Rapid Introduction to Machine Learning / Deep Learning


Professor Hyeong In Choi (Seoul National University, Math)


Lectures: Seven lectures delivered biweekly in the fall of 2015 (for dates, see Lecture Schedule)

      Lecture hour starts at 7:30 pm and will last for about two to three hours

Venue: Room 104, Sangsan Mathematical Sciences Bldg., Seoul National University
б а (129) 104ȣ]

Aim & Scope:
The aim of this lecture series is to introduce the basics of current machine learning to those who are interested in deep learning. The intended audience is diverse: they range from practitioners in various industries to researchers in mathematics, statistics, computer science, and engineering. Due to time limitation, we will go over various topics only very briefly, leaving out many important details. Our approach is somewhat theoretical/mathematical, although we skip the proofs in most places.

Audience: Anyone is welcome.



Lecture Schedule (tentative; subject to change)


Lecture 1: (Sept. 18, 2015)

Unit 1a : Introduction

-        Very brief history of machine learning from deep learning perspective

-        Course outline

-        Bare minimum of probability: marginalization and Bayes rule

-        Short demo

Unit 1b : Logistic regression & neural network

-        Probabilistic formalism of logistic regression

-        Symmetric(redundant)-form of log-likelihood function

-        Training

-        Exponential family of distributions and generalized linear model

-        Softmax regression

-        XOR problem & neural network with hidden layer

-      Universal approximation theorem


Lecture 2: (Oct. 2, 2015)

 Unit 2a : SVM & kernel machine

-      Support Vector Machine (SVM)

-      Reproducing Kernel Hilbert Space (RKHS) and kernel machine

-      Deep vs. shallow learning

Unit 2b : Statistical learning theory and its consequences

-      Brief overview of statistical learning theory

-      Overfitting

-      Regularization


Lecture 3: (Oct. 16, 2015)

 Unit 3a : Model selection

-      Bias-variance tradeoff

-      Testing,,validation and model selection

 Unit 3b : Aggregation and randomization

-      Bootstrap

-      Bagging

-      Random Forests


Lecture 4: (Oct. 30, 2015)

 Unit 4a : Feedforward neural network

-      Multilayer perceptron (MLP)

-      Backpropagation algorithm

 Unit 4b: Convolutional network & computer vision

-      Convolution (filter)

-      Convolutional network


Lecture 5: (Nov. 13, 2015)

Unit 5a : Bayesian network

-      Dependency and independency model

-      D-separation

-      Examples

 Unit 5b : Markov random field

-      Energy and probability

-      Boltzmann machine

-      Restricted Boltzmann machine (RBM)


Lecture 6: (Nov. 27, 2015)

 Unit 6a : MCMC 

-      Markov chain and its limiting probability distribution

-      Gibbs sampling

-      Metropolis-Hasting algorithm

 Unit 6b : Unsupervised feature learning : RBM

-      Contrastive divergence (CD)

-      Stacked RMB

-      Deep belief network (DBN)


Lecture 7: (Dec. 11, 2015)

 Unit 7a : Unsupervised feature learning : Auto-encoder & Sparse coding

-      Auto-encoder and denoising auto-encoder (DAE)

Unit 7b : Sparse coding



Post Scriptum:

If demand warrants it, we may organize more lectures/seminars in 2016 on various aspects of deep learning. For example, we can cover some important topics that are left untouched in this lecture series; we may also delve into theoretical/mathematical aspects. But most of all, we want to together explore more practical issues arising in the implementation of or in programming with the likes of Caffe, Torch and Theano.