Chainer: A Powerful, Flexible, and Intuitive Framework of Neural Networks
Project Website: http://chainer.org/
Github Link: https://github.com/pfnet/chainer
Chainer is a flexible framework for neural networks. One major goal is flexibility, so it must enable us to write complex architectures simply and intuitively.
Most existing deep learning frameworks are based on the “Define-and-Run” scheme. That is, first a network is defined and fixed, and then the user periodically feeds it with mini-batches. Since the network is statically defined before any forward/backward computation, all the logic must be embedded into the network architecture as data. Consequently, defining a network architecture in such systems (e.g. Caffe) follows a declarative approach. Note that one can still produce such a static network definition using imperative languages (e.g. Torch7 and Theano-based frameworks).
In contrast, Chainer adopts a “Define-by-Run” scheme, i.e., the network is defined on-the-fly via the actual forward computation. More precisely, Chainer stores the history of computation instead of programming logic. This strategy enables to fully leverage the power of programming logic in Python. For example, Chainer does not need any magic to introduce conditionals and loops into the network definitions. The Define-by-Run scheme is the core concept of Chainer. We will show in this tutorial how to define networks dynamically.