This Video Of Two AI Cars Fighting Will Give You An Idea Of How Machines Think
Aadhya Khatri - Oct 11, 2019
The two cars in the video are two neural networks and their fight gives us an idea of how machine learning works
- Health-Technology Startups Is Emerging In India
- These Tactical Glasses Are Equipped With AI That Can Learn Soldiers' Tactics And Use It Against Them
- Maybe A Solution For All Glass-Made Products From IIT-Delhi
Sebastian Schuchmann, a machine learning developer, has created two neural networks recently. When they are first made, they were like brothers. They learned and developed together in a strange world with just the other for company. However, Schuchmann then made them fought with each other to amuse his YouTube followers.
The two networks are represented as a green jeep and a police cruiser competing in a video game-like world. They were made in Unity3D:
“The Unity guys have a Machine Learning Framework called ML-Agents, which I use for most of my stuff. It implements algorithms like PPO (this is the same stuff OpenAI uses). I am currently in the process of moving more of my training over in the cloud to speed up the production time.. but of course this comes at a cost,” said Schuchmann.
The two neural networks are created in a way that they leverage reinforcement learning for completing different tasks. So in the video, the police car must find a way to stop the other vehicle from crossing the white line. If it fails, there will be a punishment; if it succeeds, it will get rewarded. It is the case for the Jeep too.
The performance is not exactly impressive, but it gives us an initial idea of the way reinforcement learning works. Machines like these learn by making use of brute force. Early in the video, the two cars did not do much, but later; they joined the game.
If they get enough time, we might have the chance to see more complicated tactics perform by these networks as they try to complete the assigned tasks. Many experts in this field fear that someday, this kind of learning may result in a doomsday for us.
For now, the video is a cool example of how machines think: