Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)

Abstract

One of the challenges in machine learning research is to ensure that presented and published results are sound and reliable. Reproducibility, that is obtaining similar results as presented in a paper or talk, using the same code and data (when available), is a necessary step to verify the reliability of research findings. Reproducibility is also an important step to promote open and accessible research, thereby allowing the scientific community to quickly integrate new findings and convert ideas to practice. Reproducibility also promotes the use of robust experimental workflows, which potentially reduce unintentional errors. In 2019, the Neural Information Processing Systems (NeurIPS) conference, the premier international conference for research in machine learning, introduced a reproducibility program, designed to improve the standards across the community for how we conduct, communicate, and evaluate machine learning research. The program contained three components: a code submission policy, a community-wide reproducibility challenge, and the inclusion of the Machine Learning Reproducibility checklist as part of the paper submission process. In this paper, we describe each of these components, how it was deployed, as well as what we were able to learn from this initiative.

arXiv commentTo appear at JMLR, 16 pages + Appendix

References in corpus (2)

• A Step Toward Quantifying Independently Reproducible Machine Learning Research
• The Scientific Method in the Science of Machine Learning

Cited by in corpus (31)

• How to avoid machine learning pitfalls: a guide for academic researchers
• Automated Evolutionary Approach for the Design of Composite Machine Learning Pipelines
• A curated dataset for data-driven turbulence modelling
• On the Value of Out-of-Distribution Testing: An Example of Goodhart's Law
• Deep Reinforcement Learning at the Edge of the Statistical Precipice
• A Scoping Study of Evaluation Practices for Responsible AI Tools: Steps Towards Effectiveness Evaluations
• The Multi-Agent Behavior Dataset: Mouse Dyadic Social Interactions
• Towards a consistent interpretation of AIOps models
• Who Is Alyx? A new Behavioral Biometric Dataset for User Identification in XR
• Between Subjectivity and Imposition: Power Dynamics in Data Annotation for Computer Vision
• DaisyRec 2.0: Benchmarking Recommendation for Rigorous Evaluation
• Experience Replay with Likelihood-free Importance Weights
• Synthetic Benchmarks for Scientific Research in Explainable Machine Learning
• Dynaboard: An Evaluation-As-A-Service Platform for Holistic Next-Generation Benchmarking
• Supervised machine learning for microbiomics: bridging the gap between current and best practices
• Navigating the challenges in creating complex data systems: a development philosophy
• FAIR AI Models in High Energy Physics
• The NCI Imaging Data Commons as a platform for reproducible research in computational pathology
• LimeSoDa: A Dataset Collection for Benchmarking of Machine Learning Regressors in Digital Soil Mapping
• Neural Temporal Point Processes For Modelling Electronic Health Records
• Using Unity to Help Solve Intelligence
• MLDev: Data Science Experiment Automation and Reproducibility Software
• Perspectives on Machine Learning from Psychology's Reproducibility Crisis
• Benchmarking the Accuracy and Robustness of Feedback Alignment Algorithms
• Larger-Context Tagging: When and Why Does It Work?
• An Experience Report on Machine Learning Reproducibility: Guidance for Practitioners and TensorFlow Model Garden Contributors
• STAMP 4 NLP -- An Agile Framework for Rapid Quality-Driven NLP Applications Development
• The MICCAI Hackathon on reproducibility, diversity, and selection of papers at the MICCAI conference
• Occlusion Aware Kernel Correlation Filter Tracker using RGB-D
• Steps Towards Best Practices For Robot Videos
• Navigating the Landscape of Reproducible Research: A Predictive Modeling Approach