Rackham Faculty Allies Speaker Series: Nina Mishra

Title:  Time-Critical Machine Learning

Abstract: How can we detect an arrhythmia in a continuous electrocardiogram in real time?  Can we use the same approach to find dips in e-commerce sales, or denial of service attacks from IP-IP communication data?  We describe algorithms that can spot and label events in continuously arriving data.  The key insight is maintaining a small synopsis or sketch that can identify/label events and, simultaneously, also be quickly updated as new data arrives and old data fades away.  We tell a story through real, public data.

Bio:  Nina Mishra has been a Principal Scientist at Amazon for over four years.  With an eye towards people, she identifies pain points that, if solved, could impact many.  She approaches problems from a computational lens, drawing upon tools from machine learning, data mining and CS algorithms.  Her ML work is publicly available in AWS (Amazon Web Services).  She previously worked at Microsoft Research Silicon Valley on altering web search in life-critical situations.  Her academic experience includes Associate Professor at the University of Virginia and Acting Faculty at Stanford University.  She served as Program Co-Chair for ICML and also served on journal editorial boards such as Machine Learning, IEEE Transactions on Knowledge and Data Engineering, and IEEE Intelligent Systems.