Chapter 1: Overview

Pre-requisites

In this book, we aim to teach the underlying fundamental concepts of Deep Learning. We do not assume any prior knowledge in Math and Machine Learning. This book is designed for programmers. Throughout the book, we take an approach where we learn all the fundamentals by implementing it from the scratch. Only pre-requisites we expect from the readers, is to be able to code in JAVA or PYTHON. By the end of this book, you will build your own mini Deep Learning framework in JAVA or PYTHON.

PYTHON Setup

```

# Installation links / Commands for each OS type.

```

JAVA Setup

```

# Installation links / Commands for each OS type.

```

Artificial Intelligence, Machine Learning, and Deep Learning

Artificial Intelligence (AI): Is an effort to make machines perform intellectual tasks that are normally performed by humans. Hence, AI is a very broad term used to indicate the ability to automate tasks that involve thinking, learning and decision making.

Machine Learning (ML): In a classical programming world, we input the data and program the rules to analyze the data to obtain the required Answer. However, in Machine Learning, we input the data and provide a set of Answers and make computers “learn” what rules can be used to get the expected answers. Here, machines are trying to learn the mathematical equations that best represent the Rules that can take the data and give out Answers. This mathematical equation is what we call the “Model”.

Deep Learning (DL): Is a specific subfield of Machine Learning, where the learning happens in a hierarchical (layered) mode. You can visualize it like a human trying to learn English language. First, we learn to recognize the alphabets, then we learn words, then we learn sentences and then paragraphs and more. Similarly, in DL, machines learn from data in successive multiple layers of increasingly meaningful representations.

Neural Networks (NN): In DL, these layered representations are learned via models called “Neural Networks”. You can think of DL as a broad term for a specific methodology to learn and NN is a class of algorithms that implements this learning methodology.

Current Deep Learning Capabilities

Deep Learning has achieved the following breakthroughs:

• Near-human level image classification.

• Near-human level speech recognition.

• Near-human level handwriting transcription.

• Improved machine translation.

• Improved text-to-speech conversion.

• Digital assistants such as Google Now or Amazon Alexa.

• Near-human level autonomous driving.

• Improved ad targeting, as used by Google, Baidu, and Bing.

• Improved search results on the web.

• Answering natural language questions.

• Superhuman Go playing.

In fact, we can rather say, we have barely seen the impact of Deep Learning and many more complex results are already in front of us in foreseeable future. Breakthroughs in medical field, agriculture, climate control, education are all very close to the reality.

Myths of Deep Learning

• Deep Learning / Neural Networks are not the models of the brain. Neural Networks is inspired by neurology of brain, but, there are no formal proof that our brain works like Neural Networks we see.

• Deep Learning is not yet closer to achieving “human level general intelligence” as claimed by fancy digital media. Your Amazon Alexa or Google Home is not yet closer to the way we can interact among ourselves!

AI is the future

The long-term picture for AI is looking bright. We are only just getting started in applying deep learning to many important problems in which it could prove transformative, from medical diagnoses to digital assistants. Not all the amazing research in Deep Learning has found its way in the lives of people. AI will end up being applied to nearly every process that makes up our society and our daily lives, much like the Internet today. AI is coming, and it will transform our world in a fantastic way. Hence, we set out to teach the fundamentals of Deep Learning for today’s engineers to transform in to Deep Learning practitioners.