Page 78 - Handbook of Deep Learning in Biomedical Engineering Techniques and Applications

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

Table 3.1 Keras Vs Tensorflow.

Parameters Keras TensorFlow
API Type High-level API High level & Low level
API
Prototyping Easy Difficult
Performance Slow High
Tools Uses API debugging tools like Use Tensor Board
TFDBG visualization tools
Processing Easy with simple network Good for complex models.
model models.

2.2.2 Installing keras: Amazon Web Service
Amazon Web Service (AWS) is a platform that offers cloud
computing service and products for researchers for any other
purposes. AWSs use their hardware, networking, and database
so that we can use them directly from the Internet. One of the
popular AWSs is the Deep Learning Amazon Machine Image
(DLAMI) or DL [5].
AWS DLAMI is a virtual environment that helps researchers or
practitioners to make working with DL. DLAMI offers from small
CPUs engine up to high-powered multi-GPUs engines with pre-
configured CUDA and cuDNN and comes with a variety of DL
frameworks [6]. DL is chosen AMI because it comes preinstalled
with popular DL frameworks.
If suppose custom DL framework for research, then we should
install the DL base AMI because it came up with fundamental li-
braries such as CUDA, cuDNN, GPUs drivers, and other vital li-
braries to run with DL environment.
The creation of framework can be of the following two types:
• Sequential API: This is used to implement very simple models
and is executed by adding layers to the existing model.
• Functional API
Keras functional API is very powerful, and it has built more
complex models, models with multiple output, directed acyclic
graph, and so on.