https://github.com/emalgorithm/gp-vs-cnn-uncertainty

Convolutional neural networks (CNNs) achieve state-of-the-art performance on image classification tasks, but provide no (appropriate) measure of confidence in their predictions. We show how it is possible to use Gaussian Processes on image embeddings generated by a CNN to improve variance estimation in prediction.
https://github.com/emalgorithm/gp-vs-cnn-uncertainty

Last synced: 7 days ago
JSON representation

Convolutional neural networks (CNNs) achieve state-of-the-art performance on image classification tasks, but provide no (appropriate) measure of confidence in their predictions. We show how it is possible to use Gaussian Processes on image embeddings generated by a CNN to improve variance estimation in prediction.

• Host: GitHub
• URL: https://github.com/emalgorithm/gp-vs-cnn-uncertainty
• Owner: emalgorithm
• Created: 2019-01-21T17:13:07.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2019-01-25T11:05:09.000Z (almost 6 years ago)
• Last Synced: 2024-08-02T15:18:56.977Z (3 months ago)
• Language: Jupyter Notebook
• Size: 106 MB
• Stars: 5
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project