January 2019
Intermediate to advanced
322 pages
7h 29m
English
To improve the accuracy of the Keras VGG16 pre-trained model that we have considered for the use case of this chapter, we could also decide to retrain it and apply all of the best practices we have learned from the previous chapter (running more epochs, image augmentation, and so on). Once the model has been imported into DL4J, its training can be done exactly the same way it was explained in Chapter 7, Training Neural Networks with Spark (training with DL4J and Apache Spark). After loading, an instance of org.deeplearning4j.nn.graph.ComputationGraph is created, so the exact same principles for training multilayer networks apply here.
For completeness of information, you have to know that it is possible ...
Read now
Unlock full access