# Machine Learning

This section will talk about some algorithms commonly used for machine learning.

## Articles

Here's some stuff for machine learning.

• ### A Guide to Principal Component Analysis (PCA)

PCA is a widely used operation for decorrelating data, whitening, and dimensionality reduction. Given its array of uses, a good understanding of exactly how it works is helpful.

• ### A Tutorial on Cepstrum and LPCCs

LPCCs are Linear Prediction Cepstral Coefficents, this tutorial explains what the Cepstrum is and how LPCCs are computed as well as why they are useful.

• ### A tutorial on Automatic Language Identification - ngram based

This page deals with automatically classifying a piece of text as being a certain language. A training corpus is assembled which contains examples from each of the languages we wish to identify, then we use the training information to guess what language a set of test sentences is in.

• ### A tutorial on Automatic Language Identification - word based

This page deals with automatically classifying a piece of text as being a certain language. A training corpus is assembled which contains examples from each of the languages we wish to identify, then we use the training information to guess what language a set of test sentences is in.

• ### An Intuitive Discrete Fourier Transform Tutorial

The Discrete Fourier Transform is widely used in signal processing. To beginners in the field, the equation can be confusing. This page tries to convey an intuitive understanding of how the DFT does what it does.

• ### Approximating a function with a polynomial

This page describes approximation of functions with polynomials. It is relies on using polynomials to interpolate a function, but the trick is to choose which points to use to interpolate between so that the error between the function and the interpolating polynomial is minimised.

• ### Documentation for matlab_speech_features

this is documentation for the matlab_speech_features matlab library.

• ### Encoding Variables For Neural Networks

Deep Neural Networks have become more popular over the last few years, this page will deal with methods for encoding target variables e.g. what if you want to predict categories instead of real numbers? what about angles?

• ### Fitting a polynomial to a set of points

In this page we will go over some of the main ways polynomials can be found that go through a set of points.

• ### Gaussian Mixture Model Tutorial

This is an intuitive guide to Gaussian Mixture Models along with an worked example applying them to a speaker identification problem.

• ### Graphically Determining Backpropagation Equations

The hardest part about implementing neural networks is figuring out the backpropagation equations to train the weights. This article goes through a simple graphical method for deriving the equations.

• ### Hidden Markov Model (HMM) Tutorial

Hidden Markov Models are used to model sequences probabilistically. They are related to Markov chains, but are used when the observations don't tell you exactly what state you are in.

• ### Implementing the Dolph-Chebyshev Window

The Dolph-Chebyshev Window is commonly used in signal processing. Its advantages include constant height side lobes and tunable attenuation.

• ### Linear Prediction Tutorial

Linear Prediction is an important tool for time series analysis, and is quite simple to understand too. This article provides an intuitive introduction to the topic.

• ### Mel Frequency Cepstral Coefficient (MFCC) tutorial

Mel Frequency Cepstral Coefficients, or MFCCs, are a type of feature widely used in speech and speaker recognition systems. This page tells you how to calculate MFCCs, and why MFCCs are the way they are i.e. how do steps involved in calculating MFCCs relate to psychoacoustic properties of the human ear and brain.

• ### Speech Enhancement tutorial: Spectral Subraction

Spectral Subraction is one of the simplest speech enhancement algorithms. Its results are not great, but it is a good starting point for getting into more advanced speech enhancement algorithms.

• ### Voice Activity Detection (VAD) Tutorial

The job of Voice Activity Detection (VAD) is to accurately discriminate between speech presence and speech absence. This page presents the basics of VAD.

• ### Yet Another K-Means Tutorial

The purpose of K-means is to identify groups, or clusters of data points in a multidimensional space. This tutorial covers K-means in a hopefully intuitive manner.

GQQ RPIGD GSCUWDE RGJO WDO WT IWTO WA CROEO EOJOD SGPEOE: SRGDSO, DGCPTO, SWIBPQEUWD, RGFUC, TOGEWD, BGEEUWD GDY YOEUTO