This is a companion discussion topic for the original entry at:
Reinforcement learning is one of the most intriguing fields in machine learning, and has recently made tremendous breakthroughs in a variety of domains, but perhaps most notably in un-assisted game play. In this talk, we’ll take a look at how to use the miner package to train learning agents in Minecraft using R bindings for CNTK, Keras and Tensorflow. We’ll start with simple tasks, such as learning how to ascend mountains and stairs, to more challenging tasks such as solving random mazes with obstacles using deep Q-learning. The talk will provide examples using the new R bindings for CNTK, as well as the Keras and Tensorflow packages, and describe how to use docker images for Minecraft and CUDA enabled containers with R to easily and affordably try out the examples on cloud platforms.
** Ali-Kazim Zaidi ** - Data Scientist in the AI Research Team
Ali Zaidi is data scientist in Microsoft’s AI and Research Group, where he spends his day trying to make distributed computing and machine learning in the cloud easier, more efficient, and more enjoyable for data scientists and developers alike. Previously, Ali was a research associate at NERA (National Economic Research Associates), providing statistical expertise on financial risk, securities valuation, and asset pricing. He studied statistics at the University of Toronto and computer science at Stanford University.