Computer Vision News - January 2017

Computer Vision News Tool 11 The code under # Data loads the MNIST data and reshapes it accordingly, in order to fit the network. The code under # Network defines the optimizee -- here it is a simple #MLP network. The build () function puts everything together (it feeds the optimizee network with the data) and returns its lost function to be used by the optimizer. Tool The Google DeepMind team suggests we move from hand-designed update rules to a learned update rule