Spectral clustering - Data mining lectures

Author: mutandon

Created: 2019-09-24 11:28pm

Edited: 2019-12-09 07:33pm

Keywords: clustering, spectral methods, linear algebra, graph mining

Description:

This activity focuses on spectral clustering a central masterpiece in graph analysis. During the next lecture the student will learn what spectral clustering means and how it works, but as a preliminary step the students must read a summary and answer some questions as well as being involved in a in-class discussion after presentation of the content to the class.

Intended Learning Outcomes:

Describe spectral clustering, its limitations and its strengths
Compare spectral clustering with previously presented methods for finding communities.

Resources		Tasks		Supports
Before class
1-2 pages summary of the algorithm presented during the week	→	Read and understand the summary ↓
Questions in blackboard	→	Students answer the questions about the technique ↓	←	Forum support from peers
During class
Slides	→	The lecturer present a case with the algorithm involved and comments on the answers in the test. ↓	←	Lecture presentation and feedback
Use case	→	Discussion in class ↓	←	Mediation of the discussion
After class
Practical exercises	→	Students answer theoretical questions and solves practical exercises in the TA session ↓	←	Support and feedback from the TAs
Study cafè	→	Feedback and clarifications from TAs, with possibility of communicating questions to the lecturer	←	TA and peer answers