Open Internet by MindsNet

Achieve Human-Level Sample Efficiency

Develop ML systems that can learn tasks from the same amount of data that humans require, rather than the massive datasets current systems need. Humans can learn complex tasks from relatively few examples by leveraging prior knowledge, analogical reasoning, and efficient exploration strategies. Current ML systems require orders of magnitude more data than humans for comparable performance. The challenge involves understanding and replicating the mechanisms that enable human sample efficiency. Success would make ML applicable to many more domains where large datasets are unavailable and reduce the cost and time required for ML development.

Computing & Technology, Computer Science, Machine Learning