Level UpROME DEVELOPER CONFERENCE

AIVRomeVideoGameLab

LEVEL UP:

improve
your skills

Let's Play

Meet the Devs - 2018

SVEN MIKA

Senior Software Developer at Ducandu GmbH

Senior Software Developer, founder and CEO at Ducandu GmbH, Sven Mika got a PhD in Computational Biology.

 

He wrote many articles about Machine Learning and is an active contributor to TensorForce, a python-based reinforcement learning library.

Towards Smarter NPCs with Deep Reinforcement-Learning

Deep reinforcement learning (DRL) is a subfield of machine learning that tries to solve generic problems following a learning-by-doing approach. The main technology behind DRL are so called deep neural networks, mathematical models, that can be trained fully automatically to become intelligently acting entities controlling an NPC in a game. The talk starts by quickly explaining the concepts of reinforcement learning and then introduces “MaRLEnE”, a prototype of a cloud-based reinforcement learning system for the UnrealEngine4, with which game developers will be able to smarten their NPCs overnight without the need for scripted, more classic game-AI methods such as behavior trees or utility functions.

TALK (PROGRAMMING)

06/05/2018, from 12 am to 13 pm
Track 2

Sched