We have been interviewed at NIPS-2017 on our Best Paper Award by Sam Charrington, the creator and host of This Week in Machine Learning & AI podcast (TWiML & AI).
The paper is one in a series of our works on linear-time hypothesis testing already successfully applied in numerous areas including natural language processing, computer vision and dependency testing of media annotations:
- Two-sample testing (NIPS-2016, Oral): paper, code.
- Independence testing (ICML-2017): paper, code.
- Goodness-of-fit Testing (NIPS-2017, Best Paper Award): paper, code.
The conversation sheds light on the key results, the principles applied in the contribution through an example on criminal data analysis, and how the contribution helps the interpretability and explainability of AI models.