# Meeting #1 (25/11/2021)—Introduction
11:00-12:00, Room 5001/B54

## Topics:

0. Introduction
1. Timeline, future meeting arrangements
2. Projects ideas
3. Reports write-up and presentation
4. Unsupervised learning

## 0. Introduction

## 1. Timeline and Arrangement

**Mid-term report/outline and presentation**: due on *13th Jan*
**Final Report and presentation**: due on *5th May*

### - Semester 1

* Thursday, 2nd Dec: individual meetings
*Location: 9001/B54 (my office)*
11:00-11:30 Tridham
11:30-12:00 Yuhao
12:00-12:30 Tabitha
15:00-15:30 Quentin

* Thursday, 16th Dec(11:00-12:00): group meeting
*Location: 5001/B54 (the Conference Room)*
Presentation by Chao: Density Based Clustering via Kernel Diffusion

* Thursday, 13th Jan 2022(11:00-13:00): drop-in session
*Location: 5001/B54 (the Conference Room) or MS Teams*
Drop in if you have any questions to ask/discuss (attendance strongly recommended, but not compulsory).

### - Semester 2

Biweekly meetings in following forms: (time&rooms TBD)
1. group meeting (each of your report your project progress)
2. drop-in discussion session
...

## 3. Reports Write-Up

Learn LaTeX in 30 minutes
https://www.overleaf.com/learn/latex/Learn_LaTeX_in_30_minutes

List of Math Symbols
https://www.caam.rice.edu/~heinken/latex/symbols.pdf

Overleaf (for online and collaborating projects)
http://www.overleaf.com/

R markdown:
Introduction: https://rmarkdown.rstudio.com/articles_intro.html
Cheatsheet: https://www.rstudio.com/wp-content/uploads/2015/02/rmarkdown-cheatsheet.pdf

## 4. Unsupervised Learning

Difference machine learning problems:
* Supervised Learning
* Reinforcement Learning
* Unsupervised Learning

### - Clustering
$K$-Means, density-based clustering, hierarchical clustering, spectral clustering,....

Modern Development: semi-supervised/contrastive learning,....

### - Dimension Reduction/Principal Component Analysis
PCA:
\begin{align}
\text{minimise}\quad &||X-L||_2\\
\text{subject to } \quad &rank(L)\le k,
\end{align}
MDL, random projection,...

Modern Development: sufficient dimension reduction, SIR, robust/sparse PCA, PCA for high-dimensional data, manifold learning,....

### - Applications
Video Surveillance
Face Recognition
Computational Vision
......