A Warm Welcome!
Welcome to the new course on
Deep Learning - Tools and Applications
. In
the first course,
Deep Learning - Chorale Prelude
, hope you had an introduction to
the various neural network architectures, how each architecture fits in for a specific
datatype.
In this course, you will understand the various tools that are used to implement the
architectures and the real-world applications.
Course - Introduction
You will experience the following topics during this journey into this course.
Platforms available for deep learning
Popular libraries to build a deep net
Various configuration parameters to tune our deep net
Techniques to improve the performance and various metrics available to
measure the performance
Applications of deep net
What is a Deep Net Platform?
A
deep net platform
is service that allows you to incorporate deep net in
our applications without building one from scratch.
This platform provides a
set of tools and interface to build your own custom deep net.
Types of Deep Net Platforms
Deep net platforms are of two types.
Software platform
: This platform is available as downloadable packages
that need to be deployed on your hardware.
Full platform
: It is available as online interactive UI to build and deploy
models without any coding experience.
Deep Net Platforms - Tools
The following are the tools offered by deep net platforms.
Deep Net capability
Data Munging
UI/Model Management
Infrastructure
Deep Net Capability
Using
Deep Net Capability
tool,
the user can choose among various
available deep nets depending on the use case.
For example, you can
select convolutional net for image recognition or RNTN for text analysis and
so on.
Data Munging
Data Munging
tool enables us to import data (structure or unstructured) from
various sources and manipulate it to feed into a selected deep net
UI/Model Management
This tool provides a graphical interface to customize deep net, manage deep
net models without the need for explicit coding.
Infrastructure
In order to train a significant amount of data and increase performance, some
platforms provide infrastructure facilities like built-in integration tools such
as
Amazon S3, SQL
, and
NoSQL
.
What is H2O.ai?
H2O.ai
is an opensource machine learning platform. Along with many
machine learning algorithms, this platform currently provides one deep net
capability that is
Multilayer Perceptron
.
It offers intuitive UI and powerful data munging capabilities.
It is available as a downloadable package that has to be installed and
managed in our own hardware infrastructure.
Features of H2O.ai
H2O.ai
supports machine learning models as GLM, Distributed random forest,
K-means clustering, gradient boosting machine, a Cox proportional hazard
and Naive Bayes classifier.
