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 look at the authors code, if released, in order to prevent bias toward discrepancies between code and paper.
Edward Raff (Booz Allen Hamilton)
Related Events (a corresponding poster, oral, or spotlight)
2019 Poster: A Step Toward Quantifying Independently Reproducible Machine Learning Research »
Thu Dec 12th 05:00 -- 07:00 PM Room East Exhibition Hall B + C