A new framework quantifies how model explanations influence each other.
problem Understanding how different model explanations interact and influence each other.
method Introducing the metagame, a conceptual framework for measuring second-order interaction effects of model explanations using Shapley values.
result Meta-attributions provide directional insights into how feature interactions influence model explanations.
New method interprets deep neural networks for better recommendation system understanding.
problem Making deep neural networks explainable for better user trust and understanding.
method Proposes a novel formulation of interpretable deep neural networks using masked weights and hidden features.
result Demonstrates models achieving close predictive performance with informative attributions.
Improved scalability and interpretability in training data attribution.
problem Identifying which training data drives specific behaviors, especially unintended ones.
method Leveraging interpretable structures within the model to attribute model behavior to semantic directions, not individual test examples.
result Simple probe-based attribution methods are first-order approximations of Concept Influence that achieve comparable performance while being over an order-of-magnitude faster.
This paper presents a Semantic Attribute Modulation (SAM) for language modeling and style variation. The semantic attribute modulation includes various document attributes, such as titles, authors, and document categories. We consider two types of attributes, (title attributes and category attributes), and a flexible a…
The chapter improves deep learning models by interpreting and improving their performance.
problem Deep learning models often lack interpretability, leading to poor understanding of their predictions.
method The approach involves attributing importance to features and feature groups, including interactions, to improve model performance.
result The proposed attributions provide insights across various domains and can be used to improve model generalization.
Archipelago provides interpretable explanations of feature interactions in machine learning models.
problem Interpreting the impact of feature interactions on predictions in machine learning models.
method Archipelago is a novel framework for extracting and attributing feature interactions in a scalable and interpretable manner.
result Archipelago provides significantly more interpretable explanations of feature interactions than comparable methods.
In this work we present the novel ASTRID method for investigating which attribute interactions classifiers exploit when making predictions. Attribute interactions in classification tasks mean that two or more attributes together provide stronger evidence for a particular class label. Knowledge of such interactions make…
TSInsight improves interpretability of deep time-series models.
problem Lack of interpretability methods for time-series data.
method Attach auto-encoder to classifier with sparsity-inducing norm, fine-tune based on gradients and reconstruction penalty.
result TSInsight effectively boosts interpretability of deep time-series models.
The paper formalizes feature attribution to address inconsistent definitions and evaluate methods.
problem Inconsistent definitions of feature relevance in feature attribution.
method Formalization based on relaxed functional dependence, extended to instance-wise setting.
result State-of-the-art methods often fail to verify necessary properties for candidate selection.
New method makes CNN interpretations robust to adversarial attacks.
problem Adversarial attacks on CNN interpretation maps.
method Renyi Differential Privacy (RDP) for robust interpretation.
result Certifiable top-k robustness and improved experimental robustness. A new method assigns anomaly scores to features for better interpretation.
problem Interpreting anomaly scores from feature attributions.
method Proposes a characteristic function to attribute anomaly scores using Shapley value.
result Demonstrates the potential utility of the proposed attribution methods.
Greedy PIG adapts integrated gradients for better feature attribution.
problem Interpreting deep learning model predictions.
method Unified discrete optimization framework for feature attribution and selection.
result Greedy PIG improves feature attribution on various tasks.
RoSHAP stabilizes feature attribution in machine learning models.
problem Stochastic variation in feature attribution measures.
method Modeling feature attribution score distribution and estimating it through bootstrap resampling and kernel density estimation.
result RoSHAP provides stable feature rankings and improves model performance.
Paper explores alternative cooperative game theory methods for machine learning feature attribution.
problem Debate over Shapley values' relevance in feature attribution.
method Introduces Weber and Harsanyi sets as alternative allocation schemes.
result Provides a coherent framework for designing robust feature attributions.
Improved 3D ECG feature attributions for clinical interpretation.
problem Lack of interpretability in deep learning models for 12-lead ECG analysis.
method Cross-modal mapping of feature attributions from 12-lead ECG models onto CineECG 3D space.
result Mapped feature attributions yield higher Dice scores than standard 12-lead attributions.
SWAG combines screening and wrapper methods for interpretable sparse learning.
problem Need for interpretable and replicable machine learning results in sparse settings.
method Greedy exploration of attribute space to find sparse learners.
result Library of equivalent learners for attribute availability.
This paper presents the Variation Network (VarNet), a generative model providing means to manipulate the high-level attributes of a given input. The originality of our approach is that VarNet is not only capable of handling pre-defined attributes but can also learn the relevant attributes of the dataset by itself. Thes…
Recent research has demonstrated that feature attribution methods for deep networks can themselves be incorporated into training; these attribution priors optimize for a model whose attributions have certain desirable properties -- most frequently, that particular features are important or unimportant. These attributio…
Proposes DAPr framework to learn feature importance from prior knowledge.
problem Ensuring meaningful feature attributions in deep models.
method Jointly learns feature importance from prior knowledge and biases models to rely on important features.
result Improves model generalization and provides new interpretation methods.
Interpretability is an important area of research for safe deployment of machine learning systems. One particular type of interpretability method attributes model decisions to input features. Despite active development, quantitative evaluation of feature attribution methods remains difficult due to the lack of ground t…
Interpretable ML methods for better decision-making with explanations.
problem Lack of transparency in black-box ML models.
method Use of Formal Concept Analysis and cooperative game theory to assess attribute importance and reduce attribute count.
result Developed methods to assess attribute importance and reduce attribute count in ML models.
DFI maps covariates to latent representations for feature importance.
problem Feature importance when predictors are statistically dependent.
method Disentangled Feature Importance (DFI) using entropic optimal transport.
result DFI yields stable, interpretable, uncertainty-quantified attributions of shared predictive signal.
A framework learns interpretable models with minimal accuracy loss.
problem Tackles interpretability in deep learning models.
method Jointly learns a predictive model and its interpretation model with minimal loss of accuracy.
result Generates interpretable models with minimal loss of accuracy.
New method for interpreting financial model risks.
problem Fairly allocating risk in financial models.
method Extending Shapley value framework for axiomatic risk attribution.
result Risk can be well allocated in financial models.
Semi-Non-negative Matrix Factorization is a technique that learns a low-dimensional representation of a dataset that lends itself to a clustering interpretation. It is possible that the mapping between this new representation and our original data matrix contains rather complex hierarchical information with implicit lo…
Nonparametric method measures influence of training images on diffusion model outputs.
problem Quantifying influence of individual training examples on diffusion model outputs.
method Patch-level similarity between generated and training images, using optimal score function.
result Strong attribution performance, matching gradient-based approaches and outperforming baselines.
Paper applies ANOVA decomposition for interpretable data approximation.
problem High-dimensional data interpretation and dimensionality reduction.
method ANOVA decomposition and Grouped Transformations for interpretability.
result Ability to rank variable interactions and unimportant variables.
WISCA generates consensus explanations from conflicting model-agnostic interpretability methods.
problem Conflicting explanations from diverse interpretability algorithms.
method WISCA integrates class probability and normalized attributions to generate consistent explanations.
result WISCA consistently aligns with the most reliable individual method, improving explanation reliability.
R package innsight interprets deep neural networks predictions.
problem Interpreting predictions of deep neural networks.
method Unified and user-friendly framework implementing feature attribution methods for neural networks, independent of deep learning library.
result Offers a variety of visualization tools for tabular, signal, image data or a combination.
Community detection algorithms are fundamental tools to understand organizational principles in social networks. With the increasing power of social media platforms, when detecting communities there are two possi- ble sources of information one can use: the structure of social network and node attributes. However struc…
Bayesian approach scores influential training examples for model predictions.
problem Enhance interpretability and safety of machine learning models.
method Formulate TDA as a Bayesian information-theoretic problem, scoring subsets by information loss.
result Method aligns with classical influence scores while promoting diversity for subsets.
ICAM creates interpretable feature attribution maps for brain images.
problem Challenges in predicting class relevance from brain images due to heterogeneity and background variation.
method A VAE-GAN framework for disentangling class relevance from background features.
result FA maps generated by ICAM outperform baseline methods and support phenotype variation exploration.
Unified bounds for DP risks reduce noise and improve accuracy.
problem Difficult interpretation and calibration of DP mechanisms.
method Hypothesis-testing interpretation of DP (f-DP) and unified bounds. result Unified bounds are tighter and tunable for specific risks.
New framework improves attribution of predictive uncertainties in classification models.
problem Improper attribution of predictive uncertainties in classification tasks.
method Combines path integrals, counterfactual explanations, and generative models.
result Framework outperforms existing alternatives in quantitative evaluations.
New method attributes feature uncertainty in ML models using cooperative game theory.
problem Lack of feature-level uncertainty attribution in explainable AI.
method Proposes a novel, model-agnostic uncertainty attribution method using cooperative game theory and conformal prediction.
result Demonstrates improved runtime efficiency and practical utility in real-world applications.
Study evaluates consistency of feature attribution in deep learning for multi-omics data.
problem Challenges in interpretability of deep learning models in biological research.
method Investigation of Shapley Additive Explanations (SHAP) on multi-view deep learning models applied to multi-omics data.
result SHAP rankings are sensitive to architecture and random initialization, suggesting caution.
A barrier to the wider adoption of neural networks is their lack of interpretability. While local explanation methods exist for one prediction, most global attributions still reduce neural network decisions to a single set of features. In response, we present an approach for generating global attributions called GAM, w…
Convolutional neural network based systems have largely failed to be adopted in many high-risk application areas, including healthcare, military, security, transportation, finance, and legal, due to their highly uninterpretable "black-box" nature. Towards solving this deficiency, we teach a novel multi-task capsule net…
New method separates graph structure from node attributes to recover lost signal.
problem Standard representation learning on attributed graphs merges incompatible metric spaces, leading to geometrically flawed alignment.
method Custom variational autoencoder that separates manifold learning from structural alignment.
result Transforms geometric conflict into interpretable structural descriptor, uncovering connectivity patterns and anomalies.
A new method for disentangled latent spaces in VAEs that can manipulate attributes.
problem Disentangled representation of attributes in latent spaces of VAEs.
method Attribute-based regularization loss to enforce monotonic relationships between attributes and latent codes.
result Manipulation of attributes in latent spaces post-training.
We interpret black box predictive models using causal attribution.
problem Interpreting models trained using machine learning in high-stakes applications.
method Estimate causal effects of model inputs on output using observational data.
result Effective interpretation of black box predictive models via causal attribution.
Generic generation and manipulation of text is challenging and has limited success compared to recent deep generative modeling in visual domain. This paper aims at generating plausible natural language sentences, whose attributes are dynamically controlled by learning disentangled latent representations with designated…
TimeInf estimates data contribution in time series data, improving model performance and anomaly detection.
problem Estimating data contribution in time series datasets with temporal dependencies.
method Model-agnostic data contribution estimation method using influence scores.
result TimeInf effectively detects time series anomalies and outperforms existing methods.
Captum library simplifies model interpretability for PyTorch.
problem Improving model interpretability for various PyTorch models.
method Unified, open-source library with generic attribution algorithms and metrics.
result Unified and scalable model interpretability for multimodal inputs.
Machine learning has recently been widely adopted to address the managerial decision making problems, in which the decision maker needs to be able to interpret the contributions of individual attributes in an explicit form. However, there is a trade-off between performance and interpretability. Full complexity models a…
Adversarial sample attacks perturb benign inputs to induce DNN misbehaviors. Recent research has demonstrated the widespread presence and the devastating consequences of such attacks. Existing defense techniques either assume prior knowledge of specific attacks or may not work well on complex models due to their underl…
Improved local feature attributions using neighbourhood reference distributions.
problem Misleading results from global population in local model behaviour.
method Formulation of neighbourhood reference distributions and self-normalised importance sampling.
result Neighbourhood Shapley values provide meaningful sparse feature attributions.
MACQ method explains deep learning models by analyzing feature contributions across prediction levels.
problem Explaining deep learning model predictions.
method Global gradient-based, model-agnostic approach focusing on marginal attribution.
result MACQ separates feature contributions from interaction effects and visualizes 3-way relationships.