Reduce learning rate after certain number of epoch
11 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
Farhad Abedinzadeh
el 11 de Oct. de 2022
Comentada: Nicolas CRETIN
el 29 de Jul. de 2024 a las 9:49
Hi, I have a question about reducing the learning rate or another way to enhance accuracy during training a deep learning model.
Suppose the loss metric does not improve after a defined epoch (for example 8). Is there a way to reduce the learning rate?
or
If the accuracy metric (or loss metric) fails to improve after a certain number of epochs, give another solution to improve accuracy.
Thank you all for your time and consideration
0 comentarios
Respuestas (1)
Sivylla Paraskevopoulou
el 18 de Oct. de 2022
Check out the example Specify Training Options to see how to reduce the learning rate after a certain number of epochs.
2 comentarios
Francisco Serra
el 1 de Feb. de 2024
train requires rlTrainingOptions and this one doesn't have that decayment option. Do you any have a tip?
Nicolas CRETIN
el 29 de Jul. de 2024 a las 9:49
Hi Francisco,
Check this out, this should help: Learning rate schedule - Reinforcement Learning Toolbox - MATLAB Answers - MATLAB Central (mathworks.com)
Ver también
Categorías
Más información sobre Sequence and Numeric Feature Data Workflows en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!