Big Data & Machine Learning
Fashion search with Deep Learning
ImpactHub - Atlantic room
18th November, 10:10-11:10
This is a technical talk delivered by Zimilar.ro, the Cluj-based startup which is using Deep Learning to offer visual search for online fashion retail. We discuss the emerging field of Deep Neural Networks and how we use them to index, store and retrieve similar fashion items, based solely on their visual aspect. Also we discuss the relatively heavy-weight processing requirements and how to address this with the proper infrastructure, in order to train and use Deep Nets in a real-time environment.
George is a Data Architect at Paddy Power Betfair, with more than 9 years of experience in Data Warehouses and data-driven environments. Equally passionate about Machine Learning, in his spare time he competes on Kaggle competitions. A few months ago, challenged by the idea of improving the online shopping experience by using visual search, he became one of the co-founders of Zimilar, mixing machine learning techniques into an innovative web application.