跳到正文
研踪PaperTrail

YOUR RESEARCH, IN SIGHT

让下一篇重要论文，不再错过。

覆盖计算机全学科，以 AI 为快捷入口。关注方向、作者和关键词，把发现变成持续的研究积累。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+1072新增论文
+845其中 AI 相关
+94已有论文修订
+100新增 AI 导读

arXiv 最近完成扫描：2026-11-10 01:54 · 全站最近入库：2026-11-09 13:22各信源进度 →

统计口径与数据新鲜度

统计 2026-11-09 02:47 — 2026-11-10 02:47（北京时间） · 点击数字查看对应简报；AI 相关包含在新增论文内。

最近完成检查：2026-11-10 01:54最近发现新增版本：2026-11-09 13:22库内论文最新更新时间：2026-11-07 02:59

每分钟刷新已保存的数据。空检查不代表采集异常；新收录可能包含历史论文。AI 导读按新增版本计数。

18,172 已收录论文15,506 AI 相关4,527 有修订112 HF 社区发现

浏览论文库

研究日报 / 周报 →按信源浏览 →

时间范围跟随排序：版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常，查看状态 →

找到 20 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-10 01:52（北京时间）
新论文 · v1arXiv:2611.08262

模拟视角的几何结构：在多模态语言模型中分离表征与行为

The Geometry of Simulated Perspectives: Dissociating Representation and Behaviour in a Multimodal Language Model

Mathis Immertreu, Fitim Abdullahu, Thomas Kinfe, Claus Metzner, Helmut Grabner, Patrick Krauss, Achim Schilling

AI 速读 基于摘要

作者报告在 Qwen3-VL-8B-Instruct 中，人格提示的措辞比人格身份更强地改变内部表征，而人格内容对行为评分影响更大，呈现表征与行为的分离。

cs.AI · cs.CL · cs.CV

arXiv ↗PDF ↗
查看原文摘要

Multimodal language models can simulate different observers, yet it remains unclear how assigned perspectives reshape their internal representations and whether these changes govern their judgements. Here, we investigate the representational geometry of persona-conditioned visual interestingness in Qwen3-VL-8B-Instruct. Across 100 images and 53 prompting conditions, we systematically vary persona content and the linguistic framing used to introduce it, while recording behavioural ratings and layer-wise hidden states. We observe a systematic reorganisation of representational geometry across processing depth: early layers primarily encode image identity, whereas persona-specific structures emerge in later layers. Strikingly, prompt framing reshapes internal representations more strongly than persona identity, whereas persona content has a greater impact on behavioural judgements. This dissociation is further illustrated by minimal identity statements that alter representations even without changing responses. These findings demonstrate that representational distance does not necessarily reflect behavioural relevance. More broadly, they reveal how simulated perspectives emerge as distinct structures within the representational space of artificial cognitive systems, highlighting a fundamental distinction between the geometry of internal representations and their functional consequences.

版本更新 · v2arXiv:2602.13274

ProMoral-Bench: Evaluating Prompting Strategies for Moral Reasoning and Safety in LLMs

Rohan Subramanian Thomas, Shikhar Shiromani, Abdullah Chaudhry, Ruizhe Li, Vasu Sharma, Kevin Zhu

原文摘要

Prompt design affects the moral judgments and refusal behavior produced by large language models (LLMs), yet controlled comparisons remain fragmented across datasets and models. We introduce ProMoral-Bench, a benchmark-style comparison of 10 prompting paradigms across four LLM families. Using ETHICS, Scruples, WildJailbreak, and our 200-pair ETHICS-Contrast set, we report task-level metrics together with Moral Competence Score (MCS), jailbreak/over-refusal Safety Robustness Score (SRS), and Unified Moral Safety Score (UMSS). UMSS is a study-specific relative summary of MCS and SRS, not an absolute measure of model safety or moral grounding. In this fixed-template evaluation, compact exemplar-guided scaffolds attain stronger competence-robustness trade-offs at lower token cost than the particular multi-stage variants tested, although multi-stage prompting occasionally improves jailbreak refusal. The few-shot demonstrations are non-overlapping but drawn from the evaluation distributions, so their advantage should not be read as out-of-distribution robustness. Code and data are available at https://github.com/rohanthomas334-sketch/ProMoral_Bench.

cs.AI · cs.CL

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2605.06458

Invariant Features in Language Models: Geometric Characterization and Model Attribution

Agnibh Dasgupta, Abdullah Tanvir, Xin Zhong

原文摘要

Language models exhibit strong robustness to paraphrasing, suggesting that semantic information may be encoded through stable internal representations, yet the structure and origin of such invariance remain unclear. We propose a local geometric framework in which semantically equivalent inputs produce representations that occupy structured regions in latent space, with paraphrastic variation along nuisance directions and semantic identity preserved in invariant subspaces. Building on this view, we make three contributions: (1) a geometric characterization of invariant latent features, (2) a contrastive subspace discovery method that separates semantic-changing from semantic-preserving variation, and (3) an application of invariant representations to zero-shot model attribution. Across models and layers, empirical results support these contributions. Invariant structure emerges in specific depth regions, semantic displacement lies largely outside the nuisance subspace, and representation-level interventions indicate a causal role of invariant components in model outputs. Invariant representations also capture model-specific geometric patterns, enabling accurate attribution. Together, these findings support a local geometric characterization of semantic invariance in language model representations.

cs.LG · cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.03179

Every Clause Counts: Instruction-Following Reranking by Multiplying Small, Answerable Questions

S. M. Mahbub Uddin Ahmed, Abdullah Al Musaddiq

原文摘要

People often attach instructions to their search queries, asking for pages written for beginners, documents in a certain language or official sources only. Most rerankers were trained to judge topical relevance and treat such instructions as a few extra words. On the InfoSearch benchmark, even GPT-4o used as a list-wise reranker reaches an average WISE (Weighted Instruction Sensitivity Evaluation) of only 33.5. We present DMReranker (Decision-Model Reranker), a zero-shot reranker that splits an instructed query into a topic and a short list of typed clauses. A small judge model answers one narrow question per clause, and the answer probabilities are multiplied into a single score. Nothing is trained, and the configuration was frozen before the test runs. Reranking the same top-100 candidates as the published baselines, DMReranker reaches a macro WISE of 50.9 over all six InfoSearch dimensions while keeping original-mode nDCG@10 at 82.9. It trails GPT-4o on the stricter SICR (Strict Instruction Compliance Ratio) metric, 19.2 against 32.1. We show that this gap follows from how a product of probabilities behaves when SICR compares raw scores across runs. Computed from ranks, our SICR is 67.0. On the same judge, splitting the instruction raises WISE from 32.9 to 47.6 over a single combined question, while raw-score SICR falls. Scoring all six dimensions cost about $10 at list price.

cs.IR · cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2611.02208

Test-Time Scaling for Open-Ended Reasoning at Small Model Scale

Adam Y. Ahmed, Abdullah G. Elafifi

原文摘要

A vote over a model's samples can only help when the samples disagree. For open-ended generation the binding question is therefore not how to combine samples but whether a model produces samples worth combining, a question test-time scaling has answered for math but not for free-form text. We study it at small scale with semantic cluster voting, a judge-free aggregator that groups a small instruct model's samples by meaning and returns the largest cluster's representative, with no reward model or self-judging. On a 250-prompt benchmark with three tiers of answer checkability, the same pattern holds across three model families (Qwen2.5-3B, Llama-3.2-3B, Gemma-2-2B) and two generation seeds: voting raises accuracy on the math control, where samples vary but stay checkable (a gain of +10 to +29 points across the three models), yet adds at most +0.30 judge points on open-ended text. On a larger 200-problem MATH-500 run with the primary model, the math gain is +8.5 points (95% CI [+4.0, +13.0]). Almost every open-ended sample falls into one meaning cluster, so there is little to select over. The collapse is consistent with instruction-tuning rather than model size, and raising decoding temperature does not recover usable diversity, because it costs more quality than it adds (net -0.64 judge points, 95% CI [-0.93, -0.37]).

cs.LG · cs.AI · cs.CL

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2606.24387

AutoSpecNER: A Fine-Grained Named Entity Recognition Dataset for Vehicle Specification Extraction

Jordan Lee, Filippos Ventirozos, Abdirahman Abdullahm, Ioanna Nteka, Peter Appleby, Matthew Shardlow

原文摘要

Vehicle advertisements contain rich specification information, but automotive NER resources remain limited. We introduce AutoSpecNER, an expert-annotated dataset for fine-grained entity recognition in vehicle listings. The dataset includes 659 advertisements from a popular car-selling website, with over 10,000 entities annotated across 15 categories, including MODEL, ENGINE_SPEC, and BATTERY_CAPACITY. Annotation quality was validated through inter-annotator agreement, achieving an average score of 91.5%. We benchmark rule-based extraction, fine-tuned transformer encoders, and large language models. DeBERTa achieves the best performance with a 90% micro-F1 score, outperforming the rule-based baseline (43%) and the strongest large language model (77.8%).

cs.CL

查看论文
arXiv ↗PDF ↗
新论文 · v1arXiv:2610.39590

Red-Teaming Large Language Models in Low-Resource African Languages: A Case Study of Yoruba and Nigerian Pidgin

Paula Chikaodinaka Anyanwu, Valdini Douglace Lemofouet, Blessing Ngozi Uzor, Ifeoluwa Dorcas Awotunde, Danielle Blanche Kapsa, Tassallah Abdullahi, Clemencia Siro, Idris Abdulmumin 等 10 位作者

原文摘要

As large language models (LLMs) become more widespread in Africa, their safety in low-resource languages remains largely untested since most red-teaming work focuses on English. This study evaluates six state-of-the-art LLMs using 400 explicit adversarial prompts in Yoruba and Nigerian Pidgin, categorised into four harm categories. The responses were scored by native-speaker annotators and reliability was assessed using Cohen's $κ$. While models that rejected up to 69% of English prompts rejected as few as 0% in Yoruba, this was not because they were safer, but because they failed to understand the prompts. Attack success rates increased for most models in the Nigerian languages, with GPT-4o Mini demonstrating the most significant cross-lingual safety discrepancy (+0.23 in Yoruba). When models did engage with adversarial prompts, many of the responses were harmful, with rates of harmful content reaching 0.63 in English and 0.50 in Nigerian Pidgin. The most difficult harm category to detect was cultural stereotypes and discrimination (ASR = 0.44 in Nigerian Pidgin). Comprehension in Yoruba was low across most models, and this directly suppressed apparent attack success without indicating actual safety. Gemini 2.0 Flash was the most consistent model across languages. Despite its explicit training in African languages, Aya 101 had the highest attack success rate in English (0.60) and one of the lowest Yoruba comprehension rates (0.28). These results demonstrate that English-centric safety alignment does not transfer to Nigerian languages, highlighting the urgent need for language-specific safety evaluations.

cs.CL · cs.AI

查看论文
arXiv ↗PDF ↗
版本更新 · v2arXiv:2610.35168

Do Language Models Understand Igbo Culture? Beyond Multiple Choice

Blessing Ngozi Uzor, Valdini Douglace Lemofouet, Paula Chikaodinaka Anyanwu, Mmesoma Peace Orji, Danielle Blanche Kapsa, Tassallah Abdullahi, Clemencia Siro, Idris Abdulmumin 等 10 位作者

原文摘要

The value of language models for any community depends on whether a model actually understands the culture of the people it intends to serve. We use the Igbo language as a case study. Existing benchmarks test translated Western ideas, lacking depth. To address this, we introduce CAB-Igbo, a benchmark developed from indigenous cultural sources and verified by native speakers across six domains. We evaluated diverse models under varying conditions, incorporating regional personas. Performance was assessed using multiple-choice accuracy and human evaluations across Epistemic Fidelity, Representational Richness, Linguistic Quality, and Pragmatic Proficiency. Results reveal a stark dissociation between language fluency and cultural knowledge. Through statistical testing, we show that for most models, the severe lack of Epistemic Fidelity and Representational Richness is already present in English and does not significantly worsen when switching to Igbo. Instead, native language prompting mainly degrades Linguistic Quality and Pragmatic Proficiency. Gemini 2.0 Flash is the only model whose cultural knowledge scores improve under Igbo prompting. We find that current models struggle to demonstrate the knowledge needed for true alignment with Igbo, and multiple-choice formats allow pattern-matching to appear as understanding. Furthermore, automated judges agree with humans on linguistic quality but fail on epistemic fidelity. Ultimately, cultural evaluation must be built from community knowledge, establishing human evaluation as the gold standard to prevent misrepresenting low-resource communities globally.

cs.CL

查看论文
arXiv ↗PDF ↗
首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-industry.106ACL Anthology · ACL 2026

Are Large Language Models Economically Viable for Industry Deployment?

Abdullah Mohammad, Sushant Kumar Ray, Pushkar Arora, Rafiq Ali, Ebad Shabbir, Gautam Siddharth Kashyap, Jiechao Gao, Usman Naseem

原文摘要

Generative AI—powered by Large Language Models (LLMs)—is increasingly deployed in industry across healthcare decision support, financial analytics, enterprise retrieval, and conversational automation, where reliability, efficiency, and cost control are critical. In such settings, models must satisfy strict constraints on energy, latency, and hardware utilization—not accuracy alone. Yet prevailing evaluation pipelines remain accuracy-centric, creating a Deployment–Evaluation Gap—the absence of operational and economic criteria in model assessment. To address this gap, we present EDGE-EVAL—a industry-oriented benchmarking framework that evaluates LLMs across their full lifecycle on legacy NVIDIA Tesla T4 GPUs. Benchmarking LLaMA and Qwen variants across three industrial tasks, we introduce five deployment metrics—Economic Break-Even (Nbreak), Intelligence-Per-Watt (IP W ), System Density (ρsys), Cold-Start Tax (Ctax), and Quantization Fidelity (Qret)—capturing profitability, energy efficiency, hardware scaling, serverless feasibility, and compression safety. Our results reveal a clear efficiency frontier—models in the < 2B parameter class dominate larger baselines across economic and ecological dimensions. LLaMA-3.2-1B (INT4) achieves ROI break-even in 14 requests (median), delivers 3× higher energy-normalized intelligence than 7B models, and exceeds 6,900 tokens/s/GB under 4-bit quantization. We further uncover an efficiency anomaly—while QLoRA reduces memory footprint, it increases adaptation energy by up to 7× for small models—challenging prevailing assumptions about quantization-aware training in edge deployment.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1504ACL Anthology · ACL 2026

Can AI Be a Good Peer Reviewer? A Survey of Peer Review Process, Evaluation, and the Future

Sihong Wu, Owen Jiang University of Pennsylvania, University of Pennsylvania, Yilun Zhao Yale University, Tiansheng Hu, Yiling Ma King Abdullah University of Science and Technology, Kaiyan Zhang Yale University, Manasi Patwardhan Tata Consultancy Services Limited, India, Arman Cohan Yale University and Allen Institute for Artificial Intelligence

原文摘要

Peer review is a multi-stage process involving reviews, rebuttals, meta-reviews, final decisions, and subsequent manuscript revisions. Recent advances in large language models (LLMs) have motivated methods that assist or automate different stages of this pipeline. In this survey, we synthesize techniques for (i) peer review generation, including fine-tuning strategies, agent-based systems, RL-based methods, and emerging paradigms to enhance generation; (ii) after-review tasks including rebuttals, meta-review and revision aligned to reviews; and (iii) evaluation methods spanning human-centered, reference-based, LLM-based and aspect-oriented. We catalog datasets, compare modeling choices, and discuss limitations, ethical concerns, and future directions. The survey aims to provide practical guidance for building, evaluating, and integrating LLM systems across the full peer review workflow.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.159ACL Anthology · ACL 2026

Make Mechanistic Interpretability Auditable: A Call to Develop Guidelines via Continuous Collaborative Reviewing

Michael Lan Martian, Narmeen Fatimah Oozeer, Chaithanya Bandi Northwestern University, Philip Quirke, Austin Meek University of Delaware, Fazl Barez University of Oxford, Amir Abdullah Thoughtworks

原文摘要

While mechanistic interpretability (MI) has produced important insights into neural network internals, the field has yet to establish a standardized system to audit experiments. As such, many of its findings remain underutilized in safety-critical applications such as medical AI and autonomous systems, as stakeholders cannot certify their validity. Recent work demonstrates this concretely: two papers found conflicting conclusions for the same behavior, and a third study revealed that both were partially correct but incomparable due to methodological inconsistencies. Without standardized auditing, such ambiguities hinder adoption in high-stakes contexts requiring strong correctness guarantees. We call for the MI community to work towards developing a novel reviewing system that complements peer review via: (1) Continuous reviewing supported by a Collaborative Reviewing Platform where meta-science results and discussions (such as critiques, negative results, post-hoc extensions, reproductions, replications, and partial results) that fit outside of papers are organized and discussed, allowing for comments and revisions to be made at any time (2) Generalizing good practices found on this platform into expert-verified guidelines and protocols to improve auditing efficiency, and (3) Source-based auditing systems that track arguments which claims depend on. This position paper encourages constructive debate over the necessity, design and implementation of such a framework, providing early concrete examples to help catalyze these dialogues. Overall, we propose that auditing MI itself is essential for its application in AI safety, industry, and governance.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.1915ACL Anthology · ACL 2026

Multimodal Safety Evaluation in Generative Agent Social Simulations

Alhim Adonai Vera Gonzalez, Carlos Hinojosa King Abdullah University of Science and Technology, Karen Sanchez King Abdullah University of Science and Technology, Haidar Bin Hamid University of Cincinnati, Donghoon Kim University of Cincinnati, Bernard Ghanem King Abdullah University of Science and Technology

原文摘要

Can generative agents be trusted in multimodal environments? Despite recent advances, agents remain limited in their ability to reason about safety, coherence, and trust across modalities. We introduce a reproducible simulation framework to evaluate generative agents in three aspects: (1) safety improvement over time via iterative plan revision in multimodal scenarios; (2) detection of unsafe activities across social contexts; and (3) social dynamics, measured through interaction and acceptance rates. These multimodal agents are evaluated using metrics that quantify plan revisions and unsafe-to-safe conversions. Experiments show that while agents detect direct multimodal contradictions, they often fail to align local revisions with global safety, achieving only a 55% success rate in correcting unsafe plans. We release a dataset of 1,000 multimodal plans, yielding more than 600,000 simulation steps. Notably, 45% of unsafe actions are accepted when paired with misleading visual cues, revealing a strong tendency to overtrust visual content. Code is available at https://github.com/AdonaiVera/X-CASE

cs.CL · cs.CV

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.2136ACL Anthology · ACL 2026

LogicEval: A Systematic Framework for Evaluating Automated Repair Techniques for Logical Vulnerabilities in Real-World Software

Syed Md Mukit Rashid Pennsylvania State University, Abdullah Al Ishtiaq Pennsylvania State University, Kai Tu, Yilu Dong Pennsylvania State University, Tianwei Wu, Ali Ranjbar, Tianchang Yang Pennsylvania State University, Najrin Sultana Pennsylvania State University 等 10 位作者

原文摘要

Logical vulnerabilities in software stem from flaws in program logic rather than memory safety, which can lead to critical security failures. Although existing automated program-repair techniques primarily focus on repairing memory-corruption vulnerabilities, they struggle with logical vulnerabilities because of their limited semantic understanding of the vulnerable code and its expected behavior. On the other hand, recent successes of large language models (LLMs) in understanding and repairing code are promising. However, no framework currently exists to analyze the capabilities and limitations of such techniques for logical vulnerabilities. We aim to systematically evaluate both traditional and LLM-based repair approaches for addressing real-world logical vulnerabilities. To facilitate our assessment, we created the first-ever dataset, LogicDS, comprising 122 logical vulnerabilities that reflect tangible security impact. We also developed a systematic framework, LogicEval, to evaluate patches for logical vulnerabilities. Evaluations suggest that compilation and testing failures are primarily driven by prompt sensitivity, loss of code context, and difficulty in patch localization.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.562ACL Anthology · ACL 2026

PDR: A Plug-and-Play Positional Decay Framework for LLM Pre-training Data Detection

Jinhan Liu, Yibo Yang Shanghai Jiao Tong University, Ruiying Lu, Piotr Piękos King Abdullah University of Science and Technology, Yimeng Chen King Abdullah University of Science and Technology, Peng Wang, Dandan Guo

原文摘要

Detecting pre-training data in Large Language Models (LLMs) is crucial for auditing data privacy and copyright compliance, yet it remains challenging in black-box, zero-shot settings where computational resources and training data are scarce. While existing likelihood-based methods have shown promise, they typically aggregate token-level scores using uniform weights, thereby neglecting the inherent information-theoretic dynamics of autoregressive generation. In this paper, we hypothesize and empirically validate that memorization signals are heavily skewed towards the high-entropy initial tokens, where model uncertainty is highest, and decay as context accumulates. To leverage this linguistic property, we introduce Positional Decay Reweighting (PDR), a training-free and plug-and-play framework. PDR explicitly reweights token-level scores to amplify distinct signals from early positions while suppressing noise from later ones. Extensive experiments show that PDR acts as a robust prior and can usually enhance a wide range of advanced methods across multiple benchmarks.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.731ACL Anthology · ACL 2026

Measuring What Matters!! Assessing Therapeutic Principles in Mental-Health Conversation

Abdullah Mazhar, Het Riteshkumar Shah, Aseem Srivastava Mohamed bin Zayed University of Artificial Intelligence, Smriti Joshi, Md Shad Akhtar Indraprastha Institute of Information Technology, Delhi

原文摘要

The increasing use of large language models in mental health applications calls for principled evaluation frameworks that assess alignment with psychotherapeutic best practices beyond surface-level fluency. While recent systems exhibit conversational competence, they lack structured mechanisms to evaluate adherence to core therapeutic principles. In this paper, we study the problem of evaluating AI-generated therapist-like responses for clinically grounded appropriateness and effectiveness. We assess each therapists utterance along six therapeutic principles: non-judgmental acceptance, warmth, respect for autonomy, active listening, reflective understanding, and situational appropriateness using a fine-grained ordinal scale. We introduce FAITH-M, a benchmark annotated with expert-assigned ordinal ratings, and propose CARE, a multi-stage evaluation framework that integrates intra-dialogue context, contrastive exemplar retrieval, and knowledge-distilled chain-of-thought reasoning. Experiments show that CARE achieves an F-1 score of 63.34 versus the strong baseline Qwen3 F-1 score of 38.56 which is a 64.26% improvement, which also serves as its backbone, indicating that gains arise from structured reasoning and contextual modeling rather than backbone capacity alone. Expert assessment and external dataset evaluations further demonstrate robustness under domain shift, while highlighting challenges in modeling implicit clinical nuance. Overall, CARE provides a clinically grounded framework for evaluating therapeutic fidelity in AI mental health systems.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-long.733ACL Anthology · ACL 2026

CoLA: A Choice Leakage Attack Framework to Expose Privacy Risks in Subset Training

Qi Li, Cheng-Long Wang King Abdullah University of Science and Technology, Yinzhi Cao Johns Hopkins University, Di Wang KAUST

原文摘要

Training models on a carefully chosen portion of data rather than the full dataset is now a standard preprocess for modern ML. From vision coreset selection to large-scale filtering in language models, it enables scalability with minimal utility loss. A common intuition is that training on fewer samples should also reduce privacy risks. In this paper, we challenge this assumption. We show that subset training is not privacy free: the very choices of which data are included or excluded can introduce new privacy surface and leak more sensitive information. Such information can be captured by adversaries either through side-channel metadata from the subset selection process or via the outputs of the target model. To systematically study this phenomenon, we propose CoLA (Choice Leakage Attack), a unified framework for analyzing privacy leakage in subset selection. In CoLA, depending on the adversary’s knowledge of the side-channel information, we define two practical attack scenarios: Subset-aware Side-channel Attacks and Black-box Attacks. Under both scenarios, we investigate two privacy surfaces unique to subset training: (1) Training-membership MIA (TM-MIA), which concerns only the privacy of training data membership, and (2) Selection-participation MIA (SP-MIA), which concerns the privacy of all samples that participated in the subset selection process. Notably, SP-MIA enlarges the notion of membership from model training to the entire data-model supply chain. Experiments on vision and language models show that existing threat models underestimate subset-training privacy risks: the expanded privacy surface leaks both training and selection membership, extending risks from individual models to the broader ML ecosystem.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-srw.34ACL Anthology · ACL 2026

BanglaSTEM: A Parallel Corpus and Term-Weighted Evaluation for Technical Bangla-English Translation

Kazi Reyazul Hasan Bangladesh University of Engineering and Technology, A. B. M. Alim Al Islam Bangladesh University of Engineering and Technology, Muhammad Abdullah Adnan Bangladesh University of Engineering and Technology

原文摘要

Large language models excel at technical problem solving in English but struggle when questions are posed in Bangla. While translation offers a practical solution, existing Bangla-English systems frequently mistranslate specialized terminology, altering problem semantics and degrading downstream performance. We present BanglaSTEM, a dataset of 5,000 Bangla-English sentence pairs covering computer science, mathematics, physics, chemistry, and biology. Our pipeline extracts matching passages from official bilingual curriculum textbooks using OCR, then uses LLMs to align sentences and mark technical terms. These aligned examples serve as few-shot prompts for generating over 12,000 new translation pairs from LLMs, avoiding copyright issues. Human evaluators then select the best 5,000 pairs that correctly preserve technical terminology. We also test a term-weighted BLEU metric that gives higher weight to technical words, since standard metrics treat terminology errors and common word errors equally. We show that our weighted metric correlates better with downstream accuracy in code generation and math solving, while standard BLEU gives high scores even for wrong translations. The full implementation, dataset, and model will be made publicly available.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-srw.37ACL Anthology · ACL 2026

FedPAGR: Federated Prototype Alignment via Geometric Refinement for Heterogeneous Architectures

Kris Prasad, Md Abdullah Al Hafiz Khan Kennesaw State University

原文摘要

Federated learning with heterogeneous client architectures cannot rely on parameter aggregation. Prototype-based methods address architectural heterogeneity by exchanging class-level representations, but naively averaging prototypes across non-IID clients leads to semantic drift and poor inter-class separation. We propose FedPAGR, a framework where heterogeneous clients project their features into a shared consensus space and exchange class prototypes with a central server. The server refines aggregated prototypes through a geometric regularization objective that enforces agreement with client submissions and inter-class angular separation. Clients anchor their classifiers to the refined prototypes and train with a composite objective combining classification, prototype alignment, and entropy regularization. We evaluate FedPAGR across multiple domains, including four image benchmarks and a clinical NLP task using heterogeneous ClinicalBERT variants, with five architectures per federation under severe label heterogeneity (\alpha{=}0.1). FedPAGR achieves the highest ensemble accuracy across all four image datasets and the highest local test accuracy on low-class and clinical tasks, including a 4.99-point improvement over the strongest baseline on MIMIC-IV, while remaining competitive on high-class benchmarks.

cs.CL

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop) · 2026 · 正式发表

首次观测 · 记录 #1ACL Anthology · ACL 2026:acl:2026.acl-srw.81ACL Anthology · ACL 2026

Tonal Salience in Cognitive Decline: In-Context MCI Detection with Multimodal LLMs

Christopher Song, Abdullah P. Rashed Ahmed Brown University

原文摘要

Speech-based screening for mild cognitive impairment offers a highly accessible way to detect early cognitive decline. While most existing work focuses on English, cross-linguistic research is emerging to examine how cognitive decline manifests across languages. Studies on the Interspeech 2024 TAUKADIAL dataset, comprising English and Chinese speech recordings, consistently report higher classification performance on Chinese, yet the cause of this cross-lingual discrepancy remains unexplored. We examine this gap using Gemini 2.5 Pro, a multimodal large language model, using zero-shot and in-context-learning (ICL) paradigms. We hypothesize that this disparity is rooted in language typology: in tonal languages like Chinese, pitch encodes lexical meaning in every syllable (tone), whereas in non-tonal languages like English, pitch carries no lexical function. To test this, we pitch-flattened audio from TAUKADIAL and compared how classification performance changed across both languages. We found that Chinese classification degraded significantly under both zero-shot and ICL conditions (-4.78 and -5.92 UAR, respectively), while English performance increased (+0.11 and +2.98 UAR), implicating tonal pitch as the cross-lingual advantage. These findings suggest language typology should inform the design of audio-based cognitive screening tools, with raw audio preferred for tonal languages and text for non-tonal languages, a distinction critical for developing equitable cross-linguistic screening.

cs.CL · cs.CV

Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop) · 2026 · 正式发表

版本更新 · v3arXiv:2602.02734Google Research

WAXAL: A Large-Scale Multilingual African Language Speech Corpus

Abdoulaye Diack, Perry Nelson, Kwaku Agbesi, Angela Nakalembe, MohamedElfatih MohamedKhair, Vusumuzi Dube, Tavonga Siyavora, Subhashini Venugopalan 等 43 位作者

原文摘要

The advancement of speech technology has predominantly favored high-resource languages, creating a significant digital divide for speakers of most Sub-Saharan African languages. To address this gap, we introduce WAXAL, a large-scale, openly accessible speech dataset for 24 languages representing over 100 million speakers. The collection consists of two main components: an Automated Speech Recognition (ASR) dataset containing approximately 1,250 hours of transcribed, natural speech from a diverse range of speakers, and a Text-to-Speech (TTS) dataset with around 235 hours of high-quality, single-speaker recordings reading phonetically balanced scripts. This paper details our methodology for data collection, annotation, and quality control, which involved partnerships with four African academic and community organizations. We provide a detailed statistical overview of the dataset and discuss its potential limitations and ethical considerations. The WAXAL datasets are released at https://huggingface.co/datasets/google/WaxalNLP under the permissive CC-BY-4.0 license to catalyze research, enable the development of inclusive technologies, and serve as a vital resource for the digital preservation of these languages.

eess.AS · cs.AI · cs.CL

查看论文
arXiv ↗PDF ↗