Phoeni6: a Systematic Approach for Evaluating the Energy Consumption of Neural Networks

2025-02-25

AI Summary

To address the lack of reproducibility and fair benchmarking in neural network energy consumption evaluation, this paper introduces the first standardized energy-efficiency benchmarking system that ensures portability, transparency, and interoperability. Methodologically, it integrates Docker-based containerization, a relational database, a unified energy data model, Python-based automation pipelines, and hardware-level power monitoring interfaces. Key contributions include: (1) a structured, cross-model and cross-dataset evaluation framework guaranteeing full experimental reproducibility; (2) the first systematic empirical demonstration that image format significantly impacts AI inference energy consumption—BMP reduces energy usage by up to 30% compared to PNG; and (3) quantitative validation that MobileNet achieves 2.32%–6.25% lower energy consumption than AlexNet under identical conditions. The system provides a reliable, open, and extensible infrastructure for green AI research.

Technology Category

• Machine Learning: Efficient ML / Green AI
• Computer Vision: Large Vision Models
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements

Abstract

This paper presents Phoeni6, a systematic approach for assessing the energy consumption of neural networks while upholding the principles of fair comparison and reproducibility. Phoeni6 offers a comprehensive solution for managing energy-related data and configurations, ensuring portability, transparency, and coordination during evaluations. The methodology automates energy evaluations through containerized tools, robust database management, and versatile data models. In the first case study, the energy consumption of AlexNet and MobileNet was compared using raw and resized images. Results showed that MobileNet is up to 6.25% more energy-efficient for raw images and 2.32% for resized datasets, while maintaining competitive accuracy levels. In the second study, the impact of image file formats on energy consumption was evaluated. BMP images reduced energy usage by up to 30% compared to PNG, highlighting the influence of file formats on energy efficiency. These findings emphasize the importance of Phoeni6 in optimizing energy consumption for diverse neural network applications and establishing sustainable artificial intelligence practices.

Problem

• Assess neural networks' energy consumption systematically.
• Compare energy efficiency of AlexNet and MobileNet.
• Evaluate image file formats' impact on energy usage.

Innovation

• Automates energy evaluations via containerized tools
• Manages energy data with robust database systems
• Utilizes versatile data models for efficiency