Explainability Evaluation

Transformer models achieve state-of-the-art performance across domains and tasks, yet their deeply layered representations make their predictions difficult to interpret. Existing explainability methods rely on final-layer attributions, capture either local token-level attributions or global attention patterns without unification, and lack context-awareness of inter-token dependencies and structural components. They also fail to capture how relevance evolves across layers and how structural components shape decision-making. To address these limitations, we proposed the Context-Aware Layer-wise Integrated Gradients (CA-LIG) Framework, a unified hierarchical attribution framework that computes layer-wise Integrated Gradients within each Transformer block and fuses these token-level attributions with class-specific attention gradients. This integration yields signed, context-sensitive attribution maps that capture supportive and opposing evidence while tracing the hierarchical flow of relevance through the Transformer layers. We evaluate the CA-LIG Framework across diverse tasks, domains, and transformer model families, including sentiment analysis and long and multi-class document classification with BERT, hate speech detection in a low-resource language setting with XLM-R and AfroLM, and image classification with Masked Autoencoder vision Transformer model. Across all tasks and architectures, CA-LIG provides more faithful attributions, shows stronger sensitivity to contextual dependencies, and produces clearer, more semantically coherent visualizations than established explainability methods. These results indicate that CA-LIG provides a more comprehensive, context-aware, and reliable explanation of Transformer decision-making, advancing both the practical interpretability and conceptual understanding of deep neural models.

In geophysics, hydrocarbon prospect risking involves assessing the risks associated with hydrocarbon exploration by integrating data from various sources. Machine learning-based classifiers trained on tabular data have been recently used to make faster decisions on these prospects. The lack of transparency in the decision-making processes of such models has led to the emergence of explainable AI (XAI). LIME and SHAP are two such examples of these XAI methods which try to generate explanations of a particular decision by ranking the input features in terms of importance. However, explanations of the same scenario generated by these two different explanation strategies have shown to disagree or be different, particularly for complex data. This is because the definitions of "importance" and "relevance" differ for different explanation strategies. Thus, grounding these ranked features using theoretically backed causal ideas of necessity and sufficiency can prove to be a more reliable and robust way to improve the trustworthiness of the concerned explanation strategies.We propose a unified framework to generate counterfactuals as well as quantify necessity and sufficiency and use these to perform a robustness evaluation of the explanations provided by LIME and SHAP on high dimensional structured prospect risking data. This robustness test gives us deeper insights into the models capabilities to handle erronous data and which XAI module works best in pair with which model for our dataset for hydorcarbon indication.

LLM self-explanations are often presented as a promising tool for AI oversight, yet their faithfulness to the model's true reasoning process is poorly understood. Existing faithfulness metrics have critical limitations, typically relying on identifying unfaithfulness via adversarial prompting or detecting reasoning errors. These methods overlook the predictive value of explanations. We introduce Normalized Simulatability Gain (NSG), a general and scalable metric based on the idea that a faithful explanation should allow an observer to learn a model's decision-making criteria, and thus better predict its behavior on related inputs. We evaluate 18 frontier proprietary and open-weight models, e.g., Gemini 3, GPT-5.2, and Claude 4.5, on 7,000 counterfactuals from popular datasets covering health, business, and ethics. We find self-explanations substantially improve prediction of model behavior (11-37% NSG). Self-explanations also provide more predictive information than explanations generated by external models, even when those models are stronger. This implies an advantage from self-knowledge that external explanation methods cannot replicate. Our approach also reveals that, across models, 5-15% of self-explanations are highly misleading. Despite their imperfections, we show a positive case for self-explanations: they encode information that helps predict model behavior.

SHAP explanations are widely used in high-stakes settings to justify decisions, yet they can differ substantially across repeated runs, even when the model, the input instance, and the prediction are held fixed. Prior work has documented disagreement between explanation methods; we show that substantial disagreement arises even within SHAP across reruns of the same estimator on the same trained model and instance. We call this phenomenon explanation multiplicity and develop an evaluation methodology for characterizing it under deployment-realistic computational budgets, combining a dual-seed protocol that compares model-induced and explainer-induced variability, a hierarchy of magnitude-based, rank-based, and set-based metrics, and randomized Dirichlet and Mallows null models that provide reference scales for observed disagreement. Across multiple datasets, models, and sampling strategies, we find that explanation multiplicity is pervasive and persists even for high-confidence predictions. The relative contribution of each source depends on the data regime and model: model-induced disagreement is generally greater on smaller datasets, while explainer-induced disagreement is greater on larger datasets. Commonly used L2 distance can understate this instability, while rank-based metrics reveal substantial changes in top-ranked features, including the leading feature. Improved sampling methods such as CTE do not eliminate rank-level multiplicity, and K-Means reduces run-to-run variation while its compressed-background explanations can diverge from the empirical-distribution reference. Practitioners should treat single-run SHAP outputs as realizations of a distribution rather than as authoritative artifacts.

Complex AI systems make better predictions but often lack transparency, limiting trustworthiness, interpretability, and safe deployment. Common post hoc AI explainers, such as LIME, SHAP, HSIC, and SAGE, are model agnostic but are too restricted in one significant regard: they tend to misrank correlated features and require costly perturbations, which do not scale to high dimensional data. We introduce ExCIR (Explainability through Correlation Impact Ratio), a theoretically grounded, simple, and reliable metric for explaining the contribution of input features to model outputs, which remains stable and consistent under noise and sampling variations. We demonstrate that ExCIR captures dependencies arising from correlated features through a lightweight single pass formulation. Experimental evaluations on diverse datasets, including EEG, synthetic vehicular data, Digits, and Cats-Dogs, validate the effectiveness and stability of ExCIR across domains, achieving more interpretable feature explanations than existing methods while remaining computationally efficient. To this end, we further extend ExCIR with an information theoretic foundation that unifies the correlation ratio with Canonical Correlation Analysis under mutual information bounds, enabling multi output and class conditioned explainability at scale.

Machine learning models are primarily judged by predictive performance, especially in applied genomics, where explanations are read as biological findings. In practice, reported gene panels are stabilised by averaging, ranking, or taking consensus over the many models a pipeline produces across cross-validation folds, tuning grids, and repeated runs. This raises an overlooked question: when two models achieve high accuracy, do they rely on the same internal logic, or reach the same outcome via different mechanisms? We introduce EvoXplain, a diagnostic framework that measures whether a pipeline's explanation is uniquely determined across repeated training and model selection. Rather than analysing a single trained model, EvoXplain treats explanations as samples drawn from the training and model selection pipeline itself, without aggregating predictions or constructing ensembles, and examines whether they form a single coherent explanatory basin or separate into multiple structured basins. We evaluate EvoXplain on a TCGA pan-cancer cohort and a within-cancer breast-cancer subtype task, using elastic-net Logistic Regression and gradient-boosted trees. Although all models reach about 98% accuracy, explanation structure differs across pipelines. Holding the data split fixed and varying only the regularisation strength, equally accurate Logistic Regression models separate into a few discrete, reproducible basins that recur across 100 data splits and carry distinct biological content, while the gradient-boosted pipeline converges to one basin. The same multiplicity appears within a single cancer subtype, from the ordinary tuning step alone. EvoXplain makes explanatory structure visible, revealing when an averaged consensus corresponds to no single trained model, and reframes interpretability as a property of the training pipeline rather than of any single model.

Understanding the economic intent of Ethereum transactions is critical for user safety, yet current tools expose only raw on-chain data or surface-level intent, leading to widespread ``blind signing'' (approving transactions without understanding them). Through interviews with 16 Web3 users, we find that effective explanations should be structured, risk-aware, and grounded at the token-flow level. Motivated by these findings, we formulate TxSum, a new domain-grounded NLP task for DeFi transaction explanation, and construct a dataset of 187 complex Ethereum transactions with 2,375 token-flow annotations and transaction-level summaries. We further introduce MATEX, a grounded multi-agent framework for high-stakes transaction explanation. It selectively retrieves external knowledge under uncertainty and audits explanations against raw traces to improve token-flow-level factual consistency. MATEX achieves the strongest overall explanation quality, especially on micro-level factuality and intent quality. It improves user comprehension on complex transactions from 52.9% to 76.5% over the strongest baseline and raises malicious-transaction rejection from 36.0% to 88.0%, while maintaining a low false-rejection rate on benign transactions.

Assessing the importance of individual features in Machine Learning is critical to understand the model's decision-making process. While numerous methods exist, the lack of a definitive ground truth for comparison highlights the need for alternative, well-founded measures. This paper introduces a novel post-hoc local feature importance method called Counterfactual Importance Distribution (CID). We generate two sets of positive and negative counterfactuals, model their distributions using Kernel Density Estimation, and rank features based on a distributional dissimilarity measure. This measure, grounded in a rigorous mathematical framework, satisfies key properties required to function as a valid metric. We showcase the effectiveness of our method by comparing with well-established local feature importance explainers. Our method not only offers complementary perspectives to existing approaches, but also improves performance on faithfulness metrics (both for comprehensiveness and sufficiency), resulting in more faithful explanations of the system. These results highlight its potential as a valuable tool for model analysis. Link to repository: https://github.com/EddieConti/CID

Explanations are often promoted as tools for transparency, but they can also foster confirmation bias; users may assume reasoning is correct whenever outputs appear acceptable. We study this double-edged role of Chain-of-Thought (CoT) explanations in multimodal moral scenarios by systematically perturbing reasoning chains and manipulating delivery tones. Specifically, we analyze reasoning errors in vision language models (VLMs) and how they impact user trust and the ability to detect errors. Our findings reveal two key effects: (1) users often equate trust with outcome agreement, sustaining reliance even when reasoning is flawed, and (2) the confident tone suppresses error detection while maintaining reliance, showing that delivery styles can override correctness. These results highlight how CoT explanations can simultaneously clarify and mislead, underscoring the need for NLP systems to provide explanations that encourage scrutiny and critical thinking rather than blind trust. All code will be released publicly.

AI reasoning models have exceeded human performance on the ARC-AGI-1 benchmark, but does that mean state-of-the-art models have the underlying competence (humanlike abstract reasoning) that the benchmark was designed to test? Here we investigate the abstraction abilities of AI models using the closely related but simpler ConceptARC benchmark. Our evaluations vary input modality (textual vs. visual), use of external Python tools, and reasoning effort. Beyond output accuracy, we evaluate the natural-language rules that models generate to explain their solutions, enabling us to assess whether models recognize the abstractions that ConceptARC was designed to elicit. We show that the best models' rules are frequently based on less abstract, more domain-specific concepts, capturing intended abstractions considerably less often than humans. In the visual modality, AI models' output accuracy drops sharply; however, our rule-level analysis reveals that a substantial share of their rules capture intended abstractions, even as the models struggle to apply these concepts to generate correct solutions. In short, we show that using performance (accuracy) alone can substantially overestimate AI competence in textual modalities while underestimating it in visual modalities, an illustration of the risk of mistaking performance for competence.

Perturbation-based explainability methods face criticism due to their reliance on out-of-distribution mutants. This raises doubts about the quality of the explanations. In this paper, we introduce a novel forward pass paradigm, Activation-Deactivation (AD), which obviates the need for perturbation of the input. AD replaces perturbation of input features with switching off parts of the model corresponding to to the intended perturbations. We implement ConvAD, an AD approximation algorithm for CNNs. ConvAD is a drop-in mechanism that can be easily added to any trained CNN and, without any additional training, generates more robust and more transferable explanations. We provide evaluation results across multiple architectures, datasets, methods and perturbation strategies, demonstrating the superior quality of ConvAD compared to the SOTA.

Swarm-based optimization algorithms have demonstrated remarkable success in solving complex problems, yet their widespread adoption remains limited due to poor transparency in how algorithmic components influence performance. This work presents a multi-faceted explainability framework for Particle Swarm Optimization (PSO) through two complementary perspectives: landscape-based and algorithmic explainability. From the landscape-based perspective, we develop a comprehensive characterization framework using Exploratory Landscape Analysis (ELA) to quantify problem difficulty, multimodality, and ruggedness, extracting ELA meta-features, dispersion measures, and information content statistics, while a machine learning approach employing Decision Tree and Random Forest classifiers enables prediction of optimal topology-specific hyperparameter configurations for unseen problems. From the algorithmic explainability perspective, we integrate IOHxplainer for temporal convergence profiling and Search Trajectory Networks (STN) for spatial navigation mapping, proposing three novel STN metrics-Connectivity Density, Fragmentation Score, and Search Efficiency-that enhance visual explainability by quantifying topology-specific search organization and transition effectiveness. Through systematic experimentation across 24 benchmark functions in multiple dimensions with Star, Ring, and Von Neumann topologies, we establish practical guidelines for topology selection and parameter configuration. Our findings uncover the black-box nature of PSO, providing greater transparency and interpretability to swarm intelligence systems. The source code is available at https://github.com/GitNitin02/ioh_pso.

Active learning is a general learning mechanism shared by artificial and human learners. Whether AI can teach humans such a strategy that transfers across domains is an open question. Ultra Strong Machine Learning (USML), a system whose explanations quantifiably improve human out-of-sample performance compared to self-learning, is uniquely positioned to answer this question. Prior USML work relied on hand-crafted explanation templates that require expert effort for each new domain and do not scale. We developed an explanation pipeline combining Inductive Logic Programming (ILP) with large language models (LLMs) to automate explanation generation and scoring. We tested whether these explanations achieve USML in a human trial teaching active learning strategies across three related domains. Our exploratory results show that concise, expert-written explanations benefit learners with higher initial performance, while pipeline-generated explanations provide no advantage over self-learning despite being rated as higher quality from an LLM-as-judge evaluation. This case study reveals a systematic gap that LLM quality metrics do not predict human learning outcomes. Our findings point to explanation complexity relative to task difficulty as a key factor, and call for explanation methods and evaluation criteria grounded in human cognitive constraints rather than LLM preference.

Convolutional Neural Networks (CNNs) are an effective approach for classification tasks, particularly when the training dataset is large. Although CNNs have long been considered a black-box classification method, they can be used as a white-box method through visualization techniques such as Grad-CAM. When the training samples are limited, incorporating a Principal Component Analysis (PCA) layer and/or a Support Vector Machine (SVM) classifier into a CNN can effectively improve the classification performance. However, a conventional Grad-CAM cannot be directly applied to PCA and/or SVM layers. Generating attention regions for PCA and/or SVM layers in CNNs is important to facilitate the development of white-box methods. Therefore, we propose PCA-Grad-CAM'', a method for visualizing attention regions in PCA feature vectors, and SVM-Grad-CAM'', a method for visualizing attention regions in an SVM classifier layer. Solving a closed-form Jacobian problem comprising partial derivatives from the last convolutional layer to the PCA and/or SVM layers is necessary to complete the proposed methods analytically. In this paper, we present the exact closed-form Jacobian and visualization results of the proposed methods applied to several major datasets. In addition, the insertion and deletion metrics, which are major evaluation metrics in explainable AI, were applied to PCA- and SVM-Grad-CAM. The results suggest that the proposed method can successfully visualize the attention regions of PCA and SVM.

Knowledge graphs (KGs) can be enhanced through rule mining; however, the resulting logical rules are often difficult for humans to interpret due to their inherent complexity and the idiosyncratic labeling conventions of individual KGs. This work presents Rule2Text, a comprehensive framework that leverages large language models (LLMs) to generate natural language explanations for mined logical rules, thereby improving KG accessibility and usability. We conduct extensive experiments using multiple datasets, including Freebase variants (FB-CVT-REV, FB+CVT-REV, and FB15k-237) as well as the ogbl-biokg dataset, with rules mined using AMIE 3.5.1. We systematically evaluate several LLMs across a comprehensive range of prompting strategies, including zero-shot, few-shot, variable type incorporation, and Chain-of-Thought reasoning. To systematically assess models' performance, we conduct a human evaluation of generated explanations on correctness and clarity. To address evaluation scalability, we develop and validate an LLM-as-a-judge framework that demonstrates strong agreement with human evaluators. Leveraging the best-performing model (Gemini 2.0 Flash), LLM judge, and human-in-the-loop feedback, we construct high-quality ground truth datasets, which we use to fine-tune the open-source Zephyr model. Our results demonstrate significant improvements in explanation quality after fine-tuning, with particularly strong gains in the domain-specific dataset. Additionally, we integrate a type inference module to support KGs lacking explicit type information. All code and data are publicly available at https://github.com/idirlab/KGRule2NL.

Innovations in digital intelligence are transforming robotic surgery with more informed decision-making. Real-time awareness of surgical instrument presence and actions (e.g., cutting tissue) is essential for such systems. Yet, despite decades of research, most machine learning models for this task are trained on small datasets and still struggle to generalize. Recently, vision-Language Models (VLMs) have brought transformative advances in reasoning across visual and textual modalities. Their unprecedented generalization capabilities suggest great potential for advancing intelligent robotic surgery. However, surgical VLMs remain under-explored, and existing models show limited performance, highlighting the need for benchmark studies to assess their capabilities and limitations and to inform future development. To this end, we benchmark the zero-shot performance of several advanced VLMs on two public robotic-assisted laparoscopic datasets for instrument and action classification. Beyond standard evaluation, we integrate explainable AI to visualize VLM attention and uncover causal explanations behind their predictions. This provides a previously underexplored perspective in this field for evaluating the reliability of model predictions. We also propose several explainability analysis-based metrics to complement standard evaluations. Our analysis reveals that surgical VLMs, despite domain-specific training, often rely on weak contextual cues rather than clinically relevant visual evidence, highlighting the need for stronger visual and reasoning supervision in surgical applications.

The evaluation of explainable AI (XAI) approaches often relies on three families of methods: subjective measures (e.g., questionnaires on trust or satisfaction), objective measures (e.g., task performance metrics), and mathematical metrics (e.g., for faithfulness). Yet, it remains unclear how these families align or diverge in practice. In a{preregistered} between-subjects study (N=166), we use three established saliency map techniques (LIME, Grad-CAM, Guided Backpropagation) as a testbed to examine this issue. We find that each family of methods leads to different conclusions: participants reported no differences in trust or satisfaction, Grad-CAM improved user performance, while mathematical metrics favored Guided Backpropagation. At the same time, mathematical metrics were only partially related to user performance, and these relationships were sometimes counterintuitive. Our findings highlight the methodological importance of comparing subjective, objective, and mathematical approaches when evaluating XAI, illustrating both tensions and aspects that are aligned. We discuss implications for XAI evaluation frameworks.

To develop a trustworthy AI system, it is essential to understand its behavior using attribution methods, which aim to identify the input regions that most influence the model's decisions. For discrete inputs such as images, however, accurately identifying input-prediction interactions is particularly challenging because of combinatorial explosion. We formulate black-box attribution as an ordered interpretable subset-selection problem. Inspired by the marginal-gain principle of submodular optimization, the formulation permits redundancy, positive complementarity, and negative interference among input sub-regions. We propose LiMA (Less input is More faithful for Attribution), which ranks sub-regions by their contributions along a sequence of selected subsets. LiMA employs a composite objective that evaluates subsets through semantic consistency, collaboration, prediction confidence, and semantic effectiveness. We further design a bidirectional greedy search to efficiently identify important and non-critical sub-regions. Experiments on eight foundation models (CLIP, ImageBind, LanguageBind, QuiltNet, etc.) and six datasets (ImageNet, VGG-Sound, CUB-200-2011, etc.) demonstrate strong generalization and faithful interpretations with fewer sub-regions. LiMA achieves state-of-the-art Insertion, Deletion, and average highest confidence, with average improvements of 36.3% in Insertion and 39.6% in Deletion; it is also 1.6 times faster than naive greedy search. For failure attribution on incorrect predictions, LiMA improves average highest confidence by 86.1% over state-of-the-art methods. The code is available at https://github.com/RuoyuChen10/LIMA.

When asked to explain their decisions, LLMs can often give explanations which sound plausible to humans. But are these explanations faithful, i.e. do they convey the factors actually responsible for the decision? In this work, we analyse counterfactual faithfulness across 75 models from 13 families. We analyze the tradeoff between conciseness and comprehensiveness, how correlational faithfulness metrics assess this tradeoff, and the extent to which metrics can be gamed. This analysis motivates two new metrics: the phi-CCT, a simplified variant of the Correlational Counterfactual Test (CCT) which avoids the need for token probabilities while explaining most of the variance of the original test; and F-AUROC, which eliminates sensitivity to imbalanced intervention distributions and captures a model's ability to produce explanations with different levels of detail. Our findings reveal a clear scaling trend: larger and more capable models are consistently more faithful on all metrics we consider. Our code is available at https://github.com/google-deepmind/corr_faith.

Concept-based explanations help users understand vision models through recognizable visual patterns. However, existing methods often rely on correlational signals without directly validating which image cues support prediction-relevant internal features. To this end, we introduce VisionLogic, a post-hoc framework that grounds these features in visual concepts through intervention-based validation. VisionLogic first identifies compact sets of features whose contributions reproduce the model's original prediction. It then represents their activation states as predicates using class-specific thresholds. An iterative refinement procedure grounds these predicates in visual regions through ablation tests. A region is accepted when its removal deactivates the corresponding predicate, linking the feature's numerical role to visual evidence in the input. The same predicates allow us to examine how features are activated, selected, and reused across images and classes. Across CNNs and vision transformers on ImageNet-1k, we find that only a few features are selected to explain each prediction, and frequently active features are not always selected. In a large-scale human evaluation with 465 participants, VisionLogic significantly improves participants' understanding of model behavior over established methods ACE and CRAFT. Code is available at https://github.com/allengeng123/VisionLogic.

Existing algorithms for explaining the output of image classifiers use different definitions of explanations and a variety of techniques to find them. However, none of the existing tools use a principled approach based on formal definitions of cause and explanation. In this paper we present a novel black-box approach to computing explanations grounded in the theory of actual causality. We prove relevant theoretical results and present an algorithm for computing approximate explanations based on these definitions. We prove termination of our algorithm and discuss its complexity and the amount of approximation compared to the precise definition. We implemented the framework in a tool ReX and we present experimental results and a comparison with state-of-the-art tools. We demonstrate that ReX is the most efficient black-box tool and produces the smallest explanations, in addition to outperforming other black-box tools on standard quality measures.

Post-hoc attribution is widely used to identify important model inputs, but evaluating whether these attributions identify truly informative features is difficult because real datasets rarely provide reliable ground truth. We introduce a multimodal benchmark containing symbolic tabular, vision, and audio tasks with known informative feature sets. In the main benchmark conditions, informative variables, spatial regions, or channels occupy fixed input coordinates while the remaining inputs provide irrelevant or distracting information. We use the benchmark to study how attribution quality depends on predictive performance, irrelevant-feature burden and structure, model configuration, and attribution mechanism, while separately measuring identification, stability, and computational cost. In most symbolic-data sweeps, informative-set identification co-varies with predictive performance, while the relative ordering of attribution methods remains largely stable. Across modalities, no method dominates all architectures and conditions, and greater attribution complexity does not consistently improve identification. Simple gradient attribution is often competitive at lower computational cost, while the vision and audio results show that architecture and the form of irrelevant content materially affect attribution quality.