Regularization of RNN Through Bayesian Networks
1:10 - 1:40 | Friday, May 24
Data Scientists are often subjected to wide array of data patterns and signatures. While they analyze the data in the light of a particular problem, they often face challenges when data signatures do not yield to patterns they are expecting. At this point, data scientist must improvise their techniques that are custom to the problem at hand.
While Deep Learning has shown significant promise towards model performance, it can quickly become untenable particularly when data size falls short of problem space. One such situation regularly appears when modeling with RNNs. RNNs can quickly memorize and over-fit (the problem is further aggravated when data size is small to medium). The presentation exposes shortcomings of RNNs and how a combination of RNNs and Bayesian Network (PGM) can not only overcome this shortcoming but also improvise sequence-modeling behavior of RNNs. We will learn this in the context of Marketing Channel Attribution modeling.
Principal Data Scientist, Vanguard
Vishal (‘Vish’) Hawa is Principal Data Scientist at Vanguard. Vish has over 15 years of experience in Retail and Financial services industry and works closely with Marketing Managers in designing attribution, propensity and attrition modeling.
Vish has executive management from Wharton school of business, post-graduation degrees in Information sciences, Statistics and computer engineering from Indian Statistical Institute.