| notebook.community

notebook.community



In [17]:

    
from notebook.services.config import ConfigManager
cm = ConfigManager()
cm.update('livereveal', {
              'theme': 'simple',
              'transition': 'fade',
              'start_slideshow_at': 'selected',
})









    Out[17]:





{u'scroll': True,
 u'start_slideshow_at': 'selected',
 u'theme': 'simple',
 u'transition': 'fade'}

What is Lasagne?

Lasagne is a lightweight library to build and train neural networks in Theano.

Bookkeeping device
Collection of helper functions

Design Goals

Simplicity, Transparency, Modularity
Abstract Theano as little as possible

Layers

Layer classes are the bulk of Lasagne

Abstraction of a layer of neurons in a network
But also used for other functions (pooling, reshaping, etc.)
Track connectivity and data shape
Create and manage parameters (weights)

Other Modules

init - weight initialization algorithms
objectives - loss functions and evaluation metrics
regularization - penalty functions
updates - gradient descent variants

Typical Usage

Define layer sequence

Typical Usage

Get expression for output

Typical Usage

Get expression for loss function

Typical Usage

Find gradient of loss wrt params and optimize

Typical Usage

Update parameters

Resources



In [ ]:



In [ ]:



In [ ]: