ConEx: Human-Interpretable Saliency Maps via Concept-Aware Attribution

ConEx bridges saliency maps with concept reasoning via automatic concept discovery to generate faithful, human-interpretable visual explanations.

Yehonatan Elisha, Oren Barkan, Ziv Weiss Haddad, Noam Koenigstein

Published Oct 3, 2026 · 0 citations · ▲ 1 on Hugging Face

MEA: A Reward-Driven Multi-Agent System for Faithful Model Explanations

MEA is a multi-agent framework that uses reward-driven optimization to generate faithful natural-language explanations across tabular, text, and vision data, outperforming baselines by up to 34%.

Yuyang Cheng, R. Ravi, Srivarshinee Sridhar, Sriparna Saha and 2 more

Published Oct 1, 2026 · 0 citations · ▲ 4 on Hugging Face

XAI Evaluation Cards: A Practical Method for Designing Human-Centred XAI Evaluations

XAI Evaluation Cards provide a card-sorting method to systematically design human-centered evaluations of explainable AI systems across disciplines.

Kristýna Sirka Kacafírková, Ivania Donoso-Guzmán, Denis Parra, Katrien Verbert and 1 more

Published Oct 1, 2026 · 0 citations

ProDG: Prototypes for Data-Free Generative Post-Hoc Explainability

Piotr Borycki, Magdalena Trędowicz, Jacek Tabor, Łukasz Struski and 1 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Near-optimal Explainable $k$-means Clustering under $\ell_p$ Norm

Xinyuan Cao, Konstantin Makarychev, Ilias Papanikolaou, Liren Shan

Atlanta Poster Session 2, Wed, Dec 9, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Causal Concept Explanations for Deep Neural Models

Joshua Rountree, Pulkit Verma, Oswin So, Chuchu Fan and 1 more

Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Decomposing Effects in Neural Causal Models

Matej Zečević, Devendra Singh Dhami, Kristian Kersting

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Chain-of-Thought Is Not Explainability

Fazl Barez, Tung-Yu Wu, Iván Arcuschin Moreno, Michael Lan and 12 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Interpretable but Fragile? Robustness of Concept Bottlenecks under Geometric-Semantic Perturbations

Hanwei Zhang, Tianma Hu, Gaojie Jin, Xu Cheng and 1 more

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

NORMA: Norm-Guided Explanation Subgraph Discovery

Xiangyu Fu, Wei Liu, Jun Wang, Yang Qiu and 2 more

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Counterfactual Debugging the World Model Transfer Gap

Mingxuan Li, Kai-Zhan Lee, Michael Dennis, Elias Bareinboim

Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Unpacking the Evaluators: How Configuration Shapes the Evaluation of Alignment in Explainable AI

Gizem Karagoz, Tanir Özçelebi, Nirvana Meratnia

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Can LLMs explain themselves truthfully with code?

Nhi Nguyen, Shauli Ravfogel, Rajesh Ranganath

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Introspective Coupling: LMs Learn to Explain Themselves Better Than Their Training Targets

Zifan Carl Guo, Laura Ruis, Jacob Andreas, Belinda Z Li

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Learning Global Probabilistic Explanations

Frederic Koriche, Louenas Bounia

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

Transferring Visual Explainability from Self-Explaining to Prediction-Only Vision Transformers via Task Arithmetic

Yuya Yoshikawa, Ryotaro Shimizu, Takahiro Kawashima, Yuki Saito

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Diverse Representative Rashomon Sets for Sparse Generalized Additive Models

Varun Babbar, Christopher Li, Chudi Zhong, Cynthia Rudin

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Foveated BagNet: Inherent Interpretability Does Not Exclude Global Context

Holger Heidrich, Sarah Müller, Andreas Schilling

Paris Poster Session 4, Thu, Dec 10, 5:30 PM–7:30 PM, Paris Poster Hall · Published 2026

Explanation Mechanism Influences Human Reliance on Reinforcement Learning Agents

Xuying Zhong, Daniel Beechey, Crescent Jicol, Janina A. Hoffmann and 1 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

A New Perspective on XAI: Scientific Theory Building for Auditable Artefacts

Sebastian R. Müller, Vanessa Toborek, Tamas Horvath, Brendan B Jackson and 1 more

Paris Poster Session 3, Thu, Dec 10, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026

Stop Using Plausibility as the Criterion for Explainable AI

Weina Jin, Xiaoxiao Li, Ghassan Hamarneh

Atlanta Poster Session 2, Wed, Dec 9, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Unlocking Volition: Proactive Intention Decoding via Interpretable Graph Learning of Multi-Region ECoG

Kaizhong Zheng, JiaJunMA, Hongru Liu, Xinjian LI and 3 more

Paris Poster Session 3, Thu, Dec 10, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026

Provable Explanations for Any-Order Neural Additive Models

Idan Refaeli, Shahaf Bassan, Yizhak Y. Elboher, Guy Katz

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Coherent Routing in Decision Trees: Phase-Interference Learning for Interpretable Tabular Prediction

David Li, Angela Li

Atlanta Poster Session 4, Thu, Dec 10, 4:30 PM–7:30 PM, Hall C1 · Published 2026

Robust Surrogate Modeling for Explainable Graph Neural Networks

Zhuomin Chen, Jingchao Ni, Hojat Allah Salehi, Xu Zheng and 1 more

Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Counterfactual Explanations for Time-Series Classification via Constrained Flow Matching

Akihiro Yamaguchi, Shizuo Kaji, Kaname Matsue, Ryusei Shingaki

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

xWhy: Causal Learning from Explanations

Nicholas Tagliapietra, Florian Peter Busch, Moritz Willig, Matej Zečević and 3 more

Paris Poster Session 1, Wed, Dec 9, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026

What Comes Next and Why: Interpretable Next-Event Prediction with Neuro-Symbolic Rules

Tim Nico Bauerschmidt, Isabel Valera, Jilles Vreeken

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Concepts in Motion: Temporal Concept Bottleneck Model for Interpretable Video Classification

MoTIF uses a transformer over temporally grounded concept sequences with per-concept self-attention and automatic VLM concept discovery to improve interpretable video classification.

Patrick Knab, Sascha Marton, Philipp J Schubert, Drago A Guggiana Nilo and 1 more

Sydney Poster Session 4, Wed, Dec 9, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026 · Code ★ 6

Concept frustration: Aligning human concepts and machine representations

A geometric framework defines concept frustration as contradictions from missing concepts and detects it in foundation model embeddings to align human and machine reasoning.

Christopher R. S. Banerji, Enrico Parisini, Christopher J Soelistyo, Ahab Isaac and 1 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Measuring What Matters: Synthetic Benchmarks for Concept Bottleneck Models

Synthetic benchmarks for concept bottleneck models generate controlled labeled datasets to evaluate decision support and automation use cases, diagnose failure modes, and guide testing.

Julian Skirzynski, Harry Cheon, Shreyas Kadekodi, Meredith Stewart and 1 more

Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Counterfactual Maps: What They Are and How to Find Them

Counterfactual maps use volumetric KD trees to find globally optimal counterfactual explanations for tree ensembles via nearest-region search with millisecond queries.

Awa Khouna, Julien Ferry, Thibaut Vidal

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

LoRIF: Low-Rank Influence Functions for Scalable Training Data Attribution

LoRIF exploits low-rank gradient structure to reduce storage and query I/O to O(c√D) and inverse Hessian memory to O(Dr), achieving up to 20× speedups over LoGRA at scale.

Shuangqi Li, Hieu Le, Jingyi Xu, Mathieu Salzmann

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

FrED: External Data Influence Estimation via Domain Knowledge Graph Grounding

FrED uses knowledge graphs with black-box similarities to attribute generative model outputs to specific training data, improving artistic and weather forecast localization without model access.

Theodoros Aivalis, Iraklis A Klampanos, Antonis Troumpoukis, Joemon M Jose

Paris Poster Session 1, Wed, Dec 9, 12:30 PM–2:30 PM, Paris Poster Hall · Published 2026

ORCAID: Oblique Rule-Based Continuous-Action Interpretation for Deep RL Policies

ORCAID extracts interpretable rule-based policies from continuous-action deep RL agents via efficient oblique decision trees with hyperplane splits, local linear models, and leaf merging, maintaining strong performance with few parameters and improving original policies.

Ignacio D. Lopez-Miguel, Ezio Bartocci, Thomas Eiter, Martin Tappler

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

Validating Causal Abstraction Metrics on Simulated Complex Systems

Benchmarking thirty metrics on ten simulated complex systems shows only causal metrics reliably validate high-level explanations when testing unmapped-variable faithfulness, leading to the Causal Abstraction Error metric converging with thirty interventions.

Maxime Méloux, Tiago Pimentel, François Portet, Maxime Peyrard

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

Measuring Weak-to-Strong Legibility of Reasoning Models

This paper defines weak-to-strong legibility for reasoning models and argues existing efficiency metrics miss thoroughness needed for weak monitors.

Dani Roytburg, Shreya Sridhar, Daphne Ippolito

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Training Data Attribution in Diffusion Models via Mirrored Unlearning and Noise-Consistent Skew

MUCS improves diffusion model training data attribution via mirrored unlearning and noise-consistent skew, outperforming existing methods across datasets.

Joan Serrà, Dipam Goswami, Fabio Morreale, Wei-Hsiang Liao and 1 more

Paris Poster Session 2, Wed, Dec 9, 5:00 PM–7:00 PM, Paris Poster Hall · Published 2026

Predicting Only from Selected Evidence: A Tempered Product-of-Experts Bottleneck for Auditable EEG Diagnosis

tPoE-EIB constrains EEG diagnosis to selected evidence via tempered product-of-experts fusion, improving auditable selection and integration faithfulness while preserving accuracy.

Yinghao WANG, Shujian Yu, Duc-Han LE, Zhikai Yu and 2 more

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

Measuring Cross-Modal Synergy: A Benchmark for VLM Explainability

Cross-modal redundancy causes unimodal metrics to contradict (τ=-0.06), so Synergistic Faithfulness (F_syn) isolates joint modality dividends with ρ=0.92 and 24× speedup, revealing VLM explainers over-index visual salience versus adapted attention methods.

Joël Roman Ky, Salah GHAMIZI, Maxime Cordy

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

Comparing Explanations is not Enough,Explain the Change: New Standards are Needed to Explain Behavioral Shifts in Large Language Models

Current explainability methods fail to explain LLM behavioral shifts from interventions; this paper proposes Comparative XAI (XAIΔ) to explain model transitions and defines auditing desiderata for governance.

Martino Ciaperoni, Marzio Di Vece, Roberto Pellungrini, Luca Pappalardo and 2 more

Paris Poster Session 6, Fri, Dec 11, 2:30 PM–4:30 PM, Paris Poster Hall · Published 2026

Rethinking XAI Evaluation: A Human-Centered Audit of Shapley Benchmarks in High-Stakes Settings

Standard Shapley benchmarks misalign with human decision utility, as quantitative metrics decouple from clarity and explanations inflate confidence without improving analyst performance in high-stakes risk settings.

Inês Oliveira e Silva, Sérgio Jesus, Iker Perez, Rita P. Ribeiro and 3 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

CounterFlowNet: From Minimal Changes to Meaningful Counterfactual Explanations

CounterFlowNet uses generative flow networks to generate sequential, sparse counterfactual explanations for tabular data with enforced actionability constraints and improved validity-sparsity trade-offs.

Oleksii Furman, Patryk Marszałek, Jan Masłowski, Marek Śmieja and 2 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Symb-xMIL: Symbolic Explanations for Multiple Instance Learning in Digital Pathology

Symb-xMIL quantifies alignment between MIL predictions and human-readable logical rules to expose decision patterns, recover ground-truth rules, and refine survival stratification beyond HPV status.

Yanqng Luo, Julius Hense, Niklas Prenißl, Andreas Mock and 3 more

Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Guided Data Generation for Understanding Model Behavior

A guided data generation framework generates input distributions to inspect trained model behaviors via specification functions.

Eren Mehmet KIRAL, Nursen Aydin, Ilker Birbil

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Federated Concept-Based Models: Interpretable models with distributed supervision

Federated concept-based models aggregate distributed concept annotations across institutions, adapt architectures to evolving supervision, and enable interpretable inference for locally unavailable concepts while preserving privacy.

Dario Fenoglio, Arianna Casanova Flores, Francesco De Santis, Gabriele Dominici and 5 more

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

B-XAIC Dataset: Benchmarking Explainable AI for Graph Neural Networks Using Chemical Data

B-XAIC benchmark evaluates explainable AI for graph neural networks on real molecular tasks with ground-truth rationales, revealing major limitations in current explanation methods.

Magdalena Proszewska, Tomasz Danel, Antoni Antoszek, Dawid Damian Rymarczyk

Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Multistage Defer Trees for Hybrid Interpretability: If at First You Can't Succeed, Tree Again

Multistage Defer Trees use sequential sparse decision trees to route most samples through interpretable rules before deferring to black boxes, matching ensemble accuracy with high interpretability.

Zakk Heile, Hayden McTavish, Margo Seltzer, Cynthia Rudin

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Root Cause Analysis of Measurement and Mechanistic Anomalies

A causal model distinguishes measurement errors from mechanism shifts via latent interventions, enabling robust root-cause localization and anomaly-type classification.

Hendrik Suhr, David Kaltenpoth, Jilles Vreeken

Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026

Z0-Inf: Zeroth Order Approximation for Data Influence

Z0-Inf estimates data influence via zeroth-order checkpoint losses, achieving efficient, accurate self-influence and train-test influence estimation for large language models without gradients.

Narine Kokhlikyan, Diego Garcia-Olano, Kamalika Chaudhuri, Saeed Mahloujifar

Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026

Hyperbolic Concept Bottleneck Models

HypCBM grounds concept bottlenecks in hyperbolic space via asymmetric geometric containment to yield sparse, hierarchy-aware activations and coherent interventions without extra supervision.

Daniel Uyterlinde, Swasti S Mishra, Pascal Mettes

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

Interpreting Neural Combinatorial Optimization via Evolving Programmatic Bottlenecks

EPB distills black-box neural combinatorial optimization models into interpretable program portfolios and reveals stage-dependent heuristic-like behavior.

Haocheng Duan, Yuxin Guo, Jieyi Bi, Anqi Xie and 3 more

Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026