Neural networks with Theano and Lasagne

Syllabus

1. Theano basics

• Symbolic variables/tensors, expressions
• Functions, shared variables and updates

2. Overview of Lasagne

• Layer classes and building a network
• Objectives, optimizers, and training

3. Convolutional neural networks

• Image classification
• Fine-tuning a pretrained network
• Style transfer (“Neural Art”)

4. Recurrent neural networks

• Language model (text generation)
• CNN + RNN (image captioning)

5. Extending Lasagne

• Defining custom Layers