Unsupervised Deep Learning in Python

Autoencoders and Restricted Boltzmann Machines for Deep Neural Networks in Theano / Tensorflow, plus t-SNE and PCA

Register for this Course

$29.99 $199.99 USD 85% OFF!

Login or signup to register for this course

Have a coupon? Click here.

Course Data

Lectures: 86
Length: 10h 38m
Skill Level: All Levels
Languages: English
Includes: Lifetime access, certificate of completion (shareable on LinkedIn, Facebook, and Twitter), Q&A forum

Course Description

This course is the next logical step in my deep learning, data science, and machine learning series. I’ve done a lot of courses about deep learning, and I just released a course about unsupervised learning, where I talked about clustering and density estimation. So what do you get when you put these 2 together? Unsupervised deep learning!

In these course we’ll start with some very basic stuff - principal components analysis (PCA), and a popular nonlinear dimensionality reduction technique known as t-SNE (t-distributed stochastic neighbor embedding).

Next, we’ll look at a special type of unsupervised neural network called the autoencoder. After describing how an autoencoder works, I’ll show you how you can link a bunch of them together to form a deep stack of autoencoders, that leads to better performance of a supervised deep neural network. Autoencoders are like a non-linear form of PCA.

Last, we’ll look at Restricted Boltzmann Machines (RBMs). These are yet another popular unsupervised neural network, that you can use in the same way as autoencoders to pretrain your supervised deep neural network. I’ll show you an interesting way of training restricted Boltzmann machines, known as Gibbs sampling, a special case of Markov Chain Monte Carlo, and I’ll demonstrate how even though this method is only a rough approximation, it still ends up reducing other cost functions, such as the one used for autoencoders. This method is also known as Contrastive Divergence or CD-k. As in physical systems, we define a concept called free energy> and attempt to minimize this quantity.

Finally, we’ll bring all these concepts together and I’ll show you visually what happens when you use PCA and t-SNE on the features that the autoencoders and RBMs have learned, and we’ll see that even without labels the results suggest that a pattern has been found.

All the materials used in this course are FREE. Since this course is the 4th in the deep learning series, I will assume you already know calculus, linear algebra, and Python coding. You'll want to install Numpy, Theano and Tensorflow for this course. These are essential items in your data analytics toolbox.

If you are interested in deep learning and you want to learn about modern deep learning developments beyond just plain backpropagation, including using unsupervised neural networks to interpret what features can be automatically and hierarchically learned in a deep learning system, this course is for you.

This course focuses on "how to build and understand", not just "how to use". Anyone can learn to use an API in 15 minutes after reading some documentation. It's not about "remembering facts", it's about "seeing for yourself" via experimentation. It will teach you how to visualize what's happening in the model internally. If you want more than just a superficial look at machine learning models, this course is for you.

Suggested Prerequisites:

  • calculus
  • linear algebra
  • probability
  • Python coding: if/else, loops, lists, dicts, sets
  • Numpy coding: matrix and vector operations, loading a CSV file
  • linear regression, logistic regression
  • neural networks and backpropagation
  • Can write a feedforward neural network in Theano and TensorFlow

Tips for success:

  • Use the video speed changer! Personally, I like to watch at 2x.
  • Take handwritten notes. This will drastically increase your ability to retain the information.
  • Write down the equations. If you don't, I guarantee it will just look like gibberish.
  • Ask lots of questions on the discussion board. The more the better!
  • Don't get discouraged if you can't solve every exercise right away. Sometimes it'll take hours, days, or maybe weeks!
  • Write code yourself, this is an applied course! Don't be a "couch potato".

Testimonials and Success Stories

I am one of your students. Yesterday, I presented my paper at ICCV 2019. You have a significant part in this, so I want to sincerely thank you for your in-depth guidance to the puzzle of deep learning. Please keep making awesome courses that teach us!

I just watched your short video on “Predicting Stock Prices with LSTMs: One Mistake Everyone Makes.” Giggled with delight.

You probably already know this, but some of us really and truly appreciate you. BTW, I spent a reasonable amount of time making a learning roadmap based on your courses and have started the journey.

Looking forward to your new stuff.

Thank you for doing this! I wish everyone who call’s themselves a Data Scientist would take the time to do this either as a refresher or learn the material. I have had to work with so many people in prior roles that wanted to jump right into machine learning on my teams and didn’t even understand the first thing about the basics you have in here!!

I am signing up so that I have the easy refresh when needed and the see what you consider important, as well as to support your great work, thank you.

Thank you, I think you have opened my eyes. I was using API to implement Deep learning algorithms and each time I felt I was messing out on some things. So thank you very much.

I have been intending to send you an email expressing my gratitude for the work that you have done to create all of these data science courses in Machine Learning and Artificial Intelligence. I have been looking long and hard for courses that have mathematical rigor relative to the application of the ML & AI algorithms as opposed to just exhibit some 'canned routine' and then viola here is your neural network or logistical regression. ...


I have now taken a few classes from some well-known AI profs at Stanford (Andrew Ng, Christopher Manning, …) with an overall average mark in the mid-90s. Just so you know, you are as good as any of them. But I hope that you already know that.

I wish you a happy and safe holiday season. I am glad you chose to share your knowledge with the rest of us.

Hi Sir I am a student from India. I've been wanting to write a note to thank you for the courses that you've made because they have changed my career. I wanted to work in the field of data science but I was not having proper guidance but then I stumbled upon your "Logistic Regression" course in March and since then, there's been no looking back. I learned ANNs, CNNs, RNNs, Tensorflow, NLP and whatnot by going through your lectures. The knowledge that I gained enabled me to get a job as a Business Technology Analyst at one of my dream firms even in the midst of this pandemic. For that, I shall always be grateful to you. Please keep making more courses with the level of detail that you do in low-level libraries like Theano.

I just wanted to reach out and thank you for your most excellent course that I am nearing finishing.

And, I couldn't agree more with some of your "rants", and found myself nodding vigorously!

You are an excellent teacher, and a rare breed.

And, your courses are frankly, more digestible and teach a student far more than some of the top-tier courses from ivy leagues I have taken in the past.

(I plan to go through many more courses, one by one!)

I know you must be deluged with complaints in spite of the best content around That's just human nature.

Also, satisfied people rarely take the time to write, so I thought I will write in for a change. :)

Hello, Lazy Programmer!

In the process of completing my Master’s at Hunan University, China, I am writing this feedback to you in order to express my deep gratitude for all the knowledge and skills I have obtained studying your courses and following your recommendations.

The first course of yours I took was on Convolutional Neural Networks (“Deep Learning p.5”, as far as I remember). Answering one of my questions on the Q&A board, you suggested I should start from the beginning – the Linear and Logistic Regression courses. Despite that I assumed I had already known many basic things at that time, I overcame my “pride” and decided to start my journey in Deep Learning from scratch. ...


By the way, if you are interested to hear. I used the HMM classification, as it was in your course (95% of the script, I had little adjustments there), for the Customer-Care department in a big known fintech company. to predict who will call them, so they can call him before the rush hours, and improve the service. Instead of a poem, I Had a sequence of the last 24 hours' events that the customer had, like: "Loaded money", "Usage in the food service", "Entering the app", "Trying to change the password", etc... the label was called or didn't call. The outcome was great. They use it for their VIP customers. Our data science department and I got a lot of praise.


Introduction and Outline

6 Lectures · 28min
  1. Introduction and Outline (07:28) (FREE preview available)
  2. Where does this course fit into your deep learning studies? (02:58)
  3. How to Succeed in this Course (03:04)
  4. Where to get the code and data (05:03)
  5. Tensorflow or Theano - Your Choice! (04:10)
  6. What are the practical applications of unsupervised deep learning? (05:35)

Principal Components Analysis

11 Lectures · 01hr 08min
  1. What does PCA do? (04:33)
  2. How does PCA work? (11:22)
  3. Why does PCA work? (PCA derivation) (10:13)
  4. PCA only rotates (05:30)
  5. MNIST visualization, finding the optimal number of principal components (03:40)
  6. PCA implementation (03:29)
  7. PCA for NLP (03:38)
  8. PCA objective function (02:06)
  9. PCA Application: Naive Bayes (09:51)
  10. SVD (Singular Value Decomposition) (10:59)
  11. Suggestion Box (03:10)

t-SNE (t-distributed Stochastic Neighbor Embedding)

5 Lectures · 21min
  1. t-SNE Theory (04:29)
  2. t-SNE Visualization (04:34)
  3. t-SNE on the Donut (05:52)
  4. t-SNE on XOR (04:37)
  5. t-SNE on MNIST (02:13)


12 Lectures · 01hr 10min
  1. Autoencoders (03:21)
  2. Denoising Autoencoders (01:56)
  3. Stacked Autoencoders (03:33)
  4. Writing the autoencoder class in code (Theano) (11:56)
  5. Testing our Autoencoder (Theano) (03:06)
  6. Writing the deep neural network class in code (Theano) (12:43)
  7. Autoencoder in Code (Tensorflow) (08:30)
  8. Testing greedy layer-wise autoencoder training vs. pure backpropagation (03:34)
  9. Cross Entropy vs. KL Divergence (04:41)
  10. Deep Autoencoder Visualization Description (01:33)
  11. Deep Autoencoder Visualization in Code (11:15)
  12. An Autoencoder in 1 Line of Code (04:51)

Restricted Boltzmann Machines

11 Lectures · 01hr 20min
  1. Basic Outline for RBMs (04:52)
  2. Intro to RBMs (08:22)
  3. Motivation Behind RBMs (06:52)
  4. Intractability (03:12)
  5. Neural Network Equations (07:44)
  6. Training an RBM (part 1) (11:35)
  7. Training an RBM (part 2) (06:19)
  8. Training an RBM (part 3) - Free Energy (07:21)
  9. RBM Greedy Layer-Wise Pretraining (04:51)
  10. RBM in Code (Theano) and Greedy Layer-wise Pre-training on MNIST (14:25)
  11. RBM in Code (Tensorflow) (05:04)

The Vanishing Gradient Problem

2 Lectures · 15min
  1. The Vanishing Gradient Problem Description (03:08)
  2. The Vanishing Gradient Problem Demo in Code (12:18)

Extras + Visualizing what features a neural network has learned

1 Lectures · 02min
  1. Exercises on feature visualization and interpretation (02:08)

Applications to NLP (Natural Language Processing)

3 Lectures · 21min
  1. Application of PCA and SVD to NLP (Natural Language Processing) (02:31)
  2. Latent Semantic Analysis in Code (10:08)
  3. Application of t-SNE + K-Means: Finding Clusters of Related Words (08:39)

Applications to Recommender Systems

10 Lectures · 01hr 28min
  1. Recommender Systems Section Introduction (12:31)
  2. Why Autoencoders and RBMs work (05:59)
  3. Data Preparation and Logistics (05:34)
  4. Autoencoders (AutoRec) Discussion (10:15)
  5. Autoencoders (AutoRec) Code (11:46)
  6. Categorical RBM for Recommender System Ratings (11:32)
  7. Recommender RBM Code pt 1 (07:27)
  8. Recommender RBM Code pt 2 (04:17)
  9. Recommender RBM Code pt 3 (11:43)
  10. Speeding up the Recommender RBM Code (07:54)

Basics Review

7 Lectures · 52min
  1. Theano Basics: Variables, Functions, Expressions, Optimization (07:47)
  2. Building a neural network in Theano (09:17)
  3. TensorFlow Basics: Variables, Functions, Expressions, Optimization (07:27)
  4. Building a neural network in TensorFlow (09:43)
  5. Keras Basics (06:49)
  6. Keras Neural Network in Code (06:38)
  7. Keras Functional API (04:27)

Optional - Legacy RBM Lectures

4 Lectures · 25min
  1. Restricted Boltzmann Machine Theory (09:32)
  2. Deriving Conditional Probabilities from Joint Probability (06:19)
  3. Contrastive Divergence for RBM Training (02:46)
  4. How to derive the free energy formula (06:33)

Setting Up Your Environment (Appendix/FAQ by Student Request)

3 Lectures · 42min
  1. Pre-Installation Check (04:13)
  2. Anaconda Environment Setup (20:21)
  3. How to install Numpy, Scipy, Matplotlib, Pandas, IPython, Theano, and TensorFlow (17:33)

Extra Help With Python Coding for Beginners (Appendix/FAQ by Student Request)

5 Lectures · 52min
  1. How to Code Yourself (part 1) (15:55)
  2. How to Code Yourself (part 2) (09:24)
  3. Proof that using Jupyter Notebook is the same as not using it (12:29)
  4. Python 2 vs Python 3 (04:38)
  5. Is Theano Dead? (10:04)

Effective Learning Strategies for Machine Learning (Appendix/FAQ by Student Request)

4 Lectures · 59min
  1. How to Succeed in this Course (Long Version) (10:25)
  2. Is this for Beginners or Experts? Academic or Practical? Fast or slow-paced? (22:05)
  3. What order should I take your courses in? (part 1) (11:19)
  4. What order should I take your courses in? (part 2) (16:07)

Appendix / FAQ Finale

2 Lectures · 08min
  1. What is the Appendix? (02:48)
  2. Where to get discount coupons and FREE deep learning material (05:49)
This website is using cookies. That's Fine