By Richard Xu, Associate Professor of Machine Learning & Director, Industry Analytics & Visualisation, University of Technology Sydney, Australia
Machine Learning Course for Industry
Here is the Machine Learning course I designed for Industry Data Scientists; Some of these topics also forms part of my research (e.g., Deep Learning).
Intro to Data Analytics and Machine LearningThis is a warmup with some terminologies; Along with the notes, there are heaps of demos for research and REAL WORLD industry applications.
Intro to Bayesian StatisticsExplain common probability concepts such as marginal, conditional distribution and Bayes formula etc
OptimizationGoing through the basic idea behind iterative algorithms such as Stochastic Gradient Descend
RegressionLinear, Polynomial, Mixed Effects and some classification algorithm bear the name of regression: multinomial regression
Dimensional ReductionDimensionality Reduction techniques such as Principal Component Analysis and t-SNE
Neural Networks from Shallow to DeepThe notes intend to become a complete Neural Network course from the basics to some of the state-of-the-art and with some applications.
Deep Neural Network Learning: The Last LayerThe notes is what I presented in the third Deep Learning Sydney Meetup, including Softmax, Centre, Linear, Contrastive, Triplet loss functions.
Deep Reinforcement Learning newThe notes is what I presented in the May 2017 Deep Learning Sydney Meetup, mainly explaining Google Deepmind’s Deep Q-Learning
Machine Learning Course for Researcher and PhD students
I started to develop this machine learning course since 2009, it’s constantly being updated. I enjoy sharing my knowledge with other researchers and industry practitioners. These notes are written in English.
大家好,从2015年10月开始,我为国内的同行,同学们做了一系列概率机器学习的视频公开课。我每星期都会更新。为了方便大家,我是用中文讲的. The English version of my machine learning MOOCS is also on its way…
Probabilities and EstimationsVarious common probability distribution (functions), natural parameters and Maximum Likelihood & Maximum A Posterior Estimation
Statistical PropertiesA few useful statistical properties include inequalities, convergence and uniqueness
Expectation-MaximizationProof of convergence for E-M, examples of E-M through Gaussian Mixture Model
Introduction to Monte Carlo methodsOverview of Sampling techniques, including Rejection, Adaptive Rejection, Importance, Markov Chain Monte Carlo (MCMC), Gibbs, Bootstrap Particle Filter, and Auxiliary Particle Filter
Stochastic matricesStochastic matrices, Power Method Convergence and Page Rank
Markov Chain Monte CarloMarkov Chain Monte Carlo (MCMC), Metropolis Hasting, Gibbs, Slice Sampling, Collapsed Gibbs, Swendsen-Wang, Elliptical Slice Sampling, (Collapsed) Gibbs sampling examples on LDA
Variational InferenceExplain Variational Bayes in two parts: non-exponential and exponential family distribution plus stochastic variational inference.
Statistical Models
Baysian Non-Parametrics Basics and InferenceDirichlet Process, Hieratical Dirichlet Process, HDP-HMM, Indian Buffet Process and their inference methodologies using Gibbs and Slice Gibbs.