Pruning-Based TinyML Optimization of Machine Learning Models for Anomaly Detection in Electric Vehicle Charging Infrastructure

2025-03-19

AI Summary

To address real-time anomaly detection on resource-constrained edge devices in electric vehicle charging infrastructure (EVCI), this paper proposes a lightweight modeling approach that jointly optimizes interpretability, efficiency, and accuracy. The method innovatively integrates SHAP-based feature selection with unstructured pruning for time-series anomaly detection, and further enhances compression via Optuna-driven hyperparameter optimization and collaborative model distillation across MLP, LSTM, and XGBoost architectures. Evaluated on the CICEVSE2024 dataset, the resulting model achieves a 72% reduction in model size and a 68% decrease in inference latency, while sustaining AUC degradation of less than 0.01—demonstrating substantial improvements in edge deployability and detection robustness without compromising diagnostic reliability.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

With the growing need for real-time processing on IoT devices, optimizing machine learning (ML) models' size, latency, and computational efficiency is essential. This paper investigates a pruning method for anomaly detection in resource-constrained environments, specifically targeting Electric Vehicle Charging Infrastructure (EVCI). Using the CICEVSE2024 dataset, we trained and optimized three models-Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), and XGBoost-through hyperparameter tuning with Optuna, further refining them using SHapley Additive exPlanations (SHAP)-based feature selection (FS) and unstructured pruning techniques. The optimized models achieved significant reductions in model size and inference times, with only a marginal impact on their performance. Notably, our findings indicate that, in the context of EVCI, pruning and FS can enhance computational efficiency while retaining critical anomaly detection capabilities.

Problem

• Optimize ML models for real-time IoT anomaly detection.
• Reduce model size and latency in EV charging infrastructure.
• Enhance computational efficiency with pruning and feature selection.

Innovation

• Pruning reduces model size and latency.
• SHAP-based feature selection enhances efficiency.
• Optimized models maintain anomaly detection accuracy.