New AI Training Method Leverages Crowdsourced Feedback for Faster Learning
The Human Guided Exploration (HuGE) approach, developed by researchers from MIT, Harvard, and the University of Washington, allows AI agents to learn complex tasks more efficiently, even with error-prone input from nonexpert users.
- Researchers from MIT, Harvard University, and the University of Washington have developed a new reinforcement learning approach, Human Guided Exploration (HuGE), that uses crowdsourced feedback to train AI agents.
- HuGE allows AI agents to learn more quickly, despite the fact that data crowdsourced from users are often full of errors.
- The new approach allows feedback to be gathered asynchronously, enabling nonexpert users around the world to contribute to teaching the agent.
- In real-world and simulated experiments, HuGE helped agents learn to achieve the goal faster than other methods.
- In the future, this method could help a robot learn to perform specific tasks in a user's home quickly, without the owner needing to show the robot physical examples of each task.
ScienceDaily 
MIT News 
