How to start the training of a deep net from a given state?

1 次查看（过去 30 天）
Sébastien Tosi
Sébastien Tosi 2017-10-17
I would like to start the training of a deep net from the state it was in after a previous training. In my application the user iteratively adds annotations and I want to avoid training the netwrok from scratch at each iteration. Here is the code I use to define and train the network:
layers = [ ... imageInputLayer([2*BoxRad+1 2*BoxRad+1 1]) convolution2dLayer([9 9],100,'Stride',1) reluLayer maxPooling2dLayer(2,'Stride',2) convolution2dLayer([5 5],25,'Stride',1) reluLayer maxPooling2dLayer(2,'Stride',2) fullyConnectedLayer(2) softmaxLayer classificationLayer]; options = trainingOptions('sgdm', 'MaxEpochs', 30, 'MiniBatchSize', 32); net = trainNetwork(Images,Lbl,layers,options);

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Function Approximation and Clustering 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by