Explainable Artificial Intelligence (XAI)

Derniers papiers

• Measuring What Matters: Synthetic Benchmarks for Concept Bottleneck Models

  Julian Skirzynski, Harry Cheon, Shreyas Kadekodi, Meredith Stewart, Berk Ustun · 4 juin 2026

  Concept bottleneck models predict outcomes from high-level concepts detected in inputs. Although concepts provide a simple way to reap benefits from interpretability, very few datasets include concept labels. This limits researchers' ability to determine which problems are suitable for these models,…

• Bayes-Sufficient Representations in Supervised Learning

  Vasileios Sevetlidis · 4 juin 2026

  Representation learning is often described as preserving the information in an input that is relevant for prediction. This work asks what relevance means for a fixed supervised decision problem. A representation is defined to be Bayes-sufficient for a joint distribution and loss if some prediction h…

• A Systematic Analysis of Linguistic Features in AI-Generated Text Detection Across Domains and Models

  Yassir El Attar, Esra D\"onmez, Maximilian Maurer, Agnieszka Falenska · 4 juin 2026

  Interpretable linguistic features offer a promising approach for explaining why a given text appears machine-generated, particularly for non-expert users. However, existing findings on which features reliably indicate LLM-generated text remain fragmented across feature sets, models, and text domains…

• Blessing from Human-AI Interaction: Super Reinforcement Learning in Confounded Environments

  Jiayi Wang, Zhengling Qi, Chengchun Shi · 4 juin 2026

  As AI becomes more prevalent throughout society, effective methods of integrating humans and AI systems that leverage their respective strengths and mitigate risk have become an important priority. In this paper, we introduce the paradigm of super policy learning that takes advantage of Human-AI int…

• A Geometric View of Counterfactual Behavior: Interaction of Boundary Proximity and Local Support

  Ioanna Gemou, Matteo Gamba, Randall Balestriero, Ritambhara Singh · 4 juin 2026

  Counterfactual explanations seek small, semantically meaningful changes to an input that alter a model's prediction, and are widely used to interpret and audit machine learning systems. In modern vision, language, and multimodal systems, pretrained encoders map inputs to representation spaces, and d…

• Explainably Safe Reinforcement Learning

  Sabine Rieder, Stefan Pranger, Debraj Chakraborty, Jan K\v{r}et\'insk\'y, Bettina K\"onighofer · 4 juin 2026

  Trust in a decision-making system requires both safety guarantees and the ability to interpret and understand its behavior. This is particularly important for learned systems, whose decision-making processes are often highly opaque. Shielding is a prominent model-based technique for enforcing safety…

• Tree-Based Formalization of Multi-Agent Complementarity in Human-AI Interactions

  Andrea Ferrario · 4 juin 2026

  Complementarity is the case in which a human--AI interaction (HAI) outperforms the best prediction benchmark available among its members. Although this idea is central in HAI research, formal work on complementarity remains limited. Existing frameworks do not model how agents' predictions compose in…

• SMAC-Talk: A Natural Language Extension of the StarCraft Multi-Agent Challenge for Large Language Models

  Joel Sol, Homayoun Najjaran · 4 juin 2026

  As LLMs become more widely deployed, they are increasingly expected to work alongside other AI agents rather than operating in isolation. Effective coordination in these settings requires agents to communicate, share information and make decisions under uncertainty. We introduce SMAC-Talk, a natural…

• CRAFT: Cost-aware Refinement And Front-aware Tuning of Prompts

  Shanu Kumar, Shubhanshu Khandelwal, Akhila Yesantarao Venkata, Parag Agrawal, Yova Kementchedjhieva, Manish Gupta · 4 juin 2026

  Prompts tuned for accuracy often grow long, raising inference cost on every model call. The best accuracy-cost trade-off depends on the task and the budget, so prompt optimization is a search over the Pareto front of accuracy and prompt-token cost rather than for one prompt. The usual shortcut, coll…

• GeoMin: Data-Efficient Semi-Supervised RLVR via Geometric Distribution Modeling

  Guangcheng Zhu, Shenzhi Yang, Haobo Wang, Xing Zheng, Yingfan MA, Xuening Feng, Zhongqi Chen, Kai Tang, Zhengqing Zang, Bowen Song, Weiqiang Wang, Gang Chen · 4 juin 2026

  Reinforcement learning with verifiable rewards (RLVR) significantly advances LLM reasoning, yet it faces a dilemma: standard supervised scaling is throttled by high annotation costs, while unsupervised alternatives suffer from severe model collapse. Recent semi-supervised RLVR methods address this b…

• Do Explanations Increase the Risk of Decision Logic Leakage? Explanation-Guided Stealing of Graph Models

  Bin Ma, Yuyuan Feng, Minhua Lin, Enyan Dai · 3 juin 2026

  Graph Neural Networks (GNNs) have become essential tools for analyzing graph-structured data in domains such as drug discovery and financial analysis, leading to a growing demand for model transparency. Recent advances in explainable GNNs have addressed this need by revealing important subgraphs tha…

• Building Trust in Black-box Optimization: A Comprehensive Framework for Explainability

  Nazanin Nezami, Hadis Anahideh · 3 juin 2026

  Optimizing costly black-box functions within a constrained evaluation budget presents significant challenges in many real-world applications. Surrogate Optimization (SO) is a common resolution, yet its proprietary nature introduced by the complexity of surrogate models and the sampling core (e.g., a…

• The Reliability Gap in Benchmark Auditing: Distribution Shift and Scale as Failure Modes of Contamination Detection

  Wojciech Zarzecki, Jan Dubi\'nski, Sebastian Cygert · 3 juin 2026

  Benchmark contamination, where evaluation examples appear in a model's training data, threatens the validity of LLM assessment. Statistical tools for detecting training-data membership exist, but have been validated almost exclusively in controlled academic regimes: large, homogeneous pre-training c…

• Expert-Aware Causal Tracing of Factual Recall in Sparse MoE Language Models

  Yuetian Lu, Ali Modarressi, Yihong Liu, Hinrich Sch\"utze · 3 juin 2026

  Causal tracing of factual recall has been studied predominantly in dense transformer language models, where interventions localize information flow to layers or feed-forward modules. Sparse mixture-of-experts (MoE) language models introduce a sharper question: when a factual prediction is mediated b…

• CoEval: Ranking Language Models for Custom Tasks Without Labeled Data or Trustworthy Benchmarks

  Alexander Apartsin, Yehudit Aperstein · 3 juin 2026

  Choosing or ranking language models for a specific application is hardest when no task-specific labeled data exists, and standard public benchmarks cannot be trusted, their items having likely leaked into pretraining, so scores reflect memorization rather than fitness. We present CoEval, an open-sou…

• How Quantization Changes Interpretable Features: A Sparse Autoencoder Analysis of Language Models

  Evan Duan · 3 juin 2026

  Quantization is a standard path to deploying large language models, and a quantized model is typically judged acceptable when its perplexity or downstream accuracy stays close to the full-precision original. Whether the model still computes in the same way, or whether the interpretable features iden…

• Attribution via Distributional Paths for Information Revelation

  Kieran A. Murphy, Shameen Shrestha · 3 juin 2026

  Feature attribution methods explain predictions by assigning importance scores to input features. Path-based methods such as Integrated Gradients are especially appealing because they satisfy \textit{completeness}: attributions sum to the change in model output between a reference state and the inpu…

• ScoreStop: Gradient-based early stopping using functional score tests

  Oliver J. Hines, Christian L. Hines · 3 juin 2026

  Gradient boosted decision trees require a stopping rule to avoid overfitting. The standard rule monitors a validation loss and stops if the loss fails to improve for a fixed patience period. However, the patience parameter has no interpretable scale and validation losses can be noisy or implicitly d…

• TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages

  Victor Akinode, Senyu Li, Wassim Hamidouche, Waqas Zamir, Inbal Becker-Reshef, David Ifeoluwa Adelani · 2 juin 2026

  Safety evaluation of Large Language Models (LLMs) remains heavily English-centric, leaving Low-Resource Languages (LRLs), particularly African ones, critically underexplored. We introduce TUKABENCH, a jailbreak benchmark for seven African languages that extends JailbreakBench (JBB) beyond direct tra…

• The Paradox of Outcome Optimization: A Causal Information-Theoretic Bound on Reasoning Shortcuts in LLMs

  Zihan Chen, Yiming Zhang, Wenxiang Geng, Zenghui Ding, Yining Sun · 2 juin 2026

  Large Language Models (LLMs) aligned via outcome-based Reinforcement Learning (RL) frequently exhibit a critical failure mode: they achieve high performance on in-distribution benchmarks while demonstrating brittle reasoning capabilities on out-of-distribution (OOD) tasks. We term this phenomenon Re…

• ForeSci: Evaluating LLM Agents for Forward-Looking AI Research Judgment

  Qiuyu Tian, Zequn Liu, Yingce Xia, Haojie Yin, Youyong Kong · 2 juin 2026

  AI research often requires decisions before future evidence exists: which bottleneck to attack, which direction to pursue, or where a project should be positioned. We introduce ForeSci, a temporally controlled benchmark for evaluating whether LLM agents can make such forward-looking research judgeme…

• Hidden Thoughts Are Not Secret: Reasoning Trace Exposure in LLMs

  Yu-An Lu, Ci-Yang Tsai, Yu-Lin Tsai, Raluca Ada Popa, Chia-Mu Yu · 2 juin 2026

  Reasoning traces have become a valuable form of learning signals for improving and transferring the capabilities of large language models. In particular, detailed traces can help distill reasoning behavior from stronger teacher models into weaker student models. The value of capability transfer has …

• Explainable AI Through a Democratic Lens: DhondtXAI for D'Hondt-Projected Feature Attribution

  Turker Berk Donmez · 2 juin 2026

  This study presents DhondtXAI as a SHAP-independent, D'Hondt-based attribution framework for tabular XAI. Instead of model-native feature importance or SHAP values, DhondtXAI computes background-interventional removal effects, separates positive and negative evidence, forms optional feature alliance…

• Ablating Archetypes: The Stability of Archetypal SAEs is an Artifact of Initialization and Metric Design

  Micha{\l} Brzozowski, Neo Christopher Chung · 2 juin 2026

  Dictionary learning with sparse autoencoders (SAEs) produces overcomplete bases from neural network activations that are often interpretable and reduces polysemanticity. However, features from SAEs vary substantially across random seeds -- a problem known as instability. Archetypal SAEs (Fel et al.,…

• Hoeffding Concept Bottleneck Models with Applications to Overhead Images

  Cl\'ement B\'enard, Manon Arfib, Christophe Labreuche, Victor Qu\'etu · 2 juin 2026

  Explainability of deep learning algorithms is critical for computer-vision applications with high-stake decisions. Concept bottleneck models (CBM) have recently shown promising performance to provide explainable and accurate predictions for classification problems, based on a bottleneck of high-leve…