relative gain over the Vanilla baseline, and reduces average latency by 2.91x. Additional evaluations show aggregate gains on three backbones spanning 1.7B-7B parameters and on Chinese and English role-playing tasks. These results support role-aware context management as a practical approach to maintaining conversational continuity and instruction adherence.
Read the full abstract

All authors Wanyang Hong, Zhaoning Zhang, Yi Chen, Libo Zhang, Baihui Liu, Linbo Qiao, Zhiliang Tian, Dongsheng Li

Computation and Language · Cross-listed from cs.CV

Less Is More, but Where? Dynamic Token Compression via LLM-Guided Keyframe Prior

Yulin Li, Haokun Gui, Ziyang Fan et al. (+4 more)

Recent advances in Video Large Language Models (VLLMs) have achieved remarkable video understanding capabilities, yet face critical efficiency bottlenecks due to quadratic computational growth with lengthy visual token sequences of long videos. While existing keyframe sampling methods can improve temporal modeling efficiency, additional computational cost is introduced before feature encoding, and the binary frame selection paradigm is found suboptimal. Therefore, in this work, we propose Dynamic Token compression via LLM-guided Keyframe prior (DyToK), a training-free paradigm that enables dynamic token compression by harnessing VLLMs' inherent attention mechanisms. Our analysis reveals that VLLM attention layers naturally encoding query-conditioned keyframe priors, by which DyToK dynamically adjusts per-frame token retention ratios, prioritizing semantically rich frames while suppressing redundancies. Extensive experiments demonstrate that DyToK achieves state-of-the-art efficiency-accuracy tradeoffs. DyToK shows plug-and-play compatibility with existing compression methods, such as VisionZip and FastV, attaining 4.3x faster inference while preserving accuracy across multiple VLLMs, such as LLaVA-OneVision and Qwen2.5-VL. Code is available at https://github.com/yu-lin-li/DyToK .
Read the full abstract

All authors Yulin Li, Haokun Gui, Ziyang Fan, Junjie Wang, Bin Kang, Bin Chen, Zhuotao Tian

Computation and Language

AquaFusionNet: Lightweight VisionSensor Fusion Framework for Real-Time Pathogen Detection and Water Quality Anomaly Prediction on Edge Devices

Sepyan Purnama Kristanto, Lutfi Hakim and Hermansyah

Evidence from many low and middle income regions shows that microbial contamination in small scale drinking water systems often fluctuates rapidly, yet existing monitoring tools capture only fragments of this behaviour. Microscopic imaging provides organism level visibility, whereas physicochemical sensors reveal shortterm changes in water chemistry; in practice, operators must interpret these streams separately, making realtime decision-making unreliable. This study introduces AquaFusionNet, a lightweight cross-modal framework that unifies both information sources inside a single edge deployable model. Unlike prior work that treats microscopic detection and water quality prediction as independent tasks, AquaFusionNet learns the statistical dependencies between microbial appearance and concurrent sensor dynamics through a gated crossattention mechanism designed specifically for lowpower hardware. The framework is trained on AquaMicro12K, a new dataset comprising 12,846 annotated 1000 micrographs curated for drinking water contexts, an area where publicly accessible microscopic datasets are scarce. Deployed for six months across seven facilities in East Java, Indonesia, the system processed 1.84 million frames and consistently detected contamination events with 94.8% mAP@0.5 and 96.3% anomaly prediction accuracy, while operating at 4.8 W on a Jetson Nano. Comparative experiments against representative lightweight detectors show that AquaFusionNet provides higher accuracy at comparable or lower power, and field results indicate that cross-modal coupling reduces common failure modes of unimodal detectors, particularly under fouling, turbidity spikes, and inconsistent illumination. All models, data, and hardware designs are released openly to facilitate replication and adaptation in decentralized water safety infrastructures.
Read the full abstract

Computation and Language

CAuSE: Decoding Multimodal Classifiers using Faithful Natural Language Explanation

Dibyanayan Bandyopadhyay, Soham Bhattacharjee, Mohammed Hasanuzzaman et al. (+1 more)

Multimodal classifiers function as opaque black box models. While several techniques exist to interpret their predictions, very few of them are as intuitive and accessible as natural language explanations (NLEs). To build trust, such explanations must faithfully capture the classifier's internal decision making behavior, a property known as faithfulness. In this paper, we propose CAuSE (Causal Abstraction under Simulated Explanations), a novel framework to generate faithful NLEs for any pretrained multimodal classifier. We demonstrate that CAuSE generalizes across datasets and models through extensive empirical evaluations. Theoretically, we show that CAuSE, trained via interchange intervention, forms a causal abstraction of the underlying classifier. We further validate this through a redesigned metric for measuring causal faithfulness in multimodal settings. CAuSE surpasses other methods on this metric, with qualitative analysis reinforcing its advantages. We perform detailed error analysis to pinpoint the failure cases of CAuSE. For replicability, we make the codes available at https://github.com/newcodevelop/CAuSE
Read the full abstract

All authors Dibyanayan Bandyopadhyay, Soham Bhattacharjee, Mohammed Hasanuzzaman, Asif Ekbal

Computation and Language

Large Language Model-Based Generation of Discharge Summaries

Tiago Rodrigues and Carla Teixeira Lopes

Discharge Summaries are documents written by medical professionals that detail a patient's visit to a care facility. They contain a wealth of information crucial for patient care, and automating their generation could significantly reduce the effort required from healthcare professionals, minimize errors, and ensure that critical patient information is easily accessible and actionable. In this work, we explore the use of five Large Language Models on this task, from open-source models (Mistral, Llama 2) to proprietary systems (GPT-3, GPT-4, Gemini 1.5 Pro), leveraging MIMIC-III summaries and notes. We evaluate them using exact-match, soft-overlap, and reference-free metrics. Our results show that proprietary models, particularly Gemini with one-shot prompting, outperformed others, producing summaries with the highest similarity to the gold-standard ones. Open-source models, while promising, especially Mistral after fine-tuning, lagged in performance, often struggling with hallucinations and repeated information. Human evaluation by a clinical expert confirmed the practical utility of the summaries generated by proprietary models. Despite the challenges, such as hallucinations and missing information, the findings suggest that LLMs, especially proprietary models, are promising candidates for automatic discharge summary generation as long as data privacy is ensured.
Read the full abstract

Computation and Language · Cross-listed from cs.CV

MMDuet2: Enhancing Proactive Interaction of Video MLLMs with Multi-Turn Reinforcement Learning

Yueqian Wang, Songxiang Liu, Disong Wang et al. (+4 more)

Recent advances in video multimodal large language models (Video MLLMs) have significantly enhanced video understanding and multi-modal interaction capabilities. While most existing systems operate in a turn-based manner where the model can only reply after user turns, proactively deciding when to reply during video playback presents a promising yet challenging direction for real-time applications. In this work, we propose a novel text-to-text approach to proactive interaction, where the model autonomously determines whether to respond or remain silent at each turn based on dialogue history and visual context up to current frame of an streaming video. To overcome difficulties in previous methods such as manually tuning response decision thresholds and annotating precise reply times, we introduce a multi-turn RL based training method that encourages timely and accurate responses without requiring precise response time annotations. We train our model MMDuet2 on a dataset of 52k videos with two types of dialogues via SFT and RL. Experimental results demonstrate that MMDuet2 outperforms existing proactive Video MLLM baselines in response timing and quality, achieving state-of-the-art performance on the ProactiveVideoQA benchmark.
Read the full abstract

All authors Yueqian Wang, Songxiang Liu, Disong Wang, Nuo Xu, Guanglu Wan, Huishuai Zhang, Dongyan Zhao

Computation and Language

LLM4SFC: Sequential Function Chart Generation via Large Language Models

Ofek Glick, Vladimir Tchuiev, Marah Ghoummaid et al. (+2 more)

While Large Language Models (LLMs) are increasingly used for synthesizing textual PLC programming languages like Structured Text (ST) code, other IEC 61131-3 standard graphical languages like Sequential Function Charts (SFCs) remain underexplored. Generating SFCs is challenging due to graphical nature and ST actions embedded within, which are not directly compatible with standard generation techniques, often leading to non-executable code that is incompatible with industrial tool-chains In this work, we introduce LLM4SFC, the first framework to receive natural-language descriptions of industrial workflows and provide executable SFCs. LLM4SFC is based on three components: (i) A reduced structured representation that captures essential topology and in-line ST and reduced textual verbosity; (ii) Fine-tuning and few-shot retrieval-augmented generation (RAG) for alignment with SFC programming conventions; and (iii) A structured generation approach that prunes illegal tokens in real-time to ensure compliance with the textual format of SFCs. We evaluate LLM4SFC on a dataset of real-world SFCs from automated manufacturing projects, using both open-source and proprietary LLMs. The results show that LLM4SFC reliably generates syntactically valid SFC programs effectively bridging graphical and textual PLC languages, achieving a generation generation success of 75% - 94%, paving the way for automated industrial programming.
Read the full abstract

All authors Ofek Glick, Vladimir Tchuiev, Marah Ghoummaid, Michal Moshkovitz, Dotan Di-Castro

Computation and Language

From Next-Token to Next-Block: A Principled Adaptation Path for Diffusion LLMs

Yuchuan Tian, Yuchen Liang, Shuo Zhang et al. (+10 more)

Diffusion Language Models (DLMs) enable fast generation, yet training large DLMs from scratch is costly. As a practical shortcut, adapting off-the-shelf Auto-Regressive (AR) model weights into a DLM could quickly equip the DLM with strong long-context generation capabilies. Prior "adaptation" attempts either modify logits or randomly grow attention masks to Full-Sequence diffusion, or simply transplant AR weights into a Block-Diffusion recipe, leaving two key questions unaddressed: where is the final destination of adaptation, and how to adapt better? For manifold benefits, we reframe the whole AR-to-DLM adaptation under the Block-Diffusion paradigm, transitioning from block size 1 to the final Block-Diffusion state. Concretely, the principled pathway of adaptation is designed as follows: we keep a context-causal path where causal attention is kept in the prefix, an efficient parallel adaptation procedure where an AR guidance is maintained, and gradual increment of the generation block size for a smoother transition. Built on these components, the adaptation is proved competitive on various models at different scales. With better adaptation, we propose NBDiff-7B that could inherit the long-context modeling and reasoning capabilities, and achieve state-of-the-art performance among the 7B-class DLMs. Codes: https://github.com/YuchuanTian/NBDiff.
Read the full abstract

All authors Yuchuan Tian, Yuchen Liang, Shuo Zhang, Yingte Shu, Guangwen Yang, Wei He, Sibo Fang, Tianyu Guo, Kai Han, Chao Xu, Hanting Chen, Xinghao Chen, Yunhe Wang

Computation and Language

Becoming Experienced Judges: Selective Test-Time Learning for Evaluators

Seungyeon Jwa, Daechul Ahn, Reokyoung Kim et al. (+2 more)

Automatic evaluation with large language models, commonly known as LLM-as-a-judge, is now standard across reasoning and alignment tasks. Despite evaluating many samples in deployment, these evaluators typically (i) treat each case independently, missing the opportunity to accumulate experience, and (ii) rely on a single fixed prompt for all cases, neglecting the need for sample-specific evaluation criteria. We introduce Learning While Evaluating (LWE), a framework that allows evaluators to improve sequentially at inference time without requiring training or validation sets. LWE maintains an evolving meta-prompt that (i) produces sample-specific evaluation instructions and (ii) refines itself through self-generated feedback. Furthermore, we propose Selective LWE, which updates the meta-prompt only on self-inconsistent cases, focusing computation where it matters most. This selective approach retains the benefits of sequential learning while being far more cost-effective. Across two pairwise comparison benchmarks, Selective LWE outperforms strong baselines, empirically demonstrating that evaluators can improve during sequential testing with a simple selective update, learning most from the cases they struggle with.
Read the full abstract

All authors Seungyeon Jwa, Daechul Ahn, Reokyoung Kim, Dongyeop Kang, Jonghyun Choi

Computation and Language

One Word Is Not Enough: Simple Prompts Improve Word Embeddings

Rajeev Ranjan

Text embedding models are designed for sentence-level applications like retrieval and semantic similarity, and are primarily evaluated on sentence-level benchmarks. Their behavior on isolated words is less understood. We show that simply prepending semantic prompts to words before embedding substantially improves word similarity correlations. Testing 7 text embedding models, including text-embedding-3-large (OpenAI), embed-english-v3.0 (Cohere), voyage-3(Voyage AI), all-mpnet-base-v2, and Qwen3-Embedding-8B, on 3 standard benchmarks (SimLex-999, WordSim-353, MEN-3000), we find that prompts like "meaning: {word}" or "Represent the semantic concept: {word}" improve Spearman correlations by up to +0.29 on SimLex-999. Some models fail completely on bare words (correlation = 0) but recover with prompts (+0.73 improvement). Our best results achieve correlation = 0.692 on SimLex-999 with embed-english-v3.0 (Cohere), correlation = 0.811 on WordSim-353, and correlation = 0.855 on MEN-3000 with text-embedding-3-large (OpenAI). These results outperform classic static embeddings like Word2Vec (correlation = 0.40) and even the best static method LexVec (correlation = 0.48) on SimLex-999, establishing a new state-of-the-art for pure embedding methods. This zero-shot technique requires no training and works with any text embedding model.
Read the full abstract

Computation and Language · Cross-listed from cs.LG

Arc Gradient Descent: A Geometrically Motivated Gradient Descent-based Optimiser with Phase-Aware, User-Controlled Step Dynamics (proof-of-concept)

Nikhil Verma, Joonas Linnosmaa, Leonardo Espinosa-Leal et al. (+1 more)

The paper presents the formulation, implementation, and evaluation of the ArcGD optimiser. The evaluation is conducted initially on a non-convex benchmark function and subsequently on a real-world ML dataset. The initial comparative study using the Adam optimiser is conducted on a stochastic variant of the highly non-convex and notoriously challenging Rosenbrock function, renowned for its narrow, curved valley, across dimensions ranging from 2D to 1000D and an extreme case of 50,000D. Two configurations were evaluated to eliminate learning-rate bias: (i) both using ArcGD's effective learning rate and (ii) both using Adam's default learning rate. ArcGD consistently outperformed Adam under the first setting and, although slower under the second, achieved superior final solutions in most cases. In the second evaluation, ArcGD is evaluated against state-of-the-art optimizers (Adam, AdamW, Lion, SGD) on the CIFAR-10 image classification dataset across 8 diverse MLP architectures ranging from 1 to 5 hidden layers. ArcGD achieved the highest average test accuracy (50.7%) at 20,000 iterations, outperforming AdamW (46.6%), Adam (46.8%), SGD (49.6%), and Lion (43.4%), winning or tying on 6 of 8 architectures. Notably, while Adam and AdamW showed strong early convergence at 5,000 iterations, but regressed with extended training, whereas ArcGD continued improving, demonstrating generalization and resistance to overfitting without requiring early stopping tuning. Strong performance on geometric stress tests and standard deep-learning benchmarks indicates broad applicability, highlighting the need for further exploration. Moreover, it is also shown that both a limiting variant of ArcGD and a momentum augmented ArcGD, recover sign-based momentum updates, revealing a clear conceptual link between ArcGD's phase structure and the core mechanism of the Lion Optimiser.
Read the full abstract

All authors Nikhil Verma, Joonas Linnosmaa, Leonardo Espinosa-Leal, Napat Vajragupta

Computation and Language

A Patient-Doctor-NLP-System to contest inequality for less privileged

Subrit Dikshit, Ritu Tiwari and Priyank Jain

Transfer Learning (TL) has accelerated the rapid development and availability of large language models (LLMs) for mainstream natural language processing (NLP) use cases. However, training and deploying such gigantic LLMs in resource-constrained, real-world healthcare situations remains challenging. This study addresses the limited support available to visually impaired users and speakers of low-resource languages such as Hindi who require medical assistance in rural environments. We propose PDFTEMRA (Performant Distilled Frequency Transformer Ensemble Model with Random Activations), a compact transformer-based architecture that integrates model distillation, frequency-domain modulation, ensemble learning, and randomized activation patterns to reduce computational cost while preserving language understanding performance. The model is trained and evaluated on medical question-answering and consultation datasets tailored to Hindi and accessibility scenarios, and its performance is compared against standard NLP state-of-the-art model baselines. Results demonstrate that PDFTEMRA achieves comparable performance with substantially lower computational requirements, indicating its suitability for accessible, inclusive, low-resource medical NLP applications.
Read the full abstract

Computation and Language

"The Dentist is an involved parent, the bartender is not": Revealing Implicit Biases in QA with Implicit BBQ

Aarushi Wagh and Saniya Srivastava

Existing benchmarks evaluating biases in large language models (LLMs) primarily rely on explicit cues, declaring protected attributes like religion, race, gender by name. However, real-world interactions often contain implicit biases, inferred subtly through names, cultural cues, or traits. This critical oversight creates a significant blind spot in fairness evaluation. We introduce ImplicitBBQ, a benchmark extending the Bias Benchmark for QA (BBQ) with implicitly cued protected attributes across 6 categories. Our evaluation of GPT-4o on ImplicitBBQ illustrates troubling performance disparity from explicit BBQ prompts, with accuracy declining up to 7% in the "sexual orientation" subcategory and consistent decline located across most other categories. This indicates that current LLMs contain implicit biases undetected by explicit benchmarks. ImplicitBBQ offers a crucial tool for nuanced fairness evaluation in NLP.
Read the full abstract

Computation and Language · Cross-listed from cs.CV

The Role of Entropy in Visual Grounding: Analysis and Optimization

Shuo Li, Jiajun Sun, Zhihao Zhang et al. (+10 more)

Recent advances in fine-tuning multimodal large language models (MLLMs) using reinforcement learning have achieved remarkable progress, particularly with the introduction of various entropy control techniques. However, the role and characteristics of entropy in perception-oriented tasks like visual grounding, as well as effective strategies for controlling it, remain largely unexplored. To address this issue, we focus on the visual grounding task and analyze the role and characteristics of entropy in comparison to reasoning tasks. Building on these findings, we introduce ECVGPO (Entropy Control Visual Grounding Policy Optimization), an interpretable algorithm designed for effective entropy regulation. Through entropy control, the trade-off between exploration and exploitation is better balanced. Experiments show that ECVGPO achieves broad improvements across various benchmarks and models.
Read the full abstract

All authors Shuo Li, Jiajun Sun, Zhihao Zhang, Xiaoran Fan, Senjie Jin, Hui Li, Yuming Yang, Junjie Ye, Lixing Shen, Tao Ji, Tao Gui, Qi Zhang, Xuanjing Huang

Computation and Language · Cross-listed from cs.AI

ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems in the Wild

Bufang Yang, Lilin Xu, Liekang Zeng et al. (+8 more)

Recent studies have begun to explore proactive large language model (LLM) agents that provide unobtrusive assistance by automatically leveraging contextual information, such as in code editing and in-app suggestions. However, most focus on short, task-specific episodes or on-screen contexts, rather than continuously perceiving and assisting users throughout daily life. Enabling such in-the-wild assistance requires continuous sensing of users' surroundings, which can incur substantial system overhead. In this work, we propose ProAgent, an end-to-end proactive agent system that harnesses on-demand sensory contexts to provide in-the-wild assistance. ProAgent first employs on-demand tiered perception to continuously sense users' surroundings by integrating low-cost contextual cues with richer perception on demand, and uses proactive-oriented context extraction to derive hierarchical contexts integrating both sensory contexts and human preferences. ProAgent then employs a context-aware proactive reasoner to infer user needs and invokes external tools to deliver proactive assistance. We implement ProAgent on AR glasses and evaluate it on a public dataset and a real-world dataset. Results demonstrate that ProAgent achieves up to 27.7% higher proactive prediction accuracy and 20.5% lower false detection than state-of-the-art baselines. A user study with 20 participants shows that 85% were satisfied with ProAgent and willing to use it in daily life.
Read the full abstract

All authors Bufang Yang, Lilin Xu, Liekang Zeng, Yunqi Guo, Siyang Jiang, Wenrui Lu, Kaiwei Liu, Yixuan Li, Xiaofan Jiang, Guoliang Xing, Zhenyu Yan

Computation and Language · Cross-listed from cs.AI

SIEVE: Selective Integrity Verification and Escalation for Defending LLM Agents against Indirect Prompt Injection

Zhibo Liang, Tianze Hu, Zaiye Chen et al. (+1 more)

Large Language Models (LLMs) are increasingly used as the core of agentic systems due to their strong reasoning, planning, and tool-use capabilities. By interacting with external environments, LLM agents can execute real-world tasks on behalf of users rather than merely generate text. This expanded capability also amplifies the threat of indirect prompt injection (IPI), where malicious external content can manipulate agent behavior and trigger unauthorized actions, privacy leakage, or financial loss. Existing defenses generally follow two approaches. Plan- or rule-based methods constrain agent execution using predefined plans or execution rules, but may block legitimate actions that arise from dynamic runtime context. Semantic auditing methods offer greater flexibility, yet repeatedly re-evaluating proposed actions incurs substantial token and latency overhead. These limitations motivate a selective verification strategy that applies deterministic checks whenever reliable structural evidence is available and escalates only ambiguous cases. Accordingly, we propose SIEVE, which verifies tool transitions and argument sources against an Intent Graph, escalating actions that cannot pass deterministic verification to semantic adjudication. This selective design preserves flexibility while avoiding continuous semantic auditing. We evaluate SIEVE on AgentLure and AgentDojo against MELON, DRIFT, IPIGuard, and ARGUS. SIEVE achieves 5.94% ASR with 97.5% clean utility on AgentLure, and 0.34% ASR with 87.63% clean utility on AgentDojo. Compared with DRIFT and ARGUS, SIEVE incurs the lowest token consumption and the fewest API calls.
Read the full abstract

All authors Zhibo Liang, Tianze Hu, Zaiye Chen, Mingjie Tang

Computation and Language · Cross-listed from cs.CR

Look Twice before You Leap: A Rational Framework for Localized Adversarial Anonymization

Donghang Duan, Xu Zheng, Yuefeng He et al. (+3 more)

Current LLM-based frameworks for text anonymization usually rely on remote API services from powerful LLMs, which creates an inherent privacy paradox: users must disclose the raw data to untrusted third parties for guaranteed privacy preservation. Moreover, directly migrating current solutions to local small-scale models (LSMs) offers a suboptimal solution with severe utility collapse. Our work argues that this failure stems not merely from the capability deficits of LSMs, but significantly from the inherent irrationality of the greedy adversarial strategies employed by current state-of-the-art (SOTA) methods. To address this drawback, we propose Rational Localized Adversarial Anonymization (RLAA), a fully localized and training-free framework featuring an Attacker-Arbitrator-Anonymizer architecture. We model the anonymization process as a trade-off between Marginal Privacy Gain (MPG) and Marginal Utility Cost (MUC), demonstrating that greedy strategies tend to drift into an irrational state. Instead, RLAA introduces an arbitrator that acts as a rationality gatekeeper, validating the attacker's inference to filter out ghost leaks. This mechanism promotes a rational early-stopping criterion, and structurally prevents utility collapse. Extensive experiments on different benchmarks demonstrate that RLAA achieves a superior privacy-utility trade-off compared to strong baselines.
Read the full abstract

All authors Donghang Duan, Xu Zheng, Yuefeng He, Chong Mu, Leyi Cai, Lizong Zhang

Computation and Language

Parameter-Efficient Fine-Tuning with Differential Privacy for Robust Instruction Adaptation in Large Language Models

Yulin Huang, Yaxuan Luan, Jinxu Guo et al. (+2 more)

This study addresses the issues of privacy protection and efficiency in instruction fine-tuning of large-scale language models by proposing a parameter-efficient method that integrates differential privacy noise allocation with gradient clipping in a collaborative optimization framework. The method keeps the backbone model frozen and updates parameters through a low-dimensional projection subspace, while introducing clipping and adaptive noise allocation during gradient computation. This design reduces privacy budget consumption and ensures training stability and robustness. The unified framework combines gradient constraints, noise allocation, and parameter projection, effectively mitigating performance fluctuations and privacy risks in multi-task instruction scenarios. Experiments are conducted across hyperparameter, environment, and data sensitivity dimensions. Results show that the method outperforms baseline models in accuracy, privacy budget, and parameter efficiency, and maintains stable performance under diverse and uncertain data conditions. The findings enrich the theoretical integration of differential privacy and parameter-efficient fine-tuning and demonstrate its practical adaptability in instruction tasks, providing a feasible solution for secure training in complex instruction environments.
Read the full abstract

All authors Yulin Huang, Yaxuan Luan, Jinxu Guo, Xiangchen Song, Yuchen Liu

Computation and Language

TopiCLEAR: Adaptive embedding clustering for interpretable topic discovery from short texts

Aoi Fujita, Taichi Yamamoto, Yuri Nakayama et al. (+1 more)

Topic discovery is a fundamental technique for text mining that identifies abstract topics within large document collections. A recent approach to topic discovery is to cluster document or sentence embeddings, typically obtained from pre-trained language models, and represent each cluster as a topic. Despite their strong empirical performance, the design principles linking the geometry of embedding spaces to human-interpretable topic organization remain unclear. Clarifying the relationship between these geometric structures and topic interpretability is therefore a key challenge in topic discovery. In this study, we propose TopiCLEAR (Topic discovery by CLustering Embeddings with Adaptive dimensionality Reduction), a simple framework that integrates document embeddings with iterative clustering based on adaptive dimensionality reduction. TopiCLEAR is guided by the hypothesis that human-interpretable topics correspond to low-dimensional geometric structures in embedding spaces and leverages adaptive dimensionality reduction to identify them. We evaluate topic quality using a document-level approach that combines quantitative evaluation based on human-labeled data with qualitative assessment in the absence of ground-truth labels. Experiments on four benchmark datasets, covering both formal and informal texts, show that TopiCLEAR consistently achieves strong agreement with human annotations, particularly for short and informal texts. Furthermore, a case study on Twitter data demonstrates that TopiCLEAR produces more interpretable topics than Latent Dirichlet Allocation (LDA), recovering both human-annotated topic structure and coherent sub-topic structure. These results highlight the effectiveness of clustering documents in a low-dimensional topic space for topic discovery.
Read the full abstract

All authors Aoi Fujita, Taichi Yamamoto, Yuri Nakayama, Ryota Kobayashi

Computation and Language

Think-While-Generating: On-the-Fly Reasoning for Personalized Long-Form Generation

Chengbing Wang, Yang Zhang, Wenjie Wang et al. (+4 more)

Preference alignment has enabled large language models (LLMs) to better reflect human expectations, but current methods mostly optimize for population-level preferences, overlooking individual users. Personalization is essential, yet early approaches-such as prompt customization or fine-tuning-struggle to reason over implicit preferences, limiting real-world effectiveness. Recent "think-then-generate" methods address this by reasoning before response generation. However, they face challenges in long-form generation: their static one-shot reasoning must capture all relevant information for the full response generation, making learning difficult and limiting adaptability to evolving content. To address this issue, we propose FlyThinker, an efficient "think-while-generating" framework for personalized long-form generation. FlyThinker employs a separate reasoning model that generates latent token-level reasoning in parallel, which is fused into the generation model to dynamically guide response generation. This design enables reasoning and generation to run concurrently, ensuring inference efficiency. In addition, the reasoning model is designed to depend only on previous responses rather than its own prior outputs, which preserves training parallelism across different positions-allowing all reasoning tokens for training data to be produced in a single forward pass like standard LLM training, ensuring training efficiency. Extensive experiments on real-world benchmarks demonstrate that FlyThinker achieves better personalized generation while keeping training and inference efficiency. Our code is available at https://github.com/wcb0219-sketch/FlyThinker.git.
Read the full abstract

All authors Chengbing Wang, Yang Zhang, Wenjie Wang, Xiaoyan Zhao, Fuli Feng, Xiangnan He, Tat-Seng Chua

Computation and Language

PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic Memory

Bowen Jiang, Yuan Yuan, Maohao Shen et al. (+13 more)

Personalization is one of the next milestones in advancing AI capability and alignment. We introduce PersonaMem-v2, the state-of-the-art dataset for LLM personalization that simulates 1,000 realistic user-chatbot interactions on 300+ scenarios, 20,000+ user preferences, and 128k-token context windows, where most user preferences are implicitly revealed to reflect real-world interactions. Using this data, we investigate how reinforcement fine-tuning enables a model to improve its long-context reasoning capabilities for user understanding and personalization. We also develop a framework for training an agentic memory system, which maintains a single, human-readable memory that grows with each user over time. In our experiments, frontier LLMs still struggle with implicit personalization, achieving only 37-48% accuracy. While they support long context windows, reasoning remains the bottleneck for implicit personalization tasks. Using reinforcement fine-tuning, we successfully train Qwen3-4B to outperforms GPT-5, reaching 53% accuracy in implicit personalization. Moreover, our agentic memory framework achieves state-of-the-art 55% accuracy while using 16x fewer input tokens, relying on a 2k-token memory instead of full 32k conversation histories. These results underscore the impact of our dataset and demonstrate agentic memory as a scalable path toward real-world personalized intelligence.
Read the full abstract

All authors Bowen Jiang, Yuan Yuan, Maohao Shen, Zhuoqun Hao, Zhangchen Xu, Zichen Chen, Ziyi Liu, Anvesh Rao Vijjini, Jiashu He, Hanchao Yu, Radha Poovendran, Gregory Wornell, Lyle Ungar, Dan Roth, Sihao Chen, Camillo Jose Taylor

Computation and Language

Mechanistic Interpretability of GPT-2: Lexical and Contextual Layers in Sentiment Analysis

Amartya Hatua

We present a mechanistic interpretability study of GPT-2 that causally examines how sentiment information is processed across its transformer layers. Using systematic activation patching across all 12 layers, we test the hypothesized two-stage sentiment architecture comprising early lexical detection and mid-layer contextual integration. Our experiments confirm that early layers (0-3) act as lexical sentiment detectors, encoding stable, position specific polarity signals that are largely independent of context. However, all three contextual integration hypotheses: Middle Layer Concentration, Phenomenon Specificity, and Distributed Processing are falsified. Instead of mid-layer specialization, we find that contextual phenomena such as negation, sarcasm, domain shifts etc. are integrated primarily in late layers (8-11) through a unified, non-modular mechanism. These experimental findings provide causal evidence that GPT-2's sentiment computation differs from the predicted hierarchical pattern, highlighting the need for further empirical characterization of contextual integration in large language models.
Read the full abstract

Computation and Language

CMV-Fuse: Cross Modal-View Fusion of AMR, Syntax, and Knowledge Representations for Aspect Based Sentiment Analysis

Smitha Muthya Sudheendra, Mani Deep Cherukuri and Jaideep Srivastava

Natural language understanding inherently depends on integrating multiple complementary perspectives spanning from surface syntax to deep semantics and world knowledge. However, current Aspect-Based Sentiment Analysis (ABSA) systems typically exploit isolated linguistic views, thereby overlooking the intricate interplay between structural representations that humans naturally leverage. We propose CMV-Fuse, a Cross-Modal View fusion framework that emulates human language processing by systematically combining multiple linguistic perspectives. Our approach systematically orchestrates four linguistic perspectives: Abstract Meaning Representations, constituency parsing, dependency syntax, and semantic attention, enhanced with external knowledge integration. Through hierarchical gated attention fusion across local syntactic, intermediate semantic, and global knowledge levels, CMV-Fuse captures both fine-grained structural patterns and broad contextual understanding. A novel structure aware multi-view contrastive learning mechanism ensures consistency across complementary representations while maintaining computational efficiency. Extensive experiments demonstrate substantial improvements over strong baselines on standard benchmarks, with analysis revealing how each linguistic view contributes to more robust sentiment analysis.
Read the full abstract