Robust Classifiers, Domains, and Training Loss
Walt Woods discusses the application of the Leapshift constraint to different domains, such as audio and tabular data, and the potential algorithmic changes required. He also expresses curiosity about the challenge of creating classifiers that are both accurate and robust, and explores the reasons behind the training loss never reaching zero.In this clip
From this podcast
Data Skeptic
Adversarial Explanations
Related Questions