Joseph Giampapa's IR Term Project

Basic Information

Contents

Abstract

The challenge is to apply a high-accuracy classification method (kNN, DTree or LLSF) to very large category space. The idea is to use a divide-and-conquer strategy to make a large problem more tractable, and  hopefully without significant loss and any loss of classification accuracy.
 
  1. apply kNN sampling strategy using SMART
  2. test on Reuters-21578
  3. compare your results with no-problem decomposition

Proposal and Timelines

Project Proposal and Work Plan
 
 
The below timeline is inaccurate and will be revised shortly.
 
Task
Due Date
Status
 
 1. Project Proposal (given) 
 2.  Begin comparative reading of all references, below. 
 3.  Begin program and data environment setup. 
 
Thu Feb 26
Completed
 1. Personal Project Webpage (given) 
 
Tue Mar  3
Completed
 
 1. Deadline for verifying functionality of program environment. 
 2. Begin classifier and data analysis. 
 3. Begin running systems to gain familiarity with them. 
 
Thu Mar  5
Completed
 1. Deadline for having reviewed all references, below. 
 2. Design first set of experiments for preliminary results. 
 3. Run experiments. 
 
Tue Mar 10
Completed
 1. Run experiments, continued. 
 2. Review experimental results. 
 
Thu Mar 12
Completed
 1. Deadline for completing classifier and data analysis. 
 2. Preliminary Results (given) 
 
Tue Mar 17
Completed
 1. Debugging scripts and retesting 
 
Thu Mar 19
On-going
 1. Setup experiment. 
 2. Begin running final experiment. 
 
Tue Mar 31
On-going
 1. Results analysis. 
 
Thu Apr  2
On-going
 1. Results analysis. 
 
Tue Apr  7
On-going
 1. Considerations for possible demo. 
 2. Begin final project report. 
 
Thu Apr  9
To Begin
 1. Decision about Demo (given)  - No demo.
 2. Possible demo preparation time. 
 3. Final project report (continued). 
 
Tue Apr 14
To Begin
 1. Possible demo preparation time. 
 2. Final project report (continued). 
 
Thu Apr 16
To Begin
 1. Final project report (continued). 
 
Tue Apr 21
To Begin
 1. Final Project Report (given) 
 
Thu Apr 23
To Begin
 1. Project Presentations (given) 
 
Thu Apr 30
To Begin
 1. Project Demos (given, if at all) 
 		 Mon Apr 27 
Tue Apr 28
To Begin

System Description

None Available
 
 

Experiments

... list of exp's, possibly included in Timelines-Table ...
 
 

Results

 

Demo

Not  Planned

last update:  17 April 1998