Preprint
Machine Learning

Explainable AI (XAI): Core Ideas, Techniques, and Solutions

Rudresh Dwivedi(Netaji Subhas University of Technology), Devam Dave(Pandit Deendayal Energy University), Het Naik(Pandit Deendayal Energy University), Smiti Singhal(Pandit Deendayal Energy University), Omer Rana(Cardiff University), Pankesh Patel(University of South Carolina), Bin Qian(Newcastle University), Zhenyu Wen, Tejal Shah(Newcastle University), Graham Morgan(Newcastle University), Rajiv Ranjan(Newcastle University)
September 4, 2022ACM Computing Surveys1,227 citations
PDFDOI

1.2k

Citations

23

Influential Citations

ACM Computing Surveys

Venue

2022

Year

Abstract

As our dependence on intelligent machines continues to grow, so does the demand for more transparent and interpretable models. In addition, the ability to explain the model generally is now the gold standard for building trust and deployment of artificial intelligence systems in critical domains. Explainable artificial intelligence (XAI) aims to provide a suite of machine learning techniques that enable human users to understand, appropriately trust, and produce more explainable models. Selecting an appropriate approach for building an XAI-enabled application requires a clear understanding of the core ideas within XAI and the associated programming frameworks. We survey state-of-the-art programming techniques for XAI and present the different phases of XAI in a typical machine learning development process. We classify the various XAI approaches and, using this taxonomy, discuss the key differences among the existing XAI techniques. Furthermore, concrete examples are used to describe these techniques that are mapped to programming frameworks and software toolkits. It is the intention that this survey will help stakeholders in selecting the appropriate approaches, programming frameworks, and software toolkits by comparing them through the lens of the presented taxonomy.

Analysis

Why This Paper Matters

As AI systems are deployed in high-stakes domains like healthcare, finance, and autonomous driving, the ability to explain model decisions has become a necessity rather than a luxury. This survey by Dwivedi et al. addresses a critical gap: the lack of a unified framework to help practitioners choose among the rapidly growing set of XAI techniques and their associated programming toolkits. With over 1,200 citations, the paper has become a foundational reference for both researchers and engineers building transparent AI systems.

The paper's timing is significant—it arrives when regulatory pressures (e.g., GDPR's right to explanation) and industry best practices demand interpretability. By organizing the fragmented XAI landscape into a coherent taxonomy, the authors enable stakeholders to systematically compare methods and select the right tool for their specific use case.

Technical Contributions

The paper's main innovation is its multi-dimensional taxonomy for classifying XAI techniques:

  • Scope: Global vs. local explanations (model-wide vs. per-prediction)
  • Stage: Ante-hoc (inherently interpretable models) vs. post-hoc (explanations after training)
  • Technique: Feature attribution, example-based, surrogate models, etc.
  • Mapping to frameworks: Each technique is linked to concrete libraries (e.g., LIME, SHAP, Integrated Gradients) and software toolkits (e.g., IBM AIX360, Google What-If Tool)
  • Development phases: The paper outlines where XAI fits in the ML lifecycle—data exploration, model debugging, validation, and deployment monitoring

Results

As a survey, the paper does not report quantitative experimental results. Instead, its value lies in the structured comparison of over 20 XAI techniques across multiple dimensions. The authors provide concrete code examples and framework recommendations, enabling practitioners to directly apply the taxonomy to their projects. The paper's high citation count (1,227) attests to its utility as a reference.

Significance

This survey has become a standard entry point for anyone entering the XAI field. By demystifying the landscape and providing actionable guidance, it accelerates the adoption of explainability in production AI systems. The taxonomy introduced here has influenced subsequent research and tool development, and the paper continues to be cited as a definitive overview of XAI techniques and frameworks.

explainable-aimachine-learninginterpretabilitysurveyprogramming-frameworkstaxonomytrust

Related Papers