Applying deep learning techniques to learn player representations for modelling, clustering, and visualization.
Using techniques for learning word embeddings, category embeddings, sentence embeddings, and low-dimension projections with T-SNE we are able to develop clustering and visualization tools to quickly explore and analyse player behaviours via an interactive web application. These tools allow us to identify emerging player behaviour, balance issues within the game, and player ‘metas’ in our new CCG Chronicle: Runescape Legends.
Bio: Miroslaw Horbal – Data Scientist @ Jagex. Recently moved to the UK from Canada after becoming tired of riding moose and wrestling bears on a regular basis. I’ve been working as a Data Scientist for the past 2 years, working within the gaming industry since November 2015. I reached a Kaggle Master status in early 2014 and have been focusing much of my time on bleeding edge research in the field of Deep Learning. Outside of Data Science / Machine Learning, I’m an avid gamer, played Battlefield competitively for 3 years and been gaming since the age of 4.