Black Magic in Deep Learning: How Human Skill Impacts Network Training

TL;DR

This study explores how human experience influences deep learning hyperparameter optimization, revealing that experienced users achieve better results more efficiently, highlighting the importance of human factors in AI research reproducibility.

Contribution

It provides empirical evidence that user experience significantly affects hyperparameter tuning outcomes in deep learning.

Findings

Experienced participants achieve higher accuracy.

Fewer resources are needed by experienced users.

Inexperienced users tend to follow random strategies.

Abstract

How does a user's prior experience with deep learning impact accuracy? We present an initial study based on 31 participants with different levels of experience. Their task is to perform hyperparameter optimization for a given deep learning architecture. The results show a strong positive correlation between the participant's experience and the final performance. They additionally indicate that an experienced participant finds better solutions using fewer resources on average. The data suggests furthermore that participants with no prior experience follow random strategies in their pursuit of optimal hyperparameters. Our study investigates the subjective human factor in comparisons of state of the art results and scientific reproducibility in deep learning.