This talk is about our paper accepted at the 2024 NeurIPS Datasets and Benchmarks track.
Our paper provides an analysis of dataset development practices at NeurIPS through the lens of data curation.
We present an evaluation framework for dataset documentation and we use the framework to assess the strengths and weaknesses
in current dataset development practices of 60 datasets published in the NeurIPS Datasets and Benchmarks track from 2021-2023.
Our paper, discussed in this talk, lays out how future research on the impacts of alternative design approaches on psychological distance can make data used for policy decisions more tangible and visceral. It was accepted at the Tenth Workshop on Computing within Limits (LIMITS 2024).
This talk presents our FAccT 2024 paper in which we explored how the dataset development process within ML research can be made more transparent and accountable by applying a data curation lens.
I was granted a competitive travel award to attend the Research Data Alliance 20th Plenary Meeting in 2023. As part of this, I participated in an interview for early career researchers in data management.
Supporting Responsible Machine Learning by Improving Data Curation
DECEMBER 2024
Bhardwaj, E. & Becker, C.
Abstract submission and poster presentation at the 19th Women in Machine Learning (WiML) Workshop at NeurIPS.
