stochastic gradient descent method with momentum optimizer (sgdm)  (MathWorks Inc)

Journal: IEEE access : practical innovations, open solutions

Article Title: Cross-organ, cross-modality transfer learning: feasibility study for segmentation and classification

doi: 10.1109/access.2020.3038909

Figure Lengend Snippet: Tested CNN Architectures and Training Details

Article Snippet: For other hyperparameters, we used the default values of MATLAB for SGDM and Adam optimizer. table ft1 table-wrap mode="anchored" t5 TABLE II caption a7 Base Network SegNet-VGG16 DeepLabv3plus-ResNet18 Training on Intermediate Target Intermediate Target Choice of hyperpameters Optimizer SGDM ADAM Max epoch 32 50 32 50 Minibatch Size 4 4 128 128 Learning Rate le-3 le-3 le-5 le-4 Drop Factor No No No 0.9 Drop Period [Epoch] N/A N/A N/A 1 Open in a separate window Tested CNN Architectures and Training Details We let the weights of the network updated for every iteration during training.

Techniques: Biomarker Discovery