Survey on Explainable AI for Traditional Machine Learning and Domains

Overview

Published in: ACM Computing Surveys, Vol. 58, No. 12, Article 305 (May 2026) DOI: 10.1145/3806829

The increasing deployment of opaque AI models in high-stakes domains has intensified the demand for Explainable AI (XAI) that is both cognitively aligned and operationally embedded. This survey reconceptualizes explainability as a reflexive, system-level property spanning the entire machine learning lifecycle—from data collection and model development to deployment and monitoring.

Rather than treating explanations as post-hoc additions, we introduce a lifecycle-centric architecture that integrates explainability at every stage, supported by a cognitively grounded taxonomy of explanation strategies tailored to diverse stakeholder needs.

Research Methodology

Literature Search Strategy

Databases Searched:

IEEE Xplore
ACM Digital Library
arXiv
PubMed (for healthcare applications)
SpringerLink
Google Scholar

Search Query:

("explainable AI" OR "interpretable machine learning" OR "XAI" OR
 "model interpretability" OR "transparency" OR "SHAP" OR "LIME")
AND ("deep learning" OR "neural networks" OR "machine learning")

Inclusion Criteria:

Published between 2016-2024
Peer-reviewed papers or high-quality preprints
Focus on technical methods or applications of XAI
Written in English

Exclusion Criteria:

Pure theoretical papers without practical relevance
Duplicate publications
Papers focused solely on ethics without technical content

Data Extraction & Analysis

Extracted information for each paper:

XAI technique(s) used
Application domain
Model types explained
Evaluation metrics
Key findings and limitations
Citation count and impact

Key Contributions

1. Cognitively Grounded Taxonomy of Explanation Strategies

This survey introduces a novel human-centered taxonomy of explanation types aligned with cognitive reasoning models:

Analogical Explanations

Leverage familiar examples to explain unfamiliar concepts
Example: “This credit risk is similar to cases X, Y, Z”
Aligned with case-based reasoning in human cognition

Contrastive Explanations

Answer “Why P rather than Q?” questions
Counterfactual reasoning: “If feature X changed, outcome would be Y”
Natural for human causal thinking

Conceptual Explanations

High-level abstraction through learned concepts
Example: “Detected ‘stripedness’ and ‘four legs’ → classified as zebra”
Maps to prototype theory in psychology

Narrative Explanations

Sequential, story-like explanations of decision processes
Temporal causality and process transparency
Aligns with human preference for narrative structure

Interactive Explanations

User-guided, iterative exploration
Adaptive to user expertise and information needs
Supports active learning and verification

2. Lifecycle-Centric XAI Architecture

Four interdependent layers spanning the entire ML lifecycle:

Operational Layer

Data collection, preprocessing, feature engineering
Model training, validation, deployment
Embedded explainability considerations at each stage

Explainability Layer

Technique selection based on model type and stakeholder needs
Implementation of explanation generation
Integration with operational processes

Interactivity Layer

User interface design for explanation consumption
Feedback mechanisms and refinement
Personalization and adaptation

Governance Layer

Regulatory compliance (GDPR, EU AI Act)
Audit trails and documentation
Fairness, accountability, transparency monitoring

3. Comprehensive XAI Technique Benchmark

17 XAI techniques evaluated across multiple modalities using lifecycle-aware metrics:

Tabular Data Methods

SHAP (SHapley Additive exPlanations)
LIME (Local Interpretable Model-agnostic Explanations)
Integrated Gradients
Counterfactual Explanations
Anchors

Image Data Methods

GradCAM (Gradient-weighted Class Activation Mapping)
LIME for Images
Layer-wise Relevance Propagation (LRP)
Attention Visualization
Concept Activation Vectors (CAVs)

Text Data Methods

Attention Weights
LIME for Text
Integrated Gradients for NLP
Influence Functions
Rationale Extraction

Evaluation Metrics

Lifecycle-Aware Metrics:

Fidelity: How accurately explanations reflect true model behavior
Completeness: Coverage of relevant features/factors
Monotonicity: Consistency with feature importance rankings
Stability: Robustness to input perturbations
Complexity: Cognitive load and interpretability

Cognitively Aligned Metrics:

Human comprehensibility scores
Task completion time with explanations
Decision confidence improvement
Trust calibration

4. Application Domains Analysis

Distribution of XAI Applications:

Healthcare & Medical Diagnosis: 28%
Finance & Risk Assessment: 18%
Autonomous Systems: 15%
Natural Language Processing: 14%
Computer Vision: 12%
Legal & Compliance: 7%
Other: 6%

Key Insight: Healthcare dominates due to regulatory requirements and high-stakes decision-making.

3. Evolution of XAI Research

Phase 1 (2016-2018): Foundation

Introduction of LIME and SHAP
Focus on post-hoc explanation methods
Limited evaluation frameworks

Phase 2 (2019-2021): Expansion

Domain-specific applications increase
Attention to evaluation metrics
Emergence of counterfactual explanations

Phase 3 (2022-2024): Maturation

Integration with foundation models (LLMs)
Causal interpretability gains traction
Human-centered evaluation becomes standard
Regulatory compliance drives adoption

4. Evaluation Metrics Landscape

Identified five categories of XAI evaluation:

Fidelity: How accurately explanations reflect model behavior
- Local fidelity scores
- Global consistency metrics
Interpretability: Human comprehensibility
- User study ratings
- Cognitive load measurements
Stability: Robustness to input perturbations
- Lipschitz continuity
- Adversarial stability tests
Actionability: Usefulness for decision-making
- Counterfactual proximity
- Feature actionability scores
Fairness: Bias detection and mitigation
- Disparate impact analysis
- Group fairness metrics

Critical Gap: Only 34% of papers include human evaluation; most rely solely on computational metrics.

Comprehensive Analysis

Strengths by Technique

SHAP (SHapley Additive exPlanations)

Strong theoretical foundation (game theory)
Consistent across different model types
Widely adopted in industry
Limitation: Computationally expensive for large datasets

LIME (Local Interpretable Model-agnostic Explanations)

Model-agnostic and flexible
Easy to implement and understand
Limitation: Unstable explanations for similar inputs

Gradient-Based Methods

Computationally efficient
Direct access to model internals
Limitation: Limited to differentiable models

Counterfactual Explanations

Highly actionable for users
Natural for human reasoning
Limitation: May suggest unrealistic changes

Challenges Identified

Evaluation Standardization: No consensus on what makes a “good” explanation
Faithfulness vs. Plausibility: Trade-off between accurate model representation and human comprehension
Computational Cost: Many methods don’t scale to large models or datasets
Adversarial Robustness: Explanations can be manipulated
Multi-Stakeholder Needs: Different users need different explanation types

Research Gaps & Opportunities

Identified Gaps

Limited Work on Foundation Models: Only 12% of papers address LLM interpretability
Lack of Causal Methods: Causality-based XAI represents only 8% of literature
Temporal Explanations: Time-series model explanations underexplored
Interactive Explanations: Few systems allow user-guided explanation refinement
Cultural Context: Almost no work on cross-cultural explanation preferences

Future Research Directions

XAI for Generative AI: Explaining outputs of diffusion models, LLMs, GANs
Causal XAI: Moving beyond correlations to causal explanations
Multi-Modal Explanations: Combining visual, textual, and interactive elements
Personalized Explanations: Adapting to user expertise and needs
Regulatory Compliance: Methods specifically designed for GDPR, EU AI Act requirements

Practical Recommendations

For Practitioners

Choose methods based on use case:
- High-stakes decisions → Use multiple complementary methods
- Real-time systems → Prefer computationally efficient methods
- Non-technical users → Focus on example-based explanations
Always include human evaluation for user-facing systems
Document limitations of chosen XAI techniques transparently
Test stability before deployment to avoid contradictory explanations

For Researchers

Standardize evaluation: Use established benchmarks and include multiple metric types
Consider stakeholder diversity: Design evaluations with actual end-users
Address scalability: Develop methods that work with modern large models
Explore causality: Integrate causal reasoning into XAI frameworks

Methodology Contributions

Created open-source tools for literature analysis:

class XAILiteratureSurvey:
    def __init__(self):
        self.papers = []
        self.taxonomy = TaxonomyBuilder()
        self.analyzer = TrendAnalyzer()

    def extract_insights(self):
        # Extract key information from papers
        techniques = self.extract_techniques()
        domains = self.extract_domains()
        metrics = self.extract_evaluation_metrics()

        # Analyze trends over time
        trends = self.analyzer.temporal_analysis(self.papers)

        # Build co-citation network
        network = self.build_citation_network()

        return {
            'taxonomy': self.taxonomy.build(techniques),
            'trends': trends,
            'network': network
        }

Visualizations & Outputs

Created comprehensive visualizations:

XAI Techniques Timeline: Evolution of methods from 2016-2024
Citation Network: Influential papers and their relationships
Domain Application Heatmap: Technique usage across domains
Evaluation Metrics Distribution: Gap analysis of evaluation practices
Research Trend Analysis: Emerging topics via topic modeling

Survey Statistics

Total Papers Reviewed: 202 peer-reviewed studies
XAI Techniques Benchmarked: 17 methods across 3 modalities
Explanation Strategy Types: 5 cognitively grounded categories
Architecture Layers: 4 lifecycle-centric layers
Application Domains Covered: Multiple high-stakes domains
Evaluation Dimensions: Lifecycle-aware + cognitively aligned metrics

Impact & Dissemination

Publications

Published: ACM Computing Surveys, Vol. 58, No. 12, Article 305 (May 2026)
Co-authors: Ambreen Hanif, Radwa El Shawi, Amin Beheshti, Boualem Benatallah
Pages: 42 pages
DOI: 10.1145/3806829

Community Contributions

Open Dataset: Curated bibliography with extracted metadata
Interactive Visualization: Web-based explorer of XAI landscape
Taxonomy Framework: Structured classification system for XAI methods
GitHub Repository: Analysis code and replication materials

Practical Impact

Used by 3 companies to select appropriate XAI methods
Cited in 2 regulatory consultation responses
Integrated into graduate-level XAI course curriculum

Lessons Learned

Interdisciplinary Nature: XAI requires balancing ML, HCI, psychology, and domain expertise
No One-Size-Fits-All: Different contexts require different explanation approaches
Evaluation is Hard: Human evaluation is essential but challenging to conduct rigorously
Research Velocity: Field evolving rapidly; surveys need regular updates
Practice-Research Gap: Industry needs often differ from academic focus areas

Future Work

Annual Updates: Plan to maintain living survey with yearly updates
Domain-Specific Deep Dives: Focused surveys on healthcare XAI, finance XAI
Comparative Empirical Study: Benchmark top techniques on standardized datasets
User Preference Study: Large-scale investigation of explanation preferences across demographics

Resources

Survey Paper: ACM Computing Surveys PDF: Download PDF Interactive Explorer: [xai-survey.ambreenhanif.com] GitHub Repository: [github.com/umberH/xai-literature-survey] Dataset: [Curated bibliography in BibTeX and CSV formats] Slides: [Presentation deck available on request]

Citation

@article{10.1145/3806829,
  author = {Hanif, Ambreen and Shawi, Radwa El and Beheshti, Amin and Benatallah, Boualem},
  title = {Survey on Explainable AI for Traditional Machine Learning and Domains},
  year = {2026},
  issue_date = {September 2026},
  publisher = {Association for Computing Machinery},
  address = {New York, NY, USA},
  volume = {58},
  number = {12},
  issn = {0360-0300},
  url = {https://doi.org/10.1145/3806829},
  doi = {10.1145/3806829},
  journal = {ACM Comput. Surv.},
  month = {may},
  articleno = {305},
  numpages = {42},
  keywords = {Explainable AI, cognitive alignment, human-centered explainability, evaluation metrics, applications}
}

Technologies Used