Differences

This shows you the differences between two versions of the page.

--- data_mining:neural_network:transfer_learning [2018/05/21 20:32] – [Transfer learning] phreazer
+++ data_mining:neural_network:transfer_learning [2018/05/25 21:13] (current) – [Transfer learning] phreazer
@@ Line 3: / Line 3: @@
 Using pre-trained models / their trained weights as a starting point to train a model for a different data set.
-Use pre-trained net, initialize last layers with random weights.
+Use pre-trained net, initialize last layers with random weights and e.g. create own softmax layer, **freeze** parameters in previous layers.
 Options: Train new layers of network, or even more layers.
@@ Line 9: / Line 9: @@
 Prereqs:
-* Task A and B have same input x
+  * Task A and B have same input x
-* Much data for existing model/task A, for new model only few data
+  * Much data for existing model/task A, for new model only few data
-* Low level features from A could be helpful for task B
+  * Low level features from A could be helpful for task B
-* Pre-trained model needs to generalize
+  * Pre-trained model needs to generalize
+Another trick:
+  * Precompute output of frozen layers for all samples (save computation time later)
+For larger set of samples:
+  * Only freeze first layers, train last few layers (and replace softmax)
+For large set of samples:
+  * Use weights as initialization, then train (and replace softmax)
 ===== Image-based NNs =====