Ecological Neural Architecture Search
It addresses the problem of manual hyperparameter tuning for evolutionary algorithms in neural architecture search, offering an efficient, self-regulating approach that is incremental but provides specific gains.
This paper tackles the challenge of tuning evolutionary algorithm hyperparameters in neural architecture search by introducing Neuvo Ecological Neural Architecture Search (ENAS), which integrates these parameters into candidate solutions for dynamic evolution, resulting in reduced computational time by 18.3% and improved accuracy in 3 out of 4 datasets.
When employing an evolutionary algorithm to optimize a neural networks architecture, developers face the added challenge of tuning the evolutionary algorithm's own hyperparameters - population size, mutation rate, cloning rate, and number of generations. This paper introduces Neuvo Ecological Neural Architecture Search (ENAS), a novel method that incorporates these evolutionary parameters directly into the candidate solutions' phenotypes, allowing them to evolve dynamically alongside architecture specifications. Experimental results across four binary classification datasets demonstrate that ENAS not only eliminates manual tuning of evolutionary parameters but also outperforms competitor NAS methodologies in convergence speed (reducing computational time by 18.3%) and accuracy (improving classification performance in 3 out of 4 datasets). By enabling "greedy individuals" to optimize resource allocation based on fitness, ENAS provides an efficient, self-regulating approach to neural architecture search.