Algorithmic Bias Insights
Rishabh discusses how algorithmic bias can lead to underestimations of performance, particularly due to poorly tuned hyperparameters. He highlights significant fluctuations in results based on the number of seeds used, raising questions about the validity of conclusions drawn from existing methods. The findings reveal that many published results may have been overestimated, emphasizing the need for careful benchmarking in machine learning research.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Deep Reinforcement Learning at the Edge of the Statistical Precipice with Rishabh Agarwal - #559
Related Questions