EENG 4010 “A Numerical Approach to Pattern Recognition”

Lesson 1 with all files
This includes not only the first lesson but also some other MATLAB Exercises. A broad view of the central problem of pattern recognition. /
Lesson 2 Classifiers Based on Bayes Decision Theory
Bayes Theorem and its applications is the basis for one of the fundamental principles of classification. /
Lesson 3 When we do not know the pdf
In this lesson we cover Parametric and Non Parametric Estimation. In the Parametric Estimation category we cosider Maximum Likelihood estimation (MLE), Mixture ...more /
Lesson 4 Toolbox, data fitting and an assignment
This is a review of the Statistical Toolbox in Matlab which offers a variety of programs to evaluate and graph distributions. The data fitting capabilities are...more /
Lesson 5
In this lesson we consider linear and nonlinear classifiers. We also consider multivariate classification as an extension of univariate and begin our considerat...more

Lesson 6 We continue on Gaussian Mixtures
We go in greater detail on the use of Gaussian Mixture Models and on the decomposition schemes utilizing the EM algorithm studied in earlier lessons. We work on...more /
Lesson 7 Unsupervised classification
If we do not have labels for our data we have to approach the classification with what are called unsupervised approaches. Two stand out: the k-means and the h...more /
Lesson 8: Pattern Recognition, AI and Prob and Stat
Learning Machines, Decision Theory and Pattern Recognition go to similar goals with different terminologies but the same math. /
Lesson 9: Separability, Perceptrons and SVMs
In this lesson we cover sections 3.1, 3.2, 3.3 and 3.6 of the second edition of Theodoridis, second edition. That material involves the XOR problem, Perceptron...more /
Lesson 10: Dimensionality Reduction
We study a few approaches to attempting to reduce the size of the number of features considered (the dimension of the rows of the samples collected. One is the...more

Lesson 11: More on Kernels and Duality
We define a Kernel function and consider the duality between the feature based classification approaches and the kernel based classification methods. Kernel me...more /
Lesson 12: Cleaning and Selecting Features
We have seen methods of dimensionality reduction. In general we can accomplish it by 1) reducing the number of features or by 2) selecting among the number of ...more /
Lesson 13 Some topics in Image Processing
The subject of Image Processing is complex and extensive. It is particularly complex when we move from two dimensional images to time changing frames of images....more /
Lesson 14 -HMM and Matlab
This is a brief overview of the HMMs approach to speech and other time flow /
Lessson 15 Vector quantization
This is a brief introductin to Vector Quantization Encoding. When presented with a test sample it must againg be compared with the "representatives" of the cod...more