Explainable Artificial Intelligence (XAI)

Derniers papiers

• RECAST: Model Reconstruction via Counterfactual-Aware Wasserstein Geometry under Limited Data

  Xuan Zhao, Lena Krieger, Zhuo Cao, Arya Bangun, Hanno Scharr, Ira Assent · 29 juin 2026

  Counterfactual explanations (CFs) help understand machine learning models by identifying minimal input changes that would lead to alternative model outcomes. Recent work demonstrates their utility for reconstructing black-box models, enabling third-party auditing of opaque decision systems for fairn…

• CBD: API-Only LLM Black-Box Unlearning through Controlled Behavioral Divergence

  Zhiqiang Xie, Yijing Lin, Zhipeng Gao, Dong In Kim · 29 juin 2026

  Edge devices increasingly invoke large language models (LLMs) through API services for context aware edge intelligence, while edge generated data may be collected to improve LLMs and may introduce sensitive, copyrighted, harmful, or outdated information into model behavior. Machine unlearning offers…

• Explainable AI for Biodiversity Monitoring and Ecological Image Analysis

  Brinnae Bent, Holly R. Houliston, Jiayi Zhou, G\"unel Aghakishiyeva, David W. Johnston · 29 juin 2026

  Artificial intelligence is transforming biodiversity monitoring by enabling automated analysis of ecological imagery collected from camera traps, drones, satellites, underwater platforms, and other sensing systems. These tools can expand the scale and speed of conservation assessments, yet many comp…

• Democratic ICAI: Debating Our Way to Steering Principles from Preferences

  Kevin Kingslin, Anish Natekar, Ashutosh Ranjan, Vivek Srivastava, Savita Bhat, Shirish Karande · 29 juin 2026

  Preference-based alignment often struggles to capture the reasoning that underlies human judgments. Many evaluations rely on multiple interacting criteria, yet pairwise labels reveal only the final choice rather than the considerations that shape preferences. Inverse Constitutional AI (ICAI) improve…

• The Curse of Multiple Mediators: Hidden Interaction Effects in Activation Patching

  Sankaran Vaidyanathan, David Arbour, Aaron Mueller, Scott Niekum, David Jensen · 29 juin 2026

  Activation patching is the primary tool in mechanistic interpretability. It attributes causal responsibility for a model behavior to each of its individual components by estimating its natural indirect effect (NIE). Re-deriving the activation patching estimand from causal mediation analysis, we find…

• Global Explanations for Multivariate Time Series Forecasting Models via $K$-Order Markov Approximations

  Amadeo Tunyi · 29 juin 2026

  While many explainable AI (XAI) methods have been proposed, most are not designed for time-series forecasting models and often rely on the implicit assumption that timestamp features are independent. This assumption ignores the fundamental property of temporal dependence and can lead to explanations…

• COCOLogic-V2: Identifying Logical Inconsistencies via Truly Hard-Negatives

  David Steinmann, Antonia W\"ust, Kristian Kersting, Wolfgang Stammer · 29 juin 2026

  While interpretable models such as concept bottleneck models (CBMs) and program synthesis methods enable verification of model decisions, their evaluation is typically limited to simple tasks, leaving complex reasoning on real-world images largely unexplored. We introduce COCOLogic-V2, an object-cen…

• OperatorSHAP: Fast and Accurate Shapley Value Estimation for Neural Operators

  Joshua Stiller, Santo M. A. R. Thies, Felix Czaja, Eyke H\"ullermeier · 29 juin 2026

  Understanding model predictions is essential for physical applications, where outputs often inform safety-critical decisions, such as structural load assessment, weather warnings, and clinical diagnosis. Shapley values satisfy many desirable properties as an attribution method, but their computation…

• Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs

  Fahd Seddik, Fatemeh Fard · 29 juin 2026

  We introduce an axiomatic evaluation framework for latent thought representations in LLMs, comprising metrics that are independent of downstream benchmark scores and reveal representational failures that benchmark accuracy masks. Existing evaluations conflate representation quality with model capaci…

• On the Effect of Uncertainty on Layer-wise Inference Dynamics

  Sunwoo Kim, Haneul Yoo, Alice Oh · 29 juin 2026

  Understanding how large language models (LLMs) internally represent and process their predictions is central to detecting uncertainty and preventing hallucinations. While several studies have shown that models encode uncertainty in their hidden states, it is underexplored how this affects the way th…

• Reasoning Beyond Prediction: From Data-Driven to Causal Software Engineering

  Roberto Pietrantuono, Luca Giamattei, Stefano Russo · 29 juin 2026

  Software engineering is an intellectually demanding, creative discipline that juggles a web of interdependent tasks to design, build, and assure the quality of increasingly complex systems. As our expectations from software soar - with demands spanning AI-driven products, pervasively distributed and…

• Triadic Werewolf: A Jester Role for Multi-Hop Theory of Mind in LLMs

  Avni Mittal · 29 juin 2026

  Theory-of-mind evaluations of large language models typically use dyadic social-deduction games, where every observable cue points to a single hidden side, so a model with strong language priors can score well without ever simulating opponents' incentives. We extend the Werewolf game with a Jester, …

• Hybrid Fact-Checking that Integrates Knowledge Graphs, Large Language Models, and Search-Based Retrieval Agents Improves Interpretable Claim Verification

  Shaghayegh Kolli, Richard Rosenbaum, Timo Cavelius, Lasse Strothe, Andrii Lata, Jana Diesner · 29 juin 2026

  Large language models (LLMs) excel in generating fluent utterances but can lack reliable grounding in verified information. At the same time, knowledge-graph-based fact-checkers deliver precise and interpretable evidence, yet suffer from limited coverage or latency. By integrating LLMs with knowledg…

• PEBS: Per-rater Empirical-Bayes Shrinkage for RLHF Reward-Model Calibration

  Arnav Raj · 29 juin 2026

  Reward models for Reinforcement Learning from Human Feedback (RLHF) pool preferences across thousands of annotators and fit one global affine calibrator, collapsing raters with systematically different rating-scale offsets and slopes into a single average-rater fit that does not match any individual…

• Information-Aware KV Cache Compression for Long Reasoning

  Jushi Kai, Zhuiri Xiao, Alexandra Birch, Zhouhan Lin · 26 juin 2026

  Reasoning capability has advanced rapidly in large language models (LLMs), leading to an increasing size of key-value (KV) cache in both prefilling and decoding stages. Existing KV cache compression methods mainly rely on attention weights to estimate token importance. While attention effectively ca…

• Detecting and Controlling Sycophancy with Cascading Linear Features

  Maty Bohacek, Rishub Jain, Nicholas Dufour, Thomas Leung, Chris Bregler, Roma Patel · 26 juin 2026

  Interpreting and controlling model behaviors through activation steering methods requires many pairs of contrastive samples that clearly exhibit desired or undesired behavior. These data pairs determine the degree to which interpretability frameworks can reliably detect model features responsible fo…

• Humans Disengage, Reasoning Models Persist: Separating Difficulty Registration from Deliberation Allocation

  Han-yu Wang · 26 juin 2026

  Large reasoning models (LRMs) take longer on harder problems, just as humans do. This surface similarity hides an opposite pattern within items. When an LRM gets a problem wrong, it spends more tokens than when it gets the same problem right; humans do the reverse, spending less time on the trials t…

• Interpreting "Interpretability" and Explaining "Explainability" in Machine Learning in Physics

  Rikab Gambhir, Luisa Lucie-Smith, Jesse Thaler · 26 juin 2026

  We review the concepts of interpretability and explainability as they apply to machine learning in physics. We define interpretability as concerning the structural transparency of a model (the ability to understand or approximate its inner workings) and explainability as concerning the scientific co…

• Do Safety Guardrails Need to Reason? LeanGuard: A Fast and Light Approach for Robust Moderation

  Dongbin Na · 26 juin 2026

  In order to screen a prompt or a response, the recent guardrail methods generate a chain-of-thought (CoT) before they issue a verdict. This design follows a common belief that step-by-step reasoning improves a decision. However, CoT also makes the guard heavy and slow, because the model must generat…

• OpenRCA 2.0: From Outcome Labels to Causal Process Supervision

  Aoyang Fang, Yifan Yang, Jin'ao Shang, Qisheng Lu, Junjielung Xu, Rui Wang, Songhan Zhang, Yuzhong Zhang, Boxi Yu, Pinjia He · 26 juin 2026

  Root cause analysis (RCA) poses a holistic test of LLM agentic capabilities, such as long-context understanding, multi-step reasoning, and tool use. However, existing datasets suffer from a fundamental gap: they label only the root cause, not the propagation path connecting it to the observed sympto…

• Discovering Millions of Interpretable Features with Sparse Autoencoders

  XinYang He, Wei Wang, Bing Zhao, Xuan Ren, WenBo Li, WeiXu Qiao, Hu Wei, Lin Qu · 26 juin 2026

  Sparse autoencoders (SAEs) have emerged as a powerful tool for decomposing superposed language model representations into sparse and interpretable features. However, training SAEs is computationally expensive, and available open-source SAE models remain limited. In this work, we introduce \textbf{Qw…

• Radical AI Interpretability

  Daniel A. Herrmann, Benjamin A. Levinstein · 26 juin 2026

  We develop a framework for interpreting AI systems as agents, drawing on the philosophical tradition of radical interpretation and the tools of mechanistic interpretability. The core question is: given the computational facts about a system, how do we solve for its beliefs, desires, and meanings? Th…

• Reducing Conversational Escalation in Large Language Model Dialogue with Nonviolent Communication Constraints

  Zhixing Sun, Shenghe Xu, Tao Li · 26 juin 2026

  Large language models (LLMs) are increasingly used in emotionally charged situations involving interpersonal conflict, frustration, and distress. While prior safety research has focused on preventing explicit harms such as toxic or policy-violating content, less attention has been paid to conversati…

• The Verification Horizon: No Silver Bullet for Coding Agent Rewards

  Binghai Wang, Chenlong Zhang, Dayiheng Liu, Jiajun Zhang, Jiawei Chen, Mouxiang Chen, Rongyao Fang, Siyuan Zhang, Xuwu Wang, Yuheng Jing, Zeyao Ma, Zeyu Cui · 26 juin 2026

  A classical intuition holds that verifying a solution is easier than producing one. For today's coding agents, this intuition is being inverted: as foundation models develop stronger reasoning capabilities and engineering harnesses grow more sophisticated, generating complex candidate solutions is n…

• Externalizing Research Synthesis and Validation in AI Scientists through a Research Harness

  Zijian Wang, Hanqi Li, Ziyue Yang, Zijian Hu, Shenghan Zuo, Yunzhe Zhang, Da Ma, Danyu Luo, Chenrun Wang, Jing Peng, Tiancheng Huang, Sijia Guo, Huayang Wang, Zichen Zhu, Senyu Han, Yilu Cao, Bo Chen, Xin Chen, Kai Yu, Lu Chen · 25 juin 2026

  AI systems can increasingly automate scientific workflows, but the reasoning that links prior evidence, generated ideas, experiments and final claims often remains implicit inside model inference. Here we introduce Xcientist, a research harness that externalizes research synthesis and experimental v…