Machine learning: from active experiments to intelligent teaching
•
1 like•266 views
This document discusses techniques for hedging predictions in machine learning models. It focuses on controlling similarity between a student model and teacher model to allow knowledge transfer while preventing exact replication. The goal is to leverage a teacher's expertise to help a student learn new concepts and tasks, but in a way that still allows the student to develop its own strengths.
Machine learning: from active experiments to intelligent teaching
- 1. http://www.slideshare.net/butest/hedging-predictions-in-machine-learning Learning with Intelligent Teacher: Similarity Control and Knowledge Transfer (2015) http://dx.doi.org/10.1007/978-3-319-17091-6_1, Vapnik, Vladimir, Izmailov, Rauf https://youtu.be/8OeqeTgMlq4