Product Description

• Popularity

  One of the most popular technology used by many companies.

• Practicality

  Practical projects set you ready for the industry.

• Workability

  Qualified students will be referred for positions in high-tech industry.

Course Description

Deep Learning is revolutionizing many industries like a thunderstorm, such as auto driving, AlphaGo, neural machine translation, medical diagnosis etc. Deep Learning is a generic and powerful machine learning algorithm. It is widely used by many big companies and startups. It is also one of favorite and most frequently used tool by Kaggle winners (by Kaggle CEO). This Program will provide progressive, interactive and practical lessons for anyone interested in deep learning algorithms and want to apply it to their work or research. The targeted audience is anyone who have basic knowledge in linear algebra and calculus, and knows how to code in python.

Chapter 1

Introduction and Foundation

Desc: Examples of deep learning, from the rising of deep learning and many applications of deep learnings in auto driving, strategic game play, machine translation, biology etc. Inspiration for new applications in your area of interest.

Lab:learn numpy

• Intuitive interactive play with simple neural network to proximate
• Build a simple Logistic regression with Numpy
• Approximate a given numerical function with simple Neural Network in Tensorflow

Chapter 2

Convolutional Neural Network (CNN)

Desc：you will learn convolutional neural network by lots of illustrations and interactive demonstration of each neural activation by a given picture. And how to use CNN for image classification, object detection and image segmentation.

Lab：recognize English characters with Tensorflow

• Tensorflow Installation on Windows, Mac, Linux.
• Recognize English characters in different fonts and styles
• Locate and recognize random placed English characters

Chapter 3

Recurrent Neural Network (RNN) and LSTM

Desc：Modelling sequential dependency with Recurrent Neural Network, use Long and Short Term Memory to capture long-term dependency.  And latest research of meta-learning via LSTM.

Lab：recognize a string of English characters

• Use LSTM and stacked LSTM to recognize characters
• OCR in the wild

Chapter 4

Generative Adversarial Networks (GAN)

Desc：GAN is becoming a powerful tool to tackle unsupervised learning, it can automatically discover essential parameters to generate unlabeled data.

Lab：Pix2Pix

• Modify open source code Pix2Pix to generate fun images

Chapter 5

Deep Reinforcement Learning

Desc: reinforcement learning is used to play Atari Games, self play AlpahGo, control robotic arms and flying drones. RL is an ideal tool for tasks without step by step labeled data, but have an end targeted goal.

Lab：Use RL to Play Tetris

• Setup OpenAI Gym
• Apply RL library build on top of Tensorflow

Chapter 6

Your Own Appliction and Final Project

Desc：In this course, students should be able to work on a practial project they want to do in their own fields. They should learn how to gather data, find related open source project, modify and apply to their own project.

Lab：Build and Improve your own project

Xiaoming Tian

Professor of USJ
Xiaoming (Andy) Tian is a professor of University of San Jose. Combined his Master’s Degrees from USC and CMU, with years of experiences of product demand forecasting and inventory management using big data infrastructure and deep learning technology in Walmart Labs, Andy Tian makes himself an expert on Deep Learning Technology.