After clicking “Watch Now” you will be prompted to login or join.
Insight-Driven Machine Learning Design with Human Expert Collaborations
Maithra Raghu, Google Brain
GTC 2020
The fundamental breakthroughs in machine learning, and the rapid advancements of the underlying deep neural network models have enabled the potential use of these systems in specialized, high stakes domains such as healthcare. However, despite this remarkable progress, the design of machine learning systems remains laborious, computationally expensive and opaque, sometimes resulting in catastrophic failures and significantly hindering their ability to work with human experts, who play critical roles in these settings. In this talk, I overview steps towards a more informed, intuitive design of machine learning systems, and methods to facilitate collaboration with human experts. I develop tools that enable the quantitative analysis of the complex hidden layers of deep neural networks, which in turn provides both fundamental insights and informs algorithms for efficiently training these systems. I demonstrate how these trained systems can be adapted to work effectively with human experts, resulting in better outcomes than either entity alone.