Rigorous Techniques in Data Science
Sebastian discusses the current state of rigorous techniques in data science and the potential for more principled approaches. He explores the automation of feature engineering with Neural Networks and the recent development of learning the architecture. Sebastian also highlights the importance of generating and evaluating pretrained tasks for meta learning, emphasizing the need for access to a diverse distribution of tasks for effective model training.In this clip
From this podcast
Data Skeptic
Transfer Learning
Related Questions