Skip to content

Distilling the Knowledge in a Neural Network

Authors: Hinton et al. Year: 2015 ArXiv/Link: https://arxiv.org/abs/1503.02531

Summary

Transfer knowledge from large models to smaller models through soft targets and temperature scaling.

Key Concepts

  • Knowledge distillation
  • Model compression
  • Soft targets
  • Temperature scaling
  • Student-teacher learning

Impact

Foundational technique for model compression

Category

Knowledge Distillation