Towards Secure and Interpretable AI: Scalable Methods, Interactive Visualizations, and Practical Tools


August 13, 2019


Polo Chau


Georgia Tech


The explosion of available idea repositories — scientific papers, patents, product descriptions — represents an unprecedented opportunity to accelerate innovation and lead to a wealth of discoveries. Given the scale of the problem and its ever-expanding nature, there is a need for intelligent automation to assist in the process of discovery. In this talk, I will present our work toward addressing this challenging problem.

We developed an approach for boosting people’s creativity by helping them discover analogies — abstract structural connections between ideas. We learn to decompose innovation texts into functional models that describe the components and goals of inventions, and use them to build a search engine supporting expressive inspiration queries. In ideation studies, our inspirations helped people generate better ideas with significant improvement over standard search. We also construct a commonsense ontology of purposes and mechanisms of products, mapping the landscape of ideas.

I will also describe a novel machine learning framework we developed in order to identify innovation in patents, where labels are extremely hard to obtain. In our setting, called Ballpark Learning, we are only given groups of instances with coarse constraints over label averages. We demonstrate encouraging results in classification and regression tasks across several domains.



Polo Chau

Polo Chau is an Associate Professor of Computing at Georgia Tech. He co-directs Georgia Tech’s MS Analytics program. His research group bridges machine learning and visualization to synthesize scalable interactive tools for making sense of massive datasets, interpreting complex AI models, and solving real world problems in cybersecurity, human-centered AI, graph visualization and mining, and social good. His Ph.D. in Machine Learning from Carnegie Mellon University won CMU’s Computer Science Dissertation Award, Honorable Mention. He received awards and grants from NSF, NIH, NASA, DARPA, Intel (faculty PI of ISTC-ARSA), Symantec, Google, Nvidia, IBM, Yahoo, Amazon, Microsoft, eBay, LexisNexis; Raytheon Faculty Fellowship; Edenfield Faculty Fellowship; Outstanding Junior Faculty Award; The Lester Endowment Award; Symantec fellowship (twice); Best student papers at SDM’14 and KDD’16 (runner-up); Best demo at SIGMOD’17 (runner-up); Chinese CHI’18 Best paper. His research led to open-sourced or deployed technologies by Intel (for ISTC-ARSA: ShapeShifter, SHIELD, ADAGIO, MLsploit), Google, Facebook, Symantec (Polonium, AESOP protect 120M people from malware), and Atlanta Fire Rescue Department. His security and fraud detection research made headlines.