Session Name: | Online Game Technology Summit: Machine Learning for Optimal Matchmaking |
Speaker(s): | Josh Menke |
Company Name(s): | 343 Industries |
Track / Format: | Online Game Technology Summit |
Overview: | Matchmaking has changed little since it was first automated. The general approach is to look for an ideal match across one or more preferred metrics, and then expand the ideal until a match is found. Issues with this approach make it difficult to trade-off the importance of each metric, hard to customize to specific player populations including geographic ones, and expanding forces players to wait longer than necessary. This session will present TrueMatch, a new matchmaking approach that allows developers to more intuitively express the value of each metric, and then uses machine learning to automatically optimize over the desired metrics in real-time. The results give better matches in less time and are customized to each player's characteristics and each region's real-time concurrency as it changes over time. |