from machine learning to deep learning. topics that i will cover (subject to some minor adjustment)...

7
From Machine Learning to Deep Learning

Upload: lily-bryant

Post on 01-Jan-2016

217 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

From Machine Learning to Deep Learning

Page 2: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Topics that I will Cover(subject to some minor adjustment)

Week 2: Introduction to Deep LearningWeek 3: Logistic RegressionWeek 4: Multi-Layer Perceptron and Back PropagationWeek 5: Deep Learning (1) Week 6: Deep Learning (2)

Page 3: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Week 2: Introduction to Deep Learning

• Deep learning can be viewed as a recent break-through in computer science

• In this lecture, I will introduce you a wide range of applications of Deep Learning techniques

Page 4: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Week 3: Logistic Regression

• In order to understand Deep Learning, we need to first understand Artificial Neural Network

• We will start with a simple Artificial Neural Network: Logistic Regression

• We will talk about applying Logistic Regression to classification problems, and derive a learning algorithm that is called Iteratively Reweighted Least Squares (IRLS) for Logistic Regression

Page 5: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Week 4: Multi-Layer Perceptron and Back Propagation

• In this lecture, we will develop a multi-layer perceptron based on Logistic Regression and introduce the back propagation algorithm for training multi-layer perceptrons

Page 6: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Week 5: Deep Learning (1)

• In this lecture, we will talk about the issues of applying back-propagation algorithm to neural network with deep architectures

• Then will talk about unsupervised approaches for learning the weights for a deep neural networks layer by layer, which leads to different deep learning approaches

Page 7: From Machine Learning to Deep Learning. Topics that I will Cover (subject to some minor adjustment) Week 2: Introduction to Deep Learning Week 3: Logistic

Week 6: Deep Learning (2)

• We will continue with week 5’s introduction on deep learning to further talk about the design, implementation, and applications of deep learning.

• I will talk about some of my own research work in this area, and suggest a couple of research directions that can work as Ph.D. dissertation topics.