paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

arXiv

Detector-Augmented SAMURAI for Long-Duration Drone Tracking

Tamara R. Lenhard, Andreas Weinmann, Hichem Snoussi, Tobias Koch

Robust long-term tracking of drone is a critical requirement for modern surveillance systems, given their increasing threat potential. While detector-based approaches typically achieve strong frame-level accuracy, they often suffer from temporal inconsistencies caused by frequent detection dropouts. Despite its practical relevance, research on RGB-based drone tracking is still limited and largely reliant on conventional motion models. Meanwhile, foundation models like SAMURAI have established their effectiveness across other domains, exhibiting strong category-agnostic tracking performance. However, their applicability in drone-specific scenarios has not been investigated yet. Motivated by this gap, we present the first systematic evaluation of SAMURAI's potential for robust drone tracking in urban surveillance settings. Furthermore, we introduce a detector-augmented extension of SAMURAI to mitigate sensitivity to bounding-box initialization and sequence length. Our findings demonstrate that the proposed extension significantly improves robustness in complex urban environments, with pronounced benefits in long-duration sequences - especially under drone exit-re-entry events. The incorporation of detector cues yields consistent gains over SAMURAI's zero-shot performance across datasets and metrics, with success rate improvements of up to +0.393 and FNR reductions of up to -0.475.

arXiv

HTTM: Head-wise Temporal Token Merging for Faster VGGT

Weitian Wang, Lukas Meiner, Rai Shubham, Cecilia De La Parra, Akash Kumar

The Visual Geometry Grounded Transformer (VGGT) marks a significant leap forward in 3D scene reconstruction, as it is the first model that directly infers all key 3D attributes (camera poses, depths, and dense geometry) jointly in one pass. However, this joint inference mechanism requires global attention layers that perform all-to-all attention computation on tokens from all views. For reconstruction of large scenes with long-sequence inputs, this causes a significant latency bottleneck. In this paper, we propose head-wise temporal merging (HTTM), a training-free 3D token merging method for accelerating VGGT. Existing merging techniques merge tokens uniformly across different attention heads, resulting in identical tokens in the layers' output, which hinders the model's representational ability. HTTM tackles this problem by merging tokens in multi-head granularity, which preserves the uniqueness of feature tokens after head concatenation. Additionally, this enables HTTM to leverage the spatial locality and temporal correspondence observed at the head level to achieve higher merging ratios with lower merging costs compared to existing methods. Thus, HTTM achieves up to $7\times$ acceleration over the original VGGT with negligible performance drops in a GPU-based inference.

arXiv

PickStyle: Video-to-Video Style Transfer with Context-Style Adapters

Soroush Mehraban, Vida Adeli, Jacob Rommann, Kyryl Truskovskyi, Harrison Sanborn, Babak Taati, Cole Clifford

We address the task of video style transfer with diffusion models, where the goal is to preserve the context of an input video while rendering it in a target style specified by a text prompt. A major challenge is the lack of paired video data for supervision. We propose PickStyle, a video-to-video style transfer framework that augments pretrained video diffusion backbones with style adapters and benefits from paired still image data with source-style correspondences for training. PickStyle inserts low-rank adapters into the self-attention layers of conditioning modules, enabling efficient specialization for motion-style transfer while maintaining strong alignment between video content and style. To bridge the gap between static image supervision and dynamic video, we construct synthetic training clips from paired images by applying shared augmentations that simulate camera motion, ensuring temporal priors are preserved. In addition, we introduce Context-Style Classifier-Free Guidance (CS-CFG), a novel factorization of classifier-free guidance into independent text (style) and video (context) directions. CS-CFG ensures that context is preserved in generated video while the style is effectively transferred. Experiments across benchmarks show that our approach achieves temporally coherent, style-faithful, and content-preserving video translations, outperforming existing baselines both qualitatively and quantitatively.

arXiv

CSMoE: An Efficient Remote Sensing Foundation Model with Soft Mixture-of-Experts

Leonard Hackel, Tom Burgert, Begüm Demir

Self-supervised learning (SSL) through masked autoencoders (MAEs) has recently attracted great attention for remote sensing (RS) foundation model (FM) development, enabling improved representation learning across diverse sensors and downstream tasks. However, existing RS FMs often either suffer from substantial computational complexity during both training and inference or exhibit limited representational capacity. In addition, their pretraining datasets can contain redundant images, which increases training cost without improving the learned representations. These issues restrict their practical applicability in RS. To address these limitations, we improve the computational efficiency of RS FMs along two axes: i) model efficiency; and ii) training-data efficiency. Model efficiency is achieved by integrating the Soft mixture-of-experts (MoE) mechanism into the FM, which allows modality-specific expert processing alongside shared cross-sensor representation learning while reducing computational complexity at both training and inference time. We apply this adaptation to the Cross-Sensor Masked Autoencoder (CSMAE) model, which serves as our main baseline, resulting in the Cross-Sensor Mixture-of-Experts (CSMoE) model. Training-data efficiency is achieved by a thematic-climatic descriptor-driven sampling strategy, which constructs a reduced training set from a large-scale image archive while retaining its geographic and thematic-climatic diversity, and thus reduces pretraining cost. Extensive experiments on scene classification, semantic segmentation, and content-based image retrieval (CBIR) show that CSMoE remains competitive with state-of-the-art RS FMs while requiring substantially fewer floating-point operations (FLOPs). The associated code for the model and the training set creation, as well as the pretrained model weights, will be available at https://git.tu-berlin.de/rsim/csmoe.

arXiv

Geometrically Constrained and Token-Based Probabilistic Spatial Transformers

Johann Schmidt, Sebastian Stober

Spatial transformations such as rotation and scale obscure the morphological cues needed for accurate image classification. Careful consideration is required for reliable use in high stakes settings. A model should stay robust under such transformations, expose why a correction was applied, and signal when its input is ambiguous. While geometrically equivariant architectures provide a mathematically grounded solution, they often limit model flexibility through strict symmetry constraints and incur significant computational overhead. Spatial Transformer Networks (STNs) offer a data-driven, flexible alternative for learning pseudo-equivariances to affine transformations. However, STNs have historically been restricted to convolutional architectures and suffer from training instability. To address this, we introduce a novel STN framework. It leverages the global modeling capabilities of transformers to regress the affine transformation acting on the input. For this, we decompose affine transformations into interpretable primitives, regressed under adaptable geometric constraints, thereby preventing the training instability typically caused by degenerate transformations. By sharing weights between the localization network and the classification backbone, the framework requires minimal computational overhead. Extensive experiments on challenging insect biodiversity and medical imaging benchmarks demonstrate that our approach achieves superior predictive performance under diverse spatial transformations while maintaining high efficiency. Code is available at https://github.com/johSchm/TokenSTN.

arXiv

Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning

Yuyao Ge, Shenghua Liu, Yiwei Wang, Lingrui Mei, Baolong Bi, Xuanshan Zhou, Jiayu Yao, Jiafeng Guo, Xueqi Cheng

Vision-Language Models (VLMs) have demonstrated remarkable success across diverse visual tasks, yet their performance degrades in complex visual environments. While existing enhancement approaches require additional training, rely on external segmentation tools, or operate at coarse-grained levels, they overlook the innate ability within VLMs. To bridge this gap, we investigate VLMs' attention patterns and discover that: (1) visual complexity strongly correlates with attention entropy, negatively impacting reasoning performance; (2) attention progressively refines from global scanning in shallow layers to focused convergence in deeper layers, with convergence degree determined by visual complexity. (3) Theoretically, we prove that the contrast of attention maps between general queries and task-specific queries enables the decomposition of visual signal into semantic signals and visual noise components. Building on these insights, we propose Contrastive Attention Refinement for Visual Enhancement (CARVE), a training-free method that extracts task-relevant visual signals through attention contrasting at the pixel level. Extensive experiments demonstrate that CARVE consistently enhances performance, achieving up to 75% improvement on open-source models. Our work provides critical insights into the interplay between visual complexity and attention mechanisms, offering an efficient pathway for improving visual reasoning with contrasting attention.

arXiv

RingMo-Agent: A Unified Remote Sensing Foundation Model for Multi-Platform and Multi-Modal Reasoning

Huiyang Hu, Peijin Wang, Yingchao Feng, Kaiwen Wei, Wenxin Yin, Wenhui Diao, Mengyu Wang, Hanbo Bi, Kaiyue Kang, Tong Ling, Kun Fu, Xian Sun

Remote sensing (RS) images from multiple modalities and platforms exhibit diverse details due to differences in sensor characteristics and imaging perspectives. Existing vision-language research in RS largely relies on relatively homogeneous data sources. Moreover, they still remain limited to conventional visual perception tasks such as classification or captioning. As a result, these methods fail to serve as a unified and standalone framework capable of effectively handling RS imagery from diverse sources in real-world applications. To address these issues, we propose RingMo-Agent, a model designed to handle multi-modal and multi-platform data that performs perception and reasoning tasks based on user textual instructions. Compared with existing models, RingMo-Agent 1) is supported by a large-scale vision-language dataset named RS-VL3M, comprising over 3 million image-text pairs, spanning optical, SAR, and infrared (IR) modalities collected from both satellite and UAV platforms, covering perception and challenging reasoning tasks; 2) learns modality adaptive representations by incorporating separated embedding layers to construct isolated features for heterogeneous modalities and reduce cross-modal interference; 3) unifies task modeling by introducing task-specific tokens and employing a token-based high-dimensional hidden state decoding mechanism designed for long-horizon spatial tasks. Extensive experiments on various RS vision-language tasks demonstrate that RingMo-Agent not only proves effective in both visual understanding and sophisticated analytical tasks, but also exhibits strong generalizability across different platforms and sensing modalities.

arXiv

Towards Conditional Feature Alignment for Cross-Domain Counting

Zhuonan Liang, Dongnan Liu, Jianan Fan, Yaxuan Song, Qiang Qu, Runnan Chen, Yu Yao, Peng Fu, Weidong Cai

Object counting models often degrade under cross-domain deployment because density composition varies across domains and is itself task-relevant. Standard feature alignment methods tend to suppress such variation by encouraging global domain invariance, which can be harmful when source and target domains contain different proportions of background, sparse foreground, and dense foreground. We propose Conditional Feature Alignment (CFA), a cross-domain counting framework that aligns representations within label-induced conditions rather than across full marginal feature distributions. Given density annotations or pseudo-density predictions, CFA constructs foreground/background or density-level conditions and aligns only features belonging to matching conditions. We formalise this idea through a conditional divergence perspective, characterising an ideal conditionally aligned state with no within-condition discrepancy while preserving condition-marginal density shift. For unsupervised domain adaptation, CFA estimates source conditions from annotations and target conditions from detached pseudo-density maps, then performs condition-wise adversarial alignment with full-image consistency regularisation. For source-domain generalisation, we instantiate the same principle with MPCount by enforcing condition-wise memory-consistency between generated source-domain views. Experiments on crowd and cell counting benchmarks show competitive or improved performance across diverse UDA and DG settings. For example, on JHU-CROWD++ FH->SN, CFA-DG reduces MAE/RMSE from MPCount's 216.3/421.4 to 90.5/169.9, showing a marked improvement on this large weather- and density-induced shift. These results suggest that condition-wise alignment is a promising design principle for domain-adaptive counting.

arXiv

MME-VideoOCR: Evaluating OCR-Based Capabilities of Multimodal LLMs in Video Scenarios

Yang Shi, Huanqian Wang, Wulin Xie, Huanyao Zhang, Lijie Zhao, Yi-Fan Zhang, Xinfeng Li, Chaoyou Fu, Zhuoer Wen, Wenting Liu, Zhuoran Zhang, Xinlong Chen, Bohan Zeng, Sihan Yang, Yushuo Guan, Zhang Zhang, Liang Wang, Haoxuan Li, Zhouchen Lin, Yuanxing Zhang, Pengfei Wan, Haotian Wang, Wenjing Yang

Multimodal Large Language Models (MLLMs) have achieved considerable accuracy in Optical Character Recognition (OCR) from static images. However, their efficacy in video OCR is significantly diminished due to factors such as motion blur, temporal variations, and visual effects inherent in video content. To provide clearer guidance for training practical MLLMs, we introduce the MME-VideoOCR benchmark, which encompasses a comprehensive range of video OCR application scenarios. MME-VideoOCR features 10 task categories comprising 25 individual tasks and spans 44 diverse scenarios. These tasks extend beyond text recognition to incorporate deeper comprehension and reasoning of textual content within videos. The benchmark consists of 1,464 videos with varying resolutions, aspect ratios, and durations, along with 2,000 meticulously curated, manually annotated question-answer pairs. We evaluate 18 state-of-the-art MLLMs on MME-VideoOCR, revealing that even the best-performing model (Gemini-2.5 Pro) achieves an accuracy of only 73.7%. Fine-grained analysis indicates that while existing MLLMs demonstrate strong performance on tasks where relevant texts are contained within a single or few frames, they exhibit limited capability in effectively handling tasks that demand holistic video comprehension. These limitations are especially evident in scenarios that require spatio-temporal reasoning, cross-frame information integration, or resistance to language prior bias. Our findings also highlight the importance of high-resolution visual input and sufficient temporal coverage for reliable OCR in dynamic video scenarios.

arXiv

STAG-VIO: Stabilized Prompt-to-Geometry Interface for Robust Dynamic Visual--Inertial Odometry

Rui Zhou, Jingbin Liu, Junbin Xie, Jianyu Zhang, Yingze Hu, Jiele Zhao

Dynamic visual-inertial odometry (VIO) requires reliable suppression of motion-corrupted measurements, yet prior semantic-assisted approaches depend on category-limited segmenters and degrade under partial occlusion. Promptable foundation segmentation models offer category-agnostic dynamic parsing, but their effectiveness in VIO depends critically on the temporal stability of input prompts---a factor largely overlooked in existing pipelines. When prompts derived from raw detection are jittery or intermittent under occlusion, the resulting masks flicker across frames, destabilizing geometric estimation. We propose STAG-VIO, which formulates dynamic robustness as a perception-to-geometry interface stabilization problem. We introduce uncertainty-adaptive multi-object tracking that models prompt generation as state estimation with bounded noise adaptation, producing temporally coherent box prompts. These stabilized prompts drive a lightweight foundation segmenter whose masks undergo geometry-oriented morphological refinement to establish conservative safety margins. A constraint-budget-aware feature redistribution strategy preserves well-conditioned static measurements when dynamic regions dominate the view. Experiments on VIODE and OpenLORIS-Scene show consistent gains over state-of-the-art baselines. Ablation confirms that prompt stabilization is the single most impactful component, reducing trajectory error by up to 83%.

arXiv

TIMA: Text-Image Mutual Awareness for Balancing Zero-Shot Adversarial Robustness and Generalization Ability

Fengji Ma, Hei Victor Cheng, Chenxing Li, Li Liu

Achieving zero-shot adversarial robustness without sacrificing generalization remains challenging for foundation models such as CLIP, especially under large adversarial perturbations. Through empirical analyses, we identify three critical yet overlooked issues: (1) Logit margins exhibit a stable offset between small and large adversarial perturbations, suggesting that explicitly adjusting margins could improve robustness against unseen large perturbations. (2) A significant negative correlation exists between logit margin and inter-class semantic similarity, indicating that semantic structures are insufficiently leveraged by existing methods. (3) Existing methods for adjusting text embeddings disrupt the intrinsic semantic consistency established by pre-trained models, undermining generalization capability. Motivated by these findings, we propose a novel Text-Image Mutual Awareness (TIMA) framework, including a Text-Aware Image (TAI) tuning module with an Adaptive Semantic-Aware Margin (ASAM) to explicitly calibrate logit margins, and an Image-Aware Text (IAT) tuning module with Semantic Consistent Minimum Hyperspherical Energy (SC-MHE) to preserve semantic consistency. Comprehensive experiments validate that TIMA significantly outperforms existing approaches by effectively addressing the identified limitations.

arXiv

DirMixE: Harnessing Test Agnostic Long-tail Recognition with Hierarchical Label Variations

Zhiyong Yang, Qianqian Xu, Sicong Li, Zitai Wang, Xiaochun Cao, Qingming Huang

This paper explores test-agnostic long-tail recognition, a challenging long-tail task where the test label distributions are unknown and arbitrarily imbalanced. We argue that the variation in these distributions can be broken down hierarchically into global and local levels. The global ones reflect a broad range of diversity, while the local ones typically arise from milder changes, often focused on a particular neighbor. Traditional methods predominantly use a Mixture-of-Expert (MoE) approach, targeting a few fixed test label distributions that exhibit substantial global variations. However, the local variations are left unconsidered. To address this issue, we propose a new MoE strategy, DirMixE, which assigns experts to different Dirichlet meta-distributions of the label distribution, each targeting a specific aspect of local variations. Additionally, the diversity among these Dirichlet meta-distributions inherently captures global variations. This dual-level approach also leads to a more stable objective function, allowing us to sample different test distributions better to quantify the mean and variance of performance outcomes. Building on this idea, we develop a general Latent Skill Finetuning (LSF) framework for parameter-efficient finetuning of foundation models. We provide implementations based on LoRA and Adapter. Theoretically, we derive upper bounds on the generalization error for both standard learning and PEFT. Under mild assumptions, we show that the variance-based regularization helps tighten these bounds. Furthermore, we prove that the covering number of the PEFT hypothesis class scales with the number of trainable parameters. Finally, extensive experiments on CIFAR-10-LT, CIFAR-100-LT, ImageNet-LT, and iNaturalist validate the effectiveness of DirMixE.

arXiv

Evaluating Multimodal Narrative Understanding of Popular Hollywood Films

David Bamman, Kent K. Chang, Allison Cooper, Juishan Hsu, Reina Kushihashi, Madison Mar, Arnav Podichetty, Rachael Samberg, Ipek Nil Sancak, Yuhan Shao

Multimodal language models increasingly show promise for enabling the large-scale computational analysis of film, opening up new avenues for learning about film history and the evolution of narrative techniques. But the creation of stable benchmarks built around Hollywood films is complicated by copyright protections. In this work, we address these concerns directly, by building a new collection of Hollywood films defined by two criteria: box office popularity (where we publish the first large-scale, open collection of weekly box office earnings reported by Variety magazine from 1922-1979); and likely public domain status (by researching copyright registrations and renewals in the US Catalog of Copyright Entries). We build a new multimodal MCQ benchmark on top of this collection that focuses on narrative elements that directly evaluate the abilities of models to inform meaningful research on film narrative; we find that many vision-language models struggle on this task (with many performing at near-chance levels of accuracy), while audio-visual models (including those that use audio in captioning scenes) reach a maximum accuracy of 61.1%, well below human-level performance.

arXiv

Which Source Wins? Task-Dependent Reliance in Vision-Language Models

Rodela Ghosh, Aviral Gupta, Guangjing Wang

Vision-language models (VLMs) combine images and text, but when the two conflict and one becomes harder to read, it is unclear how a model shifts its reliance between them. We study this modality reallocation with a controlled setup: we degrade either the image or the text across four levels of legibility while keeping the other clean, and track how the model's preference changes. We build conflicts from GSM8K and SVAMP by pairing the rendered image of one arithmetic problem with the text of another, so the two sources support different answers. We also introduce ChartQA-Conflict, a manually reviewed benchmark of 229 chart-report conflicts with matched chart and table-image representations. We evaluate six open-weight VLMs using both generated answers and a length-normalized conditional log-likelihood margin. On GSM8K and SVAMP, five of six models shift more strongly away from degraded text than from degraded images. On ChartQA-Conflict, all six likelihood-scored models exhibit the opposite pattern, shifting more strongly away from the degraded visual source. This reversal persists after calibrating for unimodal accuracy loss and after replacing charts with plain table images. Two frontier API models, GPT-5.6-Luna and Gemini-3.5-Flash, behaviorally replicate the ChartQA-Conflict reversal, with GPT-5.6-Luna also matching the arithmetic direction. These results show that modality reliance in VLMs is not fixed, but varies across tasks, evidence structures, models, and evaluation settings. The source code is available at https://github.com/Ro-netizen004/multimodal-arbitration-artifact.

arXiv

Token Optimization and Context Window Management in Multi-Agent AI Workflows

Dvir Shamay

Multi-agent AI workflows are limited not only by model quality but by token cost, latency, and context-window quality. This paper presents a practitioner framework for token optimization and context-window management, grounded in an internal production dashboard that extracts structured work items from meetings, email, and chat with LLMs and routes summaries across workstreams. Six patterns are described: context stratification, fetch-once/process-locally architecture, schema-contracted prompts, token-aware fallback chains, semantic caching, and inter-agent communication compression. In production they cut measured cold-load latency to 61-116 seconds (six timed runs) from an operational baseline of roughly 3.5-10.5 minutes, with an estimated 60-70% token reduction. It also reports a controlled context-composition study: 2,420 confirmatory trials across 11 model configurations, using 661 anonymized workplace items scored for relevance. Holding the prompt at a fixed ten items, replacing some high-relevance items with same-domain low-relevance items improves the model's relevance-score concordance on the target items, versus high-relevance items only; we call this relevance-contrast context. In the all-11 paired analysis, the 50:50 signal/noise condition improved relevance accuracy by +0.077 over the 100% condition (naive 95% CI [+0.056, +0.098], Cohen's d = 0.49, Holm-adjusted p < .001, n = 220). These cells are not independent; by the nine model families the effect is +0.084 (95% interval [+0.064, +0.103]), reported as a within-corpus descriptive comparison, not a population inference. A Fusion-of-N follow-up found that learned synthesis did not beat the mechanical set union of item IDs. The contribution is a measured engineering layer between model research and production agent practice: repeatable patterns and evaluation methods for faster, cheaper, more reliable workflows.

arXiv

AISA: AI Safety Assistant Framework for Continuous Improvement of Highway Construction

Mason Smetana, Trevor Neece, Lev Khazanovich

Job Safety Analysis (JSA) and pre-task planning can benefit from prior incident records, yet historical accident data is often stored as unstructured narratives that are difficult to consult at the point of planning. A novel framework centered on large language models (LLMs) for highway construction safety reporting and planning is proposed as a foundation for future agentic applications, prioritizing deterministic, local inferencing. The first aim is to enable classification and quality scoring of incident narratives for existing and future reporting purposes. The second is to evaluate retrieval of relevant historical accidents, related imagery, and trusted industry documents for incorporation into daily safety plans. Neural probes were trained to classify incidents along four multiclass and two binary Occupational Injury and Illness Classification System (OIICS) fields and to derive an overall quality score, evaluated on a test set of over 15,000 narratives and a held-out set of 100 author-labeled records, benchmarked against a majority-vote LLM ensemble. The retrieval of historical accidents, reference imagery, and industry documents was benchmarked across embedding models using standard information retrieval metrics. OIICS classification reached 75% held-out accuracy, though the two binary flags were degenerate. The quality score, while meaningful on one database, was distorted on out-of-distribution fatalities in the held-out dataset. Accident retrieval recovered relevant incidents far above chance, performing best on lexically distinct construction activities. On document question answering, an open-weight decoder embedding model surpassed proprietary models. Overall, this work provides a new framework rooted in local inferencing and text embedding models for future agentic applications, with emphasis on bridging external data to JSA reports.

arXiv

Polaris: Learning to Generate Table Descriptions from Retrieval Feedback

Ting Cai, Tuan Minh Phan, AnHai Doan

Many table-centric NLP tasks such as NL2SQL first retrieve relevant tables from large collections using keyword search. Recent work uses LLMs to generate natural-language table descriptions to improve retrieval, but they are typically optimized for fluency rather than retrieval effectiveness. We present Polaris, a system that trains an LLM to generate table descriptions directly from retrieval feedback. Our key insight is that existing table retrieval benchmarks already contain the supervision needed for this task: given query-table relevance judgments, we generate multiple candidate descriptions for each table, rank them by their BM25 retrieval effectiveness, and use the resulting preference pairs to fine-tune the LLM with Direct Preference Optimization (DPO). Polaris further expands abbreviated table and column names before generation to reduce vocabulary mismatch. Extensive experiments show that Polaris outperforms the state-of-the-art AutoDDG solution, often by a significant margin. More broadly, our results demonstrate that retrieval benchmarks can be repurposed as supervision for training LLMs to generate retrieval-oriented metadata.

arXiv

Can LLMs Reason in a Legally Meaningful Manner? A Small-scale Study on European Court of Human Rights Cases

Amogh Raina, Ilias Chalkidis, Daniel Hershcovich, Henrik Palmer Olsen

Reasoning has become a standard technique and feature for contemporary LLMs; however, its application and quality in the context of demanding legal-oriented tasks, such as legal case forecasting, remain under explored. We investigate how LLMs reason in the context of legal case forecasting, using legal cases from the European Court of Human Rights (ECtHR) as a testbed. We evaluate OpenAI GPT 5.4, a recent top-tier LLM, by exploring alternative prompting strategies that are more or less suggestive of what counts as legally meaningful reasoning in the context of ECtHR jurisprudence. We present our findings derived from assessing the model's responses with both human and LLM evaluation. We find that the examined model scores far from ideal in legal reasoning, the model produces structurally complete but substantively shallow analyses, and that LLM-as-a-Judge evaluators are internally consistent yet align only weakly with our trained annotators, i.e., reliable but not a valid substitute for human evaluation. Overall, the expert-curated prompt leads to more comprehensive reasoning, which does not result in more accurate predictions compared to the other examined settings. Based on our findings, we urge the community not to rely solely on automated LLM-based evaluation and to avoid using task accuracy as an appropriate proxy for reasoning quality.

arXiv

Towards Safer RAG: Only Agents Capable of System 2 Thinking may Access Untrusted Documents

Mehrdad Ghassabi

Retrieval-Augmented Generation (RAG) has significantly enhanced the performance of large language models (LLMs), yet these systems remain vulnerable to knowledge-poisoning attacks, in which misinformation in retrieved documents can influence the model's final outputs. Notably, an LLM may correctly detect that a document contains incorrect information while nevertheless being influenced by it. Prior work has addressed this vulnerability through the Cordon Principle, which prevents models responsible for final answer synthesis from directly accessing raw evidence. Although effective, this strict isolation can introduce substantial computational overhead. In this work, we propose a refined security principle: only agents capable of deliberative System 2 reasoning may access untrusted documents. To evaluate this principle, we introduce novel metrics that quantify the discrepancy between misinformation detection and downstream influence. We then empirically compare state-of-the-art reasoning language models with standard language models across these metrics. Our results show that reasoning-capable models are substantially more robust to corrupted evidence, without requiring the strict isolation imposed by the Cordon Principle. These findings provide empirical support for our refined principle and suggest a more practical foundation for secure RAG system design.

arXiv

KnowSim: Evaluating Information Calibration in LLM Assistants with User Simulators that Learn

Yoonjoo Lee, Hyoungwook Jin, Tae Soo Kim, Shaoyang Zhang, Philippe Laban, Q. Vera Liao

To effectively collaborate with users on knowledge-intensive tasks, Large Language Models (LLMs) must perform information calibration: matching content to a user's evolving understanding and cognitive capacity. Yet user simulators used to evaluate and train LLMs do not explicitly model user knowledge so they neither produce realistic interactions across knowledge levels nor reflect how interactions unfold as that knowledge evolves. To close this gap, we introduce KNOWSIM, an evaluation framework built around a user simulator that maintains explicit knowledge states, represented as a graph of Information Units with prerequisite relationships, that evolve under update rules grounded in learning theory. KNOWSIM computes three metrics (Knowledge Gain, Delivery Calibration, Cognitive Overload) directly from the knowledge state trajectory, reflecting key mechanistic aspects of information calibration. We validate KNOWSIM against 705 human-AI sessions across two domains, stratified by knowledge level: its rankings align significantly with human judgments (73-74% sign agreement), outperforming three baseline simulators. Applied to 9 LLMs, KNOWSIM reveals that the best model shifts by user knowledge level, revealing aptitude-treatment interactions invisible to standard evaluation.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.