Explainable Artificial Intelligence (XAI)

Derniers papiers

• RelP: Faithful and Efficient Circuit Discovery in Language Models via Relevance Patching

  Farnoush Rezaei Jafari, Oliver Eberle, Ashkan Khakzar, Neel Nanda · 31 octobre 2025

  Activation patching is a standard method in mechanistic interpretability for localizing the components of a model responsible for specific behaviors, but it is computationally expensive to apply at scale. Attribution patching offers a faster, gradient-based approximation, yet suffers from noise and …

• Experiments with Optimal Model Trees

  Sabino Francesco Roselli, Eibe Frank · 31 octobre 2025

  Model trees provide an appealing way to perform interpretable machine learning for both classification and regression problems. In contrast to ``classic'' decision trees with constant values in their leaves, model trees can use linear combinations of predictor variables in their leaf nodes to form p…

• Towards Piece-by-Piece Explanations for Chess Positions with SHAP

  Francesco Spinnato · 31 octobre 2025

  Contemporary chess engines offer precise yet opaque evaluations, typically expressed as centipawn scores. While effective for decision-making, these outputs obscure the underlying contributions of individual pieces or patterns. In this paper, we explore adapting SHAP (SHapley Additive exPlanations) …

• Faithful and Fast Influence Function via Advanced Sampling

  Jungyeon Koh, Hyeonsu Lyu, Jonggyu Jang, Hyun Jong Yang · 31 octobre 2025

  How can we explain the influence of training data on black-box models? Influence functions (IFs) offer a post-hoc solution by utilizing gradients and Hessians. However, computing the Hessian for an entire dataset is resource-intensive, necessitating a feasible alternative. A common approach involves…

• Activation Matching for Explanation Generation

  Pirzada Suhail, Aditya Anand, Amit Sethi · 30 octobre 2025

  In this paper we introduce an activation-matching--based approach to generate minimal, faithful explanations for the decision-making of a pretrained classifier on any given image. Given an input image $x$ and a frozen model $f$, we train a lightweight autoencoder to output a binary mask $m$ such tha…

• Tree Ensemble Explainability through the Hoeffding Functional Decomposition and TreeHFD Algorithm

  Cl\'ement B\'enard · 30 octobre 2025

  Tree ensembles have demonstrated state-of-the-art predictive performance across a wide range of problems involving tabular data. Nevertheless, the black-box nature of tree ensembles is a strong limitation, especially for applications with critical decisions at stake. The Hoeffding or ANOVA functiona…

• FaCT: Faithful Concept Traces for Explaining Neural Network Decisions

  Amin Parchami-Araghi, Sukrut Rao, Jonas Fischer, Bernt Schiele · 30 octobre 2025

  Deep networks have shown remarkable performance across a wide range of tasks, yet getting a global concept-level understanding of how they function remains a key challenge. Many post-hoc concept-based approaches have been introduced to understand their workings, yet they are not always faithful to t…

• Right for the Right Reasons: Avoiding Reasoning Shortcuts via Prototypical Neurosymbolic AI

  Luca Andolfi, Eleonora Giunchiglia · 30 octobre 2025

  Neurosymbolic AI is growing in popularity thanks to its ability to combine neural perception and symbolic reasoning in end-to-end trainable models. However, recent findings reveal these are prone to shortcut reasoning, i.e., to learning unindented concepts--or neural predicates--which exploit spurio…

• PULSE: Practical Evaluation Scenarios for Large Multimodal Model Unlearning

  Tatsuki Kawakami, Kazuki Egashira, Atsuyuki Miyai, Go Irie, Kiyoharu Aizawa · 29 octobre 2025

  In recent years, unlearning techniques, which are methods for inducing a model to "forget" previously learned information, have attracted attention as a way to address privacy and copyright concerns in large language models (LLMs) and large multimodal models (LMMs). While several unlearning benchmar…

• Einsum Networks: Fast and Scalable Learning of Tractable Probabilistic Circuits

  Robert Peharz, Steven Lang, Antonio Vergari, Karl Stelzner, Alejandro Molina, Martin Trapp, Guy Van den Broeck, Kristian Kersting, Zoubin Ghahramani · 29 octobre 2025

  Probabilistic circuits (PCs) are a promising avenue for probabilistic modeling, as they permit a wide range of exact and efficient inference routines. Recent ``deep-learning-style'' implementations of PCs strive for a better scalability, but are still difficult to train on real-world data, due to th…

• Comparison of generalised additive models and neural networks in applications: A systematic review

  Jessica Doohan, Lucas Kook, Kevin Burke · 29 octobre 2025

  Neural networks have become a popular tool in predictive modelling, more commonly associated with machine learning and artificial intelligence than with statistics. Generalised Additive Models (GAMs) are flexible non-linear statistical models that retain interpretability. Both are state-of-the-art i…

• Enhancing Pre-trained Representation Classifiability can Boost its Interpretability

  Shufan Shen, Zhaobo Qi, Junshu Sun, Qingming Huang, Qi Tian, Shuhui Wang · 29 octobre 2025

  The visual representation of a pre-trained model prioritizes the classifiability on downstream tasks, while the widespread applications for pre-trained visual models have posed new requirements for representation interpretability. However, it remains unclear whether the pre-trained representations c…

• Now you see me! Attribution Distributions Reveal What is Truly Important for a Prediction

  Nils Philipp Walter, Jilles Vreeken, Jonas Fischer · 28 octobre 2025

  Neural networks are regularly employed in high-stakes decision-making, where understanding and transparency is key. Attribution methods have been developed to gain understanding into which input features neural networks use for a specific prediction. Although widely used in computer vision, these me…

• Minimizing False-Positive Attributions in Explanations of Non-Linear Models

  Anders Gj{\o}lbye, Stefan Haufe, Lars Kai Hansen · 28 octobre 2025

  Suppressor variables can influence model predictions without being dependent on the target outcome, and they pose a significant challenge for Explainable AI (XAI) methods. These variables may cause false-positive feature attributions, undermining the utility of explanations. Although effective remed…

• Transformer Key-Value Memories Are Nearly as Interpretable as Sparse Autoencoders

  Mengyu Ye, Jun Suzuki, Tatsuro Inaba, Tatsuki Kuribayashi · 28 octobre 2025

  Recent interpretability work on large language models (LLMs) has been increasingly dominated by a feature-discovery approach with the help of proxy modules. Then, the quality of features learned by, e.g., sparse auto-encoders (SAEs), is evaluated. This paradigm naturally raises a critical question: …

• FlightKooba: A Fast Interpretable FTP Model

  Jing Lu, Xuan Wu, Yizhun Tian, Songhan Fan, Yali Fang · 28 octobre 2025

  Flight trajectory prediction (FTP) and similar time series tasks typically require capturing smooth latent dynamics hidden within noisy signals. However, existing deep learning models face significant challenges of high computational cost and insufficient interpretability due to their complex black-…

• Distributional Training Data Attribution: What do Influence Functions Sample?

  Bruno Mlodozeniec, Isaac Reid, Sam Power, David Krueger, Murat Erdogdu, Richard E. Turner, Roger Grosse · 28 octobre 2025

  Randomness is an unavoidable part of training deep learning models, yet something that traditional training data attribution algorithms fail to rigorously account for. They ignore the fact that, due to stochasticity in the initialisation and batching, training on the same dataset can yield different…

• Representational Difference Explanations

  Neehar Kondapaneni, Oisin Mac Aodha, Pietro Perona · 28 octobre 2025

  We propose a method for discovering and visualizing the differences between two learned representations, enabling more direct and interpretable model comparisons. We validate our method, which we call Representational Differences Explanations (RDX), by using it to compare models with known conceptua…

• Measuring the (Un)Faithfulness of Concept-Based Explanations

  Shubham Kumar, Narendra Ahuja · 28 octobre 2025

  Post-hoc, unsupervised concept-based explanation methods (U-CBEMs) translate a vision model's internal reasoning into human-understandable concepts, leading to interpretable explanations. However, we find that many state-of-the-art (SOTA) U-CBEMs are not faithful: their concepts seem interpretable b…

• FaithUn: Toward Faithful Forgetting in Language Models by Investigating the Interconnectedness of Knowledge

  Nakyeong Yang, Minsung Kim, Seunghyun Yoon, Joongbo Shin, Kyomin Jung · 28 octobre 2025

  Various studies have attempted to remove sensitive or private knowledge from a language model to prevent its unauthorized exposure. However, prior studies have overlooked the complex and interconnected nature of knowledge, where related knowledge must be carefully examined. Specifically, they have f…

• Combining Deep Learning and Explainable AI for Toxicity Prediction of Chemical Compounds

  Eduard Popescu, Adrian Groza, Andreea Cernat · 28 octobre 2025

  The task here is to predict the toxicological activity of chemical compounds based on the Tox21 dataset, a benchmark in computational toxicology. After a domain-specific overview of chemical toxicity, we discuss current computational strategies, focusing on machine learning and deep learning. Seve…

• From Black-box to Causal-box: Towards Building More Interpretable Models

  Inwoo Hwang, Yushu Pan, Elias Bareinboim · 28 octobre 2025

  Understanding the predictions made by deep learning models remains a central challenge, especially in high-stakes applications. A promising approach is to equip models with the ability to answer counterfactual questions -- hypothetical ``what if?'' scenarios that go beyond the observed data and prov…

• ProfileXAI: User-Adaptive Explainable AI

  Gilber A. Corrales, Carlos Andr\'es Ferro S\'anchez, Reinel Tabares-Soto, Jes\'us Alfonso L\'opez Sotelo, Gonzalo A. Ruz, Johan Sebastian Pi\~na Dur\'an · 28 octobre 2025

  ProfileXAI is a model- and domain-agnostic framework that couples post-hoc explainers (SHAP, LIME, Anchor) with retrieval - augmented LLMs to produce explanations for different types of users. The system indexes a multimodal knowledge base, selects an explainer per instance via quantitative criteria…

• Prediction-Powered Causal Inferences

  Riccardo Cadei, Ilker Demirel, Piersilvio De Bartolomeis, Lukas Lindorfer, Sylvia Cremer, Cordelia Schmid, Francesco Locatello · 27 octobre 2025

• SHAP Meets Tensor Networks: Provably Tractable Explanations with Parallelism

  Reda Marzouk, Shahaf Bassan, Guy Katz · 27 octobre 2025