How to create a continuous learning ML/DL model for Computer Vision? Which approach would be ameliorated to train the ML/DL model up-to-date with respect to the new images/dataset? Let's suppose if new images could have come on a daily basis, then would I need to re-train the complete model again with the new dataset(Images)? Please let me know ,Is there anyway to train the model with new images & merge into the old trained model?
https://scholar.google.com/scholar?hl=ar&as_sdt=0%2C5&as_vis=1&q=How+to+create+a+continuous+learning+ML%2FDL+model+for+Computer+Vision%3F&btnG=#d=gs_qabs&u=%23p%3D6cbpK7Cc9K4J
https://scholar.google.com/scholar?hl=ar&as_sdt=0%2C5&as_vis=1&q=How+to+create+a+continuous+learning+ML%2FDL+model+for+Computer+Vision%3F&btnG=#d=gs_qabs&u=%23p%3D6cbpK7Cc9K4J
Well, tensorflow provide this process because it's designed to be for production, check tensorflow 2.0 live training.
The best way to machine learning (ML) is to use deep learning(DL) and to make this learning continuous we use deep reinforcement learning for the machine and it is one of the applications currently used in training the machine to win games
- Badges
- Science topic