- [Q] Considering transfer learning, why there are (very) large weight updates if the 'Dense' layers on top are randomly initialized while convolutional base stays frozen? https://keras.io/guides/transfer_learning/ 3 comments learnmachinelearning
Linking pages
- Building powerful image classification models using very little data http://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html 6 comments
- Building powerful image classification models using very little data http://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html?cmp=em-na-na-na-newsltr_four_short_links_20160608&imm_mid=0e4984 0 comments
- A Comprehensive Study Guide for the Google Professional Machine Learning Engineer Certification | by Jeffrey Luppes | Towards Data Science https://towardsdatascience.com/a-comprehensive-study-guide-for-the-google-professional-machine-learning-engineer-certification-1e411db4d2cf 0 comments
- UniTune: Google's Alternative Neural Image Editing Technique - Unite.AI https://www.unite.ai/unitune-googles-alternative-neural-image-editing-technique/ 0 comments
Related searches:
Search whole site: site:keras.io
Search title: Transfer learning & fine-tuning
See how to search.