Course Abstract

Training duration : 4 hours

Relatively obscure a few short years ago, deep learning is ubiquitous today across data-driven applications as diverse as machine vision, natural language processing, artistic creativity, and complex sequential decision-making. This deep learning primer brings the revolutionary approach behind contemporary artificial intelligence to life with interactive demos featuring TensorFlow 2 and PyTorch, the two leading deep learning libraries. To facilitate an intuitive understanding of deep learning’s artificial-neural-network foundations, essential theory will be introduced visually and pragmatically. Paired with tips for overcoming common pitfalls and hands-on code run-throughs provided in Python Jupyter notebooks, this foundational knowledge empowers you to build powerful state-of-the-art deep neural network models. Many resources will be provided for digging further into any deep learning-related topic that piques your interest.

DIFFICULTY LEVEL: INTERMEDIATE

Learning Objectives

  • Understand the essential theory of artificial neural networks, including which deep learning approach is most appropriate for solving a given problem

  • Build production-ready deep neural networks with the NumPy-esque PyTorch library as well as with the heavyweight TensorFlow 2 library (by taking advantage of its in-built, easy-to-use Keras module)

  • Interpret the output of deep learning models to troubleshoot and improve results

Instructor

Instructor Bio:

Chief Data Scientist, Author of Deep Learning Illustrated | untapt

Dr. Jon Krohn

Jon Krohn is Chief Data Scientist at the machine learning company untapt. He authored the 2019 book Deep Learning Illustrated, an instant #1 bestseller that was translated into six languages. Jon is renowned for his compelling lectures, which he offers in-person at Columbia University, New York University, and the NYC Data Science Academy. Jon holds a Ph.D. in neuroscience from Oxford and has been publishing on machine learning in leading academic journals since 2010; his papers have been cited over a thousand times.

Course Outline

Segment 1: The Unreasonable Effectiveness of Deep Learning (40 min)


  • Training Overview 

  • A Brief History of the Rise of Deep Learning

  • Deep Learning vs Other Machine Learning Approaches

  • Dense Feedforward Networks

  • Convolutional Networks for Machine Vision

  • Recurrent Networks for Natural Language Processing and Time-Series Predictions

  • Deep Reinforcement Learning for Sequential Decision-Making

  • Generative Adversarial Networks for Creativity

  • Overview of the Leading Deep Learning Libraries, including TensorFlow 2, Keras, PyTorch, MXNet, CNTK, and Caffe


Segment 2: Essential Deep Learning Theory (80 min)


  • An Artificial Neural Network with Keras

  • The Essential Math of Artificial Neurons

  • The Essential Math of Neural Networks

  • Activation Functions

  • Cost Functions, including Cross-Entropy

  • Stochastic Gradient Descent

  • Backpropagation

  • Mini-Batches

  • Learning Rate

  • Fancy Optimizers (e.g., Adam, Nadam)

  • Glorot/He Weight Initialization

  • Dense Layers

  • Softmax Layers

  • Dropout

  • Data Augmentation

  • TensorFlow Playground: Visualizing a Deep Net in Action 


Segment 3: TensorFlow 2 and PyTorch (90 min)


  • Revisiting our Shallow Neural Network

  • Deep Neural Nets in TensorFlow 2

  • Deep Neural Nets in PyTorch

  • Tuning Model Hyperparameters

  • Creating Your Own Deep Learning Project

  • What to Study Next, Depending on Your Interests

Background knowledge

  • Some experience with machine learning would make this workshop easier to follow, but is by no means necessary.

  • All code demos during the training will be in Python, so experience with it or another object-oriented programming language would be helpful.