data structure, ml, machine learning, ml, python, artificial intelligence, ai, computer science, data science, data analytics, technology training at bay area, algorithm, deep learning, gan, general adversarial network, neural learning, deep learning, application, online program, online degree, cstu, online courses
CSE606 AI Application with GAN

This course focuses on deep neural network learning with Generative Adversarial Network (GAN) and introduces some key concepts in deep neural learning. Training Deep learning networks re-quires a good understanding of the nature of gradient descent and its variant, and different forms of loss functions.
  • » 23 hours (8 weeks) in class lecturing plus dedicated mentoring sessions from our faculty of industry experts
  • » 1.5 semester credits for both certificate and master’s degree
  • » Access to high-quality live class recording
  • » Online live classroom available for all classes
  • » Lifetime learning resources for our students
  • ETTP Program $7200
Course Description

This course focuses on deep neural network learning with Generative Adversarial Network (GAN) and introduces some key concepts in deep neural learning. Training Deep learning networks re-quires a good understanding of the nature of gradient descent and its variant, and different forms of loss functions. GAN is a class of machine learning frameworks. Given a training set, GAN learns to generate new data with the same statistics as the training set. A GAN trained on photographs can generate new photographs that look at least superficially authentic to human observers. Though originally proposed as a form of generative model for unsupervised learning, GANs have also proven useful for semi-supervised learning, fully supervised learning, and reinforcement learning. The core idea of a GAN is based on the "indirect" training through the discriminator, which itself is also being updated dynamically.

Course Objectives
  • Learn the basic concepts and algorithms of deep learning;
  • Learn TensorFlow 2.0 step by step by using practical examples and practical tips;
  • Learn how to use TensorFlow framework to build and train GAN and its variant.
University-wide Student Learning Outcomes

The University Student Learning Outcomes assessed and reinforced in this course in-clude but are not limited to the following:

  • Can build a strong connection between the basic concepts/algorithms and TensorFlow API;
  • Can use TensorFlow 2.0 to build various neural network structures, like CNN, DCGAN, pix2pix;
  • Can know how to tune hyper parameters related to both model and algorithm to improve network performance.
Weekly Schedule
Week 1:
  • Introduction of deep learning, from TensorFlow 1.x to 2.0
  • Get familiar with TensorFlow API and learn how to tune the hyper parameters.
  • Understand the basic concepts in deep learning, including overfit/underfit, data preprocessing, weight initialization.
  • First Homework project assignment.
Week2:
  • Quiz 1 of TensorFlow programming.
  • Bet familiar with some important learning algorithms used in deep learning, including back propagation, Gradient Decent.
  • Learn how to use TensorFlow API to compute the gradients and optimize the network weights.
  • 1st Homework submission.
Week3:
  • Review 1st Homework and Quiz.
  • Be familiar with basic and advanced Network structures used in deep learning.
  • Learn how to choose the network structures and estimate the network size based on the datasets and targets.
  • Create your own CNN or RNN, LSTM based on different data sets.
  • Second Homework project assignment.
Week4:
  • Midterm exam
  • Be familiar with Deep Convolutional GAN (DCGAN)
  • Learn how to tune the parameters to train the GAN networks.
  • Hands-on exercise of generating hand written digits based on MNIST datasets.
  • 2nd Homework submission.
Week5.
  • Review 2nd Homework and Quiz.
  • Learn how to visualize the performance of the Generator and Discriminator using Tensorboard.
  • Learn to use different cost functions to optimize the performance.
  • Be familiar with different training datasets and learn how to adapt your GANs based on data sets.
  • 3rd Homework assignment
  • Final Project Kickoff.
Week6.
  • Quiz of GAN basics.
  • Learn how to build conditional GAN to control the generated images.
  • Be familiar with different metrics to measure the network performance.
  • Discussion of project idea and plan.
  • 3rd Homework submission.
Week7.
  • Review of Quiz and homework.
  • Learn Pix2Pix GAN network.
  • Understand backtrace technique to get the perception field.
  • Go through both Generators and Discriminators of Pix2Pix GAN.
  • Project discussion
Week8.
  • Review Final Project submission and presentation.
  • Final exam.
About the Instructor

Gong Danian

Currently work at CADENCE Design Systems as Sr. Principle Engineer,responsible for developing high performance microprocessor; also serves as adjunct professor in CSTU, where he teaches Artificial intelligence and deep learning.

Study and work Experience:

  • 1996 Graduate from Zhejiang University, Hanzhou,bachelor degree in EE major.
  • 2001 Graduate from Tsinghua University, Beijing, Ph.D and master degree in EE major.
  • 2001-present, R&D management and development in Silicon Valley High Tech companies