Grokking Deep Learning in Motion
Beau Carnes
  • Course duration: 2h 41m
    Estimated full duration: 5h 25m
  • MEAP began August 2018
  • Publication in November 2018 (estimated)
Despite being one of the biggest technical leaps in AI in decades, building an understanding in deep learning doesn't mean you need a math degree. All it takes is the right intuitive approach, and you'll be writing your own neural networks in pure Python in no time!
Table of Contents detailed table of contents

Introducing Deep Learning

Introduction

What you need to get started

Fundamental concepts

What is Deep Learning and Machine Learning?

Supervised vs. unsupervised learning

Parametric vs. non-parametric learning

Introduction to neural prediction

Making a prediction

What does a Neural Network do?

Multiple inputs

Multiple outputs and stacking predictions

Primer on NumPy

Introduction to neural learning

Compare and learn

Why measure error?

Hot and cold learning

Practice exercises

Gradient descent

Learning with gradient decent

The secret to learning

How to use a derivative to learn

Alpha

Learning multiple weights at a time

Gradient descent learning with multiple inputs

Several steps of learning

Gradient descent with multiple outputs

Visualizing weight values

Building your first "deep" neural network

The streetlight problem

Building our neural network

Up and down pressure

Correlation and backpropagation

Linear vs. non-linear

Our first "deep" neural network

How to picture neural networks

Simplifying

Simplified visualization

Seeing the network predict

Learning signal and ignoring noise

3-layer network on MNIST

Memorization vs. generalization

Overfitting

Early stopping

Dropout

Batch gradient descent

Modeling probabilities and non-linearities

Activation function constraints

Standard activation functions

Softmax and implementation in code

Conclusion

About the subject

Machine learning has made remarkable progress in recent years. Deep-learning systems now enable previously impossible smart applications, revolutionizing image recognition and natural-language processing, and identifying complex patterns in data. To really get the most out of deep learning, you need to understand it inside and out, but where do you start? This liveVideo course is the perfect jumping off point!

About the video

Grokking Deep Learning in Motion is a new liveVideo course that takes you on a journey into the world of deep learning. Rather than just learn how to use a single library or framework, you’ll actually discover how to build these algorithms completely from scratch!

Professional instructor Beau Carnes breaks deep learning wide open, drawing together his expertise in video instruction and Andrew Trask's unique, intuitive approach from Grokking Deep Learning! As you move through this course, you’ll learn the fundamentals of deep learning from a unique standing! Using Python, as well as Jupyter Notebooks, you’ll get stuck right in to the basics of neural prediction and learning, and teach your algorithms to visualize things like different weights. Throughout, you’ll train your neural network to be smarter, faster, and better at its job in a variety of ways, ready for the real world!

Packed with great animations and explanations that bring the world of deep learning to life in a way that just makes sense, Grokking Deep Learning in Motion is exactly what anyone needs to build an intuitive understanding of one of the hottest techniques in machine learning.

This liveVideo also works perfectly alongside the original Grokking Deep Learning by Andrew Trask, bringing his unique way to teaching to life.

Prerequisites

This liveVideo course is perfect for anyone with high school-level math and basic programming skills with a language like Python. Experience with Calculus is helpful but NOT required.

What you will learn

  • The differences between deep and machine learning
  • An introduction to neural prediction
  • Building your first deep neural network
  • The importance of visualization tools
  • Memorization vs Generalization
  • Modeling probabilities and non-linearities

About the instructor

liveVideo instructor Beau Carnes is a software developer and a recognized authority in software instruction. Besides teaching in-person workshops and classes, Beau has recently joined the team at freeCodeCamp as their lead video instructor, helping to teach over 2 million people around the world to code. Beau also teaches Manning's best-selling video course, Algorithms in Motion.