Pattern Recognition
Course Outline
Pattern recognition deals with the problem of finding the characteristics of a given data set by using a prior knowledge or statistical information of the data, and to classify the data into several categories on the basis of the characteristics. The dimension of the data is usually very high to visualize. When there is a priori information about the data, some typical methods in pattern recognition are the Bayes classifier, maximum likelihood method, discriminant analysis and artificial neural networks. When there is no prior information, data can be classified by clustering methods. Pattern recognition is used in bio recognition, such as face and fingerprint, and also in search for new knowledge in vastly accumulated data. This course introduces several typical methods used in pattern recognition and deals with its application.
Goals
Fundamental techniques for pattern recognition are introduced, including Learning Rules (Cross Entropy, Mutual Information, ICA, SupportVector Machines, Error Backpropagation rules), Bayesian Frameworks(Bayesian Decision/Inference, Bayesian Networks, Parametric /NonparametricDensity Estimation), Regression Analysis(Linear/Nonlinear Regression, PCA, Partial Regression, Gaussian Process Regression, Kalman Filter).
Syllabus
Week 01: Lecture Introduction, Linear Algebra for Machine Learning
Week 02: Matrix-Vector Derivative, Cross Entropy, Mutual Information
Week 03: Optimization for Machine Learning
Week 04: Support Vector Machine, Kernel Tricks
Week 05: Error Backpropagation Rule for Neural Network Training
Week 06: Bayesian Decision, Parametric Density Estimation
Week 07: Nonparametric Density Estimation, Expectation-Maximization
Week 08: Midterm Exam
Week 09: Makov-Chain Monte Carlo(MCMC) Method, Boltzmann Machine
Week 10: Bayesian Networks
Week 11: Bayesian Inference via MCMC for Unsupervised Clustering
Week 12: Variational Inference, Traffic Pattern Analysis
Week 13: Linear/Nonlinear Regression, Principal Component Analysis
Week 14: Gaussian Process Regression, Kalman Filter
Week 15: Final Exam
