Characterizing Data Scientists’ Mental Models of Local Feature Importance

Collaris, Dennis; Weerts, Hilde J. P.; Miedema, Daphne; Wijk, Jarke J. van; Pechenizkiy, Mykola

doi:10.1145/3546155.3546670

Cited by 1 publication

(2 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Useful in privacy-preserving applications as only information around the vicinity of the instance is explored [32,33,[48][49][50][51][59][60][61][65][66][67][68][69][70]73,[75][76][77][78][79][80][81]83,84,96,[107][108][109][110][111]126] Global Useful to explain the complete working logic of the AI system to business stakeholders who decide to adopt the AI system into the business pipeline [20,[52][53][54][55][56][57][58] It is to be noted that an XAI method can fall under multiple categories based on the aspect used for categorization. Our forthcoming discussions are structured by considering the stage of incorporation of explainability as the basis for sub-categorization.…”

Section: Localmentioning

confidence: 99%

“…Zafar & Khan [59] propose a deterministic approach to sampling neighbors utilizing agglomerative hierarchical clustering and sampling k-nearest neighbors, using which an interpretable approximator is constructed. Collaris et al [65] hint at the possibility of sampling fewer neighbors when sampling is performed independent of the queried instance to be explained and propose to sample from a hypersphere around the instance to obtain a robust local explanation. Anchors [50] generate explanations for individual predictions using if-then rules constructed in a bottom-up fashion such that the rule precisely covers the local neighbors of the instance to be explained.…”

Section: Model-agnostic Explanationsmentioning

confidence: 99%

See 1 more Smart Citation

Explainable Image Classification: The Journey So Far and the Road Ahead

Kamakshi,

Krishnan

2023

View full text Add to dashboard Cite

Explainable Artificial Intelligence (XAI) has emerged as a crucial research area to address the interpretability challenges posed by complex machine learning models. In this survey paper, we provide a comprehensive analysis of existing approaches in the field of XAI, focusing on the tradeoff between model accuracy and interpretability. Motivated by the need to address this tradeoff, we conduct an extensive review of the literature, presenting a multi-view taxonomy that offers a new perspective on XAI methodologies. We analyze various sub-categories of XAI methods, considering their strengths, weaknesses, and practical challenges. Moreover, we explore causal relationships in model explanations and discuss approaches dedicated to explaining cross-domain classifiers. The latter is particularly important in scenarios where training and test data are sampled from different distributions. Drawing insights from our analysis, we propose future research directions, including exploring explainable allied learning paradigms, developing evaluation metrics for both traditionally trained and allied learning-based classifiers, and applying neural architectural search techniques to minimize the accuracy–interpretability tradeoff. This survey paper provides a comprehensive overview of the state-of-the-art in XAI, serving as a valuable resource for researchers and practitioners interested in understanding and advancing the field.

show abstract