The Scientific Method in the Science of Machine Learning

Abstract

In the quest to align deep learning with the sciences to address calls for rigor, safety, and interpretability in machine learning systems, this contribution identifies key missing pieces: the stages of hypothesis formulation and testing, as well as statistical and systematic uncertainty estimation -- core tenets of the scientific method. This position paper discusses the ways in which contemporary science is conducted in other domains and identifies potentially useful practices. We present a case study from physics and describe how this field has promoted rigor through specific methodological practices, and provide recommendations on how machine learning researchers can adopt these practices into the research ecosystem. We argue that both domain-driven experiments and application-agnostic questions of the inner workings of fundamental building blocks of machine learning models ought to be examined with the tools of the scientific method, to ensure we not only understand effect, but also begin to understand cause, which is the raison d'être of science.

arXiv comment4 pages + 1 appendix. Presented at the ICLR 2019 Debugging Machine Learning Models workshop

References in corpus (13)

• Confounding variables can degrade generalization performance of radiological deep learning models
• Rivet user manual
• Deep Reinforcement Learning that Matters
• The RooFit toolkit for data modeling
• Simple random search provides a competitive approach to reinforcement learning
• Measuring the Effects of Data Parallelism on Neural Network Training
• ELF OpenGo: An Analysis and Open Reimplementation of AlphaZero
• Deterministic Implementations for Reproducibility in Deep Reinforcement Learning
• Natural Environment Benchmarks for Reinforcement Learning
• Definition and Treatment of Systematic Uncertainties in High Energy Physics and Astrophysics
• Avoiding a Tragedy of the Commons in the Peer Review Process
• Expanding search in the space of empirical ML
• Distilling Information from a Flood: A Possibility for the Use of Meta-Analysis and Systematic Review in Machine Learning Research

Cited by in corpus (11)

• A Survey and Critique of Multiagent Deep Reinforcement Learning
• How I failed machine learning in medical imaging -- shortcomings and recommendations
• Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)
• On the Value of Out-of-Distribution Testing: An Example of Goodhart's Law
• Towards falsifiable interpretability research
• When we can trust computers (and when we can't)
• Learning from learning machines: a new generation of AI technology to meet the needs of science
• Individual predictions matter: Assessing the effect of data ordering in training fine-tuned CNNs for medical imaging
• Showing Your Work Doesn't Always Work
• Expected Validation Performance and Estimation of a Random Variable's Maximum
• Semantic coordinates analysis reveals language changes in the AI field