Page 79 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

Chapter 3 Application, algorithm, tools directly related to deep learning 67

Consider the following eight steps to create DL model in Keras:
• Load the meta data.
• Preprocess all the loaded data.
• Define the model.
• Compile the model.
• Fit with the specified model.
• Evaluate it.
• Make the required predictions.
• Save the model.

2.3 CAFFE
CAFFE (Convolutional Architecture for Fast Feature Embed-
ding) is a DL framework, originally developed at the University
of California, Berkeley. It is open source, under a BSD license.
It is written in Cþþ, with a Python interface.
Yahoo has integrated CAFFE into Spark and enables DL on
distributed architectures. With CAFFE's high learning and pro-
cessing speed and the use of CPUs and GPUs, DL models can
be trained in just a few hours. Latest Nvidia Pascal GPUs support
CAFFE and offer up to 65% faster speeds. Up to 60 million images
per day can be processed (as of 2018). The intended program-
ming interfaces of the framework are Python and MATLAB.

2.3.1 The main features of CAFFE
Key features of CAFFE support CPUs, GPUs, Nvidia's Compute
Unified Device Architecture (CUDA), and the cuDNN Library
(CUDA deep neural network), also from this vendor. Thus, the
framework is designed primarily for speed [7].
CAFFE platform comes with Linux distributions such as
Ubuntu, MacOS, and Docker. For Windows installations, solu-
tions are mostly available on GitHub. For the Amazon AWS
Cloud, CAFFE is used as a preconfigured AMI [8].
CAFFE can be suitable for all DL architectures. The framework
is suitable for various architectures such as CNN (convolutional
neural network), long-term recurrent convolutional network
(LRCN), long short-term memory (LSTM), and fully connected
neural networks. A large number of preconfigured models are
hands on to the user, allowing a rapid introduction to machine
learning and the neural networks.