AI methods for Formal Reasoning

Overview of deep learning and AI methods for higher order logic mathematical reasoning.
Who is this presentation for?
Technical audience with research interest AI/deep learning or formal methods background.
Prerequisite knowledge:
Basics of deep learning
What you'll learn?
A novel direction in exploiting current deep learning techniques like graph neural networks and constrastive optiomization to augment sophisticated search processes for practical mathematical reasoning.

Christian Szegedy is a research scientist at Google. He is known for the first paper on adversarial examples, batch-normalization and the Inception architecture for computer vision. He has also worked on pose estimation and object detection. His current work is focused on AI assisted theorem proving, especially formal mathematics in higher order logic.