Synergistic Integration of Neural Architecture Search, Ensemble Learning, and Meta-Learning with Explainable AI

Migunani Migunani; Joseph Teguh Santoso; Budi  Raharjo

doi:10.51903/jtie.v5i1.207

Authors

Migunani Migunani Universitas Sains dan Teknologi Komputer, Semarang, Indonesia https://orcid.org/0000-0002-8551-2157
Joseph Teguh Santoso Universitas Sains dan Teknologi Komputer, Semarang, Indonesia https://orcid.org/0000-0001-6227-1111
Budi Raharjo Universitas Sains dan Teknologi Komputer, Semarang, Indonesia https://orcid.org/0009-0005-3425-5479

DOI:

https://doi.org/10.51903/jtie.v5i1.207

Keywords:

Explainable AI, Ensemble Learning, Heterogeneous Ensemble, Meta Learning, Neural Architecture Search, SHAP

Abstract

The automation of deep learning model design through Neural Architecture Search (NAS) has emerged as a transformative paradigm, yet the integration of NAS with ensemble learning and meta-learning remains underexplored. This study presents a comprehensive framework that synergistically combines (1) NAS for automated neural network architecture discovery, (2) homogeneous ensemble learning through NAS-generated architectures, (3) heterogeneous ensemble learning integrating diverse base learners, and (4) meta-learning with stacked generalization for adaptive model fusion, all complemented by Explainable AI (XAI) via SHAP for model interpretability. The proposed framework is evaluated on four benchmark datasets: Iris, Wine, Breast Cancer, and Digits. Experimental results demonstrate that the heterogeneous ensemble achieves competitive or superior performance across all datasets, with cross-validation accuracy reaching 98.87% (wine), 97.72% (breast cancer), 98.05% (digits), and 95.33% (iris). SHAP-based explainability analysis reveals consistent feature importance patterns across NAS architectures, providing valuable interpretability insights. The framework establishes a robust pipeline for automated, accurate, and interpretable machine learning, addressing the growing demand for transparent AI systems in critical applications.

References

Atlas, L., & Devi, A. S. (2025). Neural Architecture Search (NAS): A survey of methods, challenges, and future directions. Semantic Scholar, September 2025.

Chen, Y., et al. (2025). AgenticRS-EnsNAS: Ensemble-decoupled self-evolving architecture search. arXiv:2603.20014.

Ensemble Learning: Importance, applications, challenges, and future work. (2025). IEEE Xplore, March 2025.

ExplaiLiver+. (2025). Multi-stage stacking framework with SHAP-based interpretability for clinical prediction. Frontiers, 16.

Foundations and innovations in data fusion and ensemble learning for effective consensus. (2025). DOAJ, February 2025.

Garouani, M., Barhrhouj, A., & Teste, O. (2025). XStacking: An effective and inherently explainable framework for stacked ensemble learning. Information Fusion, 124, 103358.

IELF (Interpretable Ensemble Learning Framework). (2025). Heart disease prediction with Explainable Boosting Machines and SHAP. Frontiers, 16.

Latent SHAP. (2025). Toward practical human-interpretable explanations. Machine Learning, August 2025.

Li, J., et al. (2025). Few shot crop disease recognition using sequence-weighted ensemble model agnostic meta learning. Frontiers in Plant Science, 16, 1615873.

Lin, Y., et al. (2021). LENAS: Learning based Neural Architecture Search and Ensemble for 3D Radiotherapy Dose Prediction. arXiv:2106.06733.

Meta ensemble learning. (2025). Multi headed models for few shot problems. ACAI, December 2025.

MetaStackD. (2025). A robust meta learning based deep ensemble model for prediction of sensors battery life. Scientific Reports, April 2025.

Rong, J., et al. (2021). Across Task Neural Architecture Search via Meta Learning. arXiv:2110.05842.

Salmani Pour Avval, S., Eskue, N. D., & Groves, R. M. (2025). Systematic review on neural architecture search. Artificial Intelligence Review, 58(3).

Sedek, A. (2025). Dynamic Meta Learning for Adaptive XGBoost Neural Ensembles. arXiv:2510.03301.

SEENAS (Semi Dynamic Ensemble for Evolutionary NAS). (2025). IEEE Xplore, December 2025.

SHAP based Explainable AI for dropout prediction. (2025). IEEE Xplore, December 2025.

Verma, S., Kumar, P., & Singh, J. P. (2025). MLP GNAS: Meta learning based predictor assisted Genetic Neural Architecture Search system. Applied Soft Computing, 112527.

BossNAS: Block wisely Self supervised Neural Architecture Search. (2025). opus.lib.uts.edu.au, July 2025.

Adaptive deep meta learning ensembles for robust daily carbon emission forecasting. (2025). ScienceDirect, October 2025.

Heterogeneous ensemble learning: modified ConvNextTiny for detecting molecular expression of breast cancer. (2025). Frontiers, June 2025.

An ensemble heterogeneous transformer model for an effective diagnosis of multiple plant diseases. (2025). Frontiers, November 2025.

Sensitivity Analysis of Neural Network Ensembles to Data Quality Issues. (2025). IEEE Xplore, September 2025.

A meta learning ensemble framework for robust and interpretable prediction of emergency medical services demand. (2025). PubMed, December 2025.

One Shot Neural Ensemble Architecture Search by Diversity Guided Search Space Shrinking. (2025). arXiv.

Multi headed Neural Ensemble Search. (2025). arXiv.

EvoREP: Evolving Reliable Ensemble of Proxies for Zero Shot Neural Architecture Search. (2025). IEEE Xplore, December 2025.

GreenFactory: Ensembling Zero Cost Proxies to Estimate Performance of Neural Networks. (2025). scirate.com, May 2025.

MedNNS: Supernet based Medical Task Adaptive Neural Network Search. (2025). MICCAI 2025, pp. 448 458.

AutoMTNAS: Automated meta reinforcement learning on graph tokenization for graph neural architecture search. (2025). Knowledge Based Systems, 310.

Fast Heterogeneous Graph Neural Network Generation via Meta Contrastive Learning. (2025). ScienceDirect, August 2025.

XStacking: Explanation Guided Stacked Ensemble Learning. (2025). Information Fusion, 124.

Metaheuristically Enhanced ANN Based Intrusion Detection System with Explainable AI Integration. (2025). IEEE Xplore, November 2025.

WISE: A Meta Learning Approach for Weighted Instance Specific Ensemble Prediction. (2025). ACAI, December 2025.

Optimizing Stacked Ensemble Machine Learning Models for Accurate Wildfire Severity Mapping. (2025). OUCI.

A novel stacking based ensemble learning framework for thermal error prediction. (2025). ScienceDirect, October 2025.

An Auto Encoder Enhanced Stacked Ensemble for Intrusion Detection in Healthcare Networks. (2025). ScienceDirect, September 2025.

Optuna Optimized Meta Learner and Ensemble Learning Models for Student Performance Prediction. (2025). ScienceDirect, November 2025.

Stacking Ensemble with Boosting and Hyperparameter Optimization for Breast Cancer Classification. (2025). DTIC.

Research progress and future prospects of neural architecture search. (2025). JUSTC, May 2025.

Neural Architecture Search for global multi step Forecasting of Energy Production Time Series. (2025). scirate.com, November 2025.

Improving Novelty Search with a Surrogate Model and Accuracy Objectives to Build High Performing Ensembles. (2025). SN Computer Science, July 2025.

Pre NAS: Efficient Neural Architecture Search. (2025). EmergentMind, September 2025.

Explainable AI using ensemble machine learning with integrated SHAP Borda approach for soil liquefaction estimation. (2025). EBSCO, September 2025.

Integrating Bayesian Optimization into Ensemble Logistic Regression for Explainable AI Based Customer Behavior Analysis. (2025). Semantic Scholar, October 2025.