İçeriğe geç

Beyond Accuracy in AI: A Multi-Objective Benchmark of Machine Learning Models

Beyond Accuracy in AI: A Multi-Objective Benchmark of Machine Learning Models

How should machine learning models be selected when predictive accuracy is not the only priority?

This study presents a multi-objective benchmarking framework comparing eight machine learning classifiers across four XOR-based scenarios: clean, noisy, rotated, and high-dimensional datasets.

Based on 640 controlled experiments, the research evaluates predictive accuracy, computational efficiency, training and inference time, memory consumption, energy usage, and model size.

Through Pareto frontier analysis, the study demonstrates that the most accurate model is not necessarily the most resource-efficient solution.

The findings provide practical insights into resource-aware machine learning, computational efficiency, model robustness, and multi-objective AI optimization.

Keywords: Multi-Objective Machine Learning, Resource-Aware AI, Computational Efficiency, Energy-Efficient Machine Learning, Inductive Bias, Pareto Optimization, Nonlinear Classification.

Published in: Applied Sciences (2026), 16(10), 4637.

DOI: https://doi.org/10.3390/app16104637

Tarih:Genel

İlk Yorumu Siz Yapın

Bir cevap yazın

E-posta hesabınız yayımlanmayacak. Gerekli alanlar * ile işaretlenmişlerdir

+ 67 = 76