Deep Learning
With Python
A Crash Course to Deep Learning with illustrations in Python Programming Language
Robert Kissinger
Copyright
Copyright2021 Robert Kissinger
All rights reserved. No part of this book may be reproduced or used in any manner without the prior written permission of the copyright owner, except for the use of brief quotations in a book review.
While the advice and information in this book are believed to be true and accurate at the date of publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein.
Printed on acid-free paper.
Printed in the United States of America
2021 by Robert Kissinger
Table of Contents
CHAPTER ONE
ARTIFICIAL INTELLIGENCE, MACHINE LEARNING AND DEEP LEARNING: HOW THEY ALL RELATE
Artificial intelligence has become the hope of humanity, in an attempt to connect every dot between machines and man. With intelligent machine, it becomes very easy to for human being to relate with them just the way human relates with human. Artificial intelligence offers many promises, much of which are already manifesting in recent time. For instance;
- Artificial intelligence forms the foundation of smart and intelligent devices, which have been used in recent time to give travel alert and weather reports. For instance, say you want to embark on an urgent trip; you can use your smart devices (iPhone, iPad, Apple watch, Fitbit devices etc) to predict weather report for the city or country that you are going. This way, you are in good position to make informed decision concerning whether you can travel or not.
- Artificial intelligence, in the form of smart bots, has been used to send invitation to large group of people, and can even send customized invites to selected group of individuals.
The examples described above are just to illustrate the simple and everyday application of Artificial intelligence to regular persons who are just interested in how the seemingly talked about field Artificial intelligence can help them ease their everyday tasks.
The History of AI and Machine Learning
You must have been wondering by now where the AI came from, well, it is good to say that it didnt evolve to what is today by just dreaming. There were and currently are realistic plans by people who had long imagined a world where things happen by mere thinking about it. Artificial intelligence is long rooted in statistics and military science, with great contribution from mathematics, philosophy, cognitive science and psychology. The original rationale behind Artificial intelligence is the quest to make computer more capable and useful of independent reasoning. Most people who are interested in history have traced the origin of AI to a particular Dartmouth research project in 1956. The project, according to history, explored hot topics such as symbolic methods and problem solving. Few years after this, the US Defense Department took up the work on AI and thereby focused on making computers mimic the ways human being processed thoughts.
For instance, the Defense Advanced Research Projects Agency (DARPA) finished a project in the 1970s when they were able to match all the streets in the Unites States. And even long before the like of Amazon, Microsoft or Google birth the idea, the Defense Advanced Research Projects Agency (DARPA) in 2003 created intelligent personal assistance. This later opened the way for the kind of automation and formal reasoning that are observable in computers we use today.
Artificial Intelligence and Machine Learning
While Artificial Intelligence entails the all-inclusive science of mimicking the abilities of human, machine learning is a particular subset of Artificial intelligence that trains machine how to incorporate learning. Machine learning and Deep learning are subset of Artificial intelligence
In general, artificial intelligence has many subfields, which are;
Machine learning : This automates analytical model building. It deploys methods from operation research, neural networks, physics and statistics to explore hidden insights in data without being told explicitly where to look for such data or the conclusions to be drawn from such data.
A neural network: This is a type of machine learning that was inspired by the way human brain works. It is actually a computing system that comprises interconnected units that help to process information just by responding to external inputs, thereby transferring information between units. The process actually needs multiple passes at the data in order to establish connections and obtain meaning from undefined data.
Deep learning : This utilizes vast neural networks with many surfaces of processing units, relying heavily on advances made in the area of computing power and enhanced training methods to learn hard patterns in huge amount of data. Common areas where this is applied include speech and image recognition.
Computer vision : This relies particularly on deep learning and pattern recognition to identify what is in video or an image. When machines are created with the ability to process, analyze and give meaning to images, then they will be able to capture videos or images in real time and give interpretation to their surroundings.
Natural language processing: This refers to the power of computers to scrutinize, understand and then generate human language, including the speech. The succeeding stage of Natural language processing is natural language interaction, which gives human the ability to talk to computers deploying everyday language to carry out tasks.
While machine learning thrives on the belief that machines should be able to adapt and learn through experience, Artificial intelligence talks about a broader concept where machines will be able to perform tasks in a smart way.
Artificial Intelligence uses deep learning, machine learning, and some other techniques to give solutions to real life problems.
Where are we now with Artificial Intelligence?
With Artificial Intelligence, people can now ask a machine questions, albeit loudly, and then expect a response about just anything from inventory, sales, customer retention, fraud detection and prevention etc. AI can also give computers the power to discover details that you never even give a thought about. AI will give you a simple summary of your data while helping you suggest other ways you can use to analyze it. It can also share details that are related to questions that have been asked before; either from you or from anyone else who asked the same question in the past. The answers are usually displayed on the screen or delivered conversationally.
How can we apply all these in the real world? For instance, in health care, effectiveness of a treatment can be determined more quickly. Also, in retails, you can quickly get suggestions about add-on items. In finance, prevention of fraud becomes very easy rather than just detecting it.
In each of the cases highlighted above, the machine knows what detail is needed, checks the relationships between all of the variables, design an answer and then give it to you letting you follow it up with further queries.
Learning representation from data
To explain deep learning, and fully understand the discrepancy between deep learning and some other machine learning approaches, you need to understand what the algorithm of machine learning is actually doing. Already, you have seen how machine learning executes tasks given some instructions from users. To fully comprehend machine learning, you will need three things;
