More Like This: similarity-based recommendations

Research Centre: Research Centre for Machine Learning

Abstract:

Targeting the right person is crucial in marketing: a correct recommendation can lead to an additional sale, whereas an inaccurate one leads to your emails being marked as spam. Given a database of customers and existing products, how do you find out which campaign should be sent to whom? In this talk, we will walk through a concrete project exploring that question, starting with the context of a commercial research centre. We will then look at how to validate the results (before we have any), and explain, at a fairly high level, some of the algorithms that can be used to resolve that question.

Bio:

Gary is a software engineer at Sparrho (www.sparrho.com), where he is helping to make scientific knowledge discoverable and accessible to everyone. Before that, he has worked for Condé Nast international as a technical lead on an e-commerce platform, for Huawei as a software researcher, and for the university of Brussels as a research assistant in databases and event stream processing. His interests cover real-world software engineering, distributed systems, big data, machine learning, programming languages and computer science.