Deep Learning for the Travel Industry
This presentation discovers some aspects of working with low examples training data sets of manmade objects and other famous tourist attractions with classifier adaptation to user generated pictures. For example choosing alternative last layer for Inception v3 convolutional net. Also a number of aspects of building the dialog systems for travel industry will be discovered.
Tom is currently a data science team lead at Find Attractions Inc (Travel Lisa). He uses Deep Convolutional Nets (CNN's) to classify user generated images of city objects with small training data samples and Recurrent Neural Nets for automatic image captioning with style change. Prior to that he was making a quantitative research at University College London and Higher School of Economics Moscow, implementing Vector Auto Regressions to analyze dynamic equilibrium on commodities and stock exchange market.