GDC is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

GDC + VRDC 2019 Session Scheduler

View, browse and sort the ever-growing list of GDC sessions by pass type, track, format, and more! With Session Scheduler, create an account to build your own GDC schedule and access it anywhere, including on the GDC app, once live.

If you've registered for GDC, you can use the same login credentials. Adding a session to your schedule does not guarantee you a seat. Sessions do fill up, so please arrive early to sessions that you would like to attend.

Math for Game Developers: Inside Neural Networks

Michael Buttner (Principal Research Engineer, Unity Technologies)

Location: Room 2006, West Hall

Date: Tuesday, March 19

Time: 3:50pm - 4:20pm

Pass Type: All Access, GDC Conference + Summits, GDC Summits - Get your pass now!

Topic: Programming

Format: Tutorial

Tutorials: Math for Game Developers

Vault Recording: Video

Neural networks are gaining attention due to their ability to solve difficult problems. But they are still not very often used to solve problems in real-time scenarios. The majority of practical examples are focused on image processing and very few examples focus on neural network inference in high-performance applications. In this presentation, the speaker will show practical examples of how to use neural networks in performance-critical code. The presentation will take a deep dive and show neural network training and inference without using any frameworks or math libraries. Instead, the talk is going to focus on bare metal C++ using a data-oriented approach and optimized SSE implementations.

Takeaway

The takeaway for attendees is to learn how to leverage neural networks in performance-critical scenarios and how to write an efficient inference engine without using any frameworks or math libraries using data-oriented design principles.

Intended Audience

This talk is for gameplay and engine programmers that are interested in using neural networks.