Corpus ID: 202577933

A Step Toward Quantifying Independently Reproducible Machine Learning Research

@inproceedings{Raff2019AST,
  title={A Step Toward Quantifying Independently Reproducible Machine Learning Research},
  author={Edward Raff},
  booktitle={NeurIPS},
  year={2019}
}
  • Edward Raff
  • Published in NeurIPS 2019
  • Computer Science, Mathematics
  • What makes a paper independently reproducible? Debates on reproducibility center around intuition or assumptions but lack empirical results. Our field focuses on releasing code, which is important, but is not sufficient for determining reproducibility. We take the first step toward a quantifiable answer by manually attempting to implement 255 papers published from 1984 until 2017, recording features of each paper, and performing statistical analysis of the results. For each paper, we did not… CONTINUE READING

    Create an AI-powered research feed to stay up to date with new papers like this posted to ArXiv

    Figures, Tables, and Topics from this paper.

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 17 REFERENCES

    Artificial intelligence faces reproducibility crisis.

    VIEW 2 EXCERPTS

    JASP (Version 0.9)[Computer software

    • JASP Team
    • 2018. [Online]. Available: https://jasp-stats.org/ 10
    • 2018
    VIEW 1 EXCERPT

    JSAT: Java Statistical Analysis Tool, a Library for Machine Learning

    VIEW 2 EXCERPTS

    NIPS 2017 Test-of-time award presentation

    • A. Rahimi, B. Recht
    • 2017.
    • 2017
    VIEW 1 EXCERPT

    Replicability Analysis for Natural Language Processing: Testing Significance with Multiple Datasets

    VIEW 1 EXCERPT