As a practicing data scientist and machine learning educator, I’m always taking courses. Suppose you want to go directly to applications, learn the most current software and get a job. In that case, I recommend my IBM Data Science Professional Certificate or IBM Machine Learning Professional Certificate.

This month’s course is Deep Learning at VU University Amsterdam. The slides break down complex topics into simple diagrams with lots of explanations. The course has several instructors, each one focusing on their specialty here is a list:

Instructors |

Jakub M. Tomczak |

Peter Bloem |

Michael Cochez |

David Romero |

Emile van Krieken |

Table of Content

- 1.0 Introduction
- 2.0 Neural Networks
- 3.0 : Convolutional Neural Networks
- 3.1: Convolutional Neural Networks – Introduction
- 3.2a: 1-Dimensional Convolutional Neural Networks: getting started
- 3.2b: 1-Dimensional Convolutional Neural Networks – formalism and solving issues
- 3.3: Two-Dimensional Convolutional Neural Networks
- 3.4 : An Example of a real-world convolutional neural network: AlexNet

- 4.0 Tools of the trade
- 5.0 Sequential Data
- 6.0 Latent Variable Models (pPCA and VAE)
- 7.0 Generative AdversarialNetwork
- 8.0 Learning with Graphs
- 9.0 Reinforcement Learning
- 10 More or Generative Modding
- Q-Learning
- Transformers

**1.0 Introduction**

Course Introduction and a primer on probability

### 1.0 Course Introduction

**1.2: A quick intro to AI**

**1.3: Probabilistic learning**

** 1.4: From logistic regression to fully-connected networks**

## 2.0 Neural Networks

An introduction to Neural Networks and how to train them.

### Lecture 2.1: A Review of Neural Networks

### Lecture 2.2: Backpropagation, scalar perspective

### Lecture 2.3: Backpropagation

### Lecture 2.4: Automatic Differentiation

## 3.0 : Convolutional Neural Networks

An introduction to Convolutional Neural Networks \.

### 3.1: Convolutional Neural Networks – Introduction

### 3.2a: 1-Dimensional Convolutional Neural Networks: getting started

### 3.2b: 1-Dimensional Convolutional Neural Networks – formalism and solving issues

### 3.3: Two-Dimensional Convolutional Neural Networks

### 3.4 : An Example of a real-world convolutional neural network: AlexNet

## 4.0 Tools of the trade

Tips and tricks to train Neural Networks

**4.1 General Deep Learning practice**

### 4.2 Why does deep Learning work?

** 4.3 Optimizers**

**4.4 The Bag of Tricks**

## 5.0 Sequential Data

An introduction to Sequential Data like natural languish processing

**5.1 Sequence Models**

** 5.2 Recurrent Neural Networks**

**5.3 Long Short-Term Memory**

** 5.4 – CNNs for Sequential Data**

**5.5 ELMo, Word2Vec**

## 6.0 Latent Variable Models (pPCA and VAE)

Generative modelling is used to create data like images.

** 6.1: Introduction to Deep Generative Modeling**

**6.2: Probabilistic PCA**

** 6.3: Variational Auto-Encoders**

## 7.0 Generative AdversarialNetwork

A popular method used for generative modelling

**Lecture 7.1 Implicit models: Density Networks**

**Lecture 7.2 Implicit models: GANs**

## 8.0 Learning with Graphs

Graphs a powerful tool in deep learning

### Lecture 8.1a: Introduction – Graphs

### Lecture 8.1b: Introduction – Embeddings

### Lecture 8.2: Graph and node embedding

### 8.3: Graph Neural Networks

### Lecture 8.4: Application – query embedding

## 9.0 Reinforcement Learning

How computers learn to do complex sequential tasks like playing games

### 9.1: Introduction to Reinforcement Learning

### 9.2: The REINFORCE algorithm

### 9.3: Gradient Estimation

## 10 More or Generative Modding

Different types of Generative Modding.

### Lecture 10.1: ARM & Flows

### 10.2: ARM & Flows

### 10.3: ARM & Flows

## Q-Learning

A better method of Reinforcement Learning

### 11.1: Deep Q-Learning

### 11.2: Variance Reduction for Policy Gradient (Actor-Critic)

### 11.3: World Models

## Transformers

A new powerful neural network architecture that was originally used in natural languish processing , but is now used everywhere