Canonical Correlation Analysis

In statistics, canonical-correlation analysis (CCA) is a way of inferring information from cross-covariance matrices. If we have two vectors X = (X_1, ..., X_n) and Y = (Y_1, ..., Y_m) of random variables, and there are correlations among the variables, then canonical-correlation analysis will find linear combinations of the X_i and Y_j which have maximum correlation with each other.

This repository contains a group of study material of CCA related algorithm.

Repo

Notes

Notebook Note.pdf

The Notebook contains topics:

1. CCA: Implementation in Numpy
2. KCCA
3. DCCA
4. DCCAE
5. VCCA: Derivation of the variational lower bound
6. VCCA-P
7. VAE: ELBO derivation
8. CTC

VCCA Pytorch Implementation

code

Paper

paper file

VCCAP Tensorflow code

code

CCA

CCA slides 1 CCA slides 2 CCA notes

Multi-view clustering via canonical correlation analysis

Kernel-CCA [KCCA]

Canonical Correlation Analysis: An Overview with application to learning methods

Deep-CCA [DCCA]

Deep Canonical Correlation Analysis

On Deep Multi-View Representation learning DCCAE

Variational CCA and Variational CCA Private [VCCA, VCCAP]

Deep Variational Canonical Correlation Analysis

Acoustic Feature Learning via Deep Variational Canonical Correlation Analysis

Variational Auto Encoder [VAE]

[Tutorial on Variational Autoencoders](https://github.com/edchengg/VCCA-StudyNotes/blob/master/paper/VAE.pdf)

Ganerative Adversarial Networks [GAN]

NIPS 2016 Tutorial: Generative Adversarial Networks

Connectionist Temporal Classification [CTC]

CTC slides 60/100

GMM/HMM

50/100