View Our Recent TechTalk
"Learning Symbolic Equations with Deep Learning" with Shirley Ho
We develop a general approach to distill symbolic representations of a learned deep model by introducing strong inductive biases. We focus on Graph Neural Networks (GNNs). The technique works as follows: we first encourage sparse latent representations when we train a GNN in a supervised setting, then we apply symbolic regression to components of the learned model to extract explicit physical relations. We find the correct known equations, including force laws and Hamiltonians, can be extracted from the neural network. We then apply our method to a non-trivial cosmology example--a detailed dark matter simulation--and discover a new analytic formula which can predict the concentration of dark matter from the mass distribution of nearby cosmic structures. The symbolic expressions extracted from the GNN using our technique also generalized to out-of-distribution data better than the GNN itself. Our approach offers alternative directions for interpreting neural networks and discovering novel physical principles from the representations they learn.
ACM award winners, leading researchers, industry veterans, thought leaders, and innovators address today and tomorrow's hottest topics and issues in computing for busy practitioners, as well as educators, students, and researchers. Check out our archive of these ACM TechTalks, free for members and non-members alike.
Talks from some of the leading visionaries and bleeding-edge researchers in AI/ML: Fei-Fei Li on visual intelligence in computers and ImageNet; Eric Horvitz on AI solutions in the open world; and Tom Mitchell on using ML to study how the brain creates and represents language.
View Latest Learning Center TechTalks
View the recent ACM TechTalk, "Horizontal Leadership: Practical Lessons for Driving Company-Wide Strategy and Action," presented by Eve Andersson, Senior Director, Accessibility, at Google, and a member of the ACM Practitioner Board. Eve also co-founded ArsDigita Corporation, an open-source software company, and co-authored two books: Software Engineering for Internet Applications (MIT Press, 2006) and Early Adopter VoiceXML (Wrox Press, 2001). Vicki Hanson, ACM CEO will moderate the questions and answers section. Continue the discussion on ACM's Discourse Page.
View the recent ACM TechTalk, "Democratizing AI: Creating Cognitive AI Assistants with No Coding," presented by Michelle Zhou, CEO of Juji, Inc., ACM Distinguished Member, and Editor in Chief of ACM Transactions on Interactive Intelligent Systems (TiiS). Wenxi Chen, AI Software Engineer at Juji, Inc., moderated the questions and answers session. Continue the discussion on ACM's Discourse Page.