NUS HomeResults
Home | Up | TSP | QAP | LABS


Results

So far, we have used IWBBA using Viz tool to design, tune, debug, improve, understand various SLS algorithms for an array of COPs:

1. Traveling Salesman Problem (TSP)
A very well-known COP.
We exploit proper intensifications (and minor diversifications)
on "Big Valley" region in the TSP fitness landscape.
SLS used: Iterated Local Search and Tabu Search.


 

2. Quadratic Assignment Problem (QAP)
There are more than one QAP fitness landscape structure!
We set more intensifications on the smooth one.
We set more stronger diversifications on the rugged one.
SLS used: Robust Tabu Search.


 

3. Low Autocorrelation Binary Sequence (LABS)
COP with nasty fitness landscape structure, is it really like finding a needle in haystack or are there any exploitable structure?
We obtained state-of-the-art LABS solver (Apr 2008-?).
SLS used: Tabu Search with frequent local restarts.


 

4. Future Works...
We are working with COP with constraints...
What if the fitness landscape has infeasible regions?
What is the best way to traverse it?
SLS used: N/A.


 

Notes: What we show here are successful usage of Viz. Sometimes, we do not manage to get any insights after analyzing our SLS using Viz. Remember that the information visualization tool in Viz is designed to maximize human (SLS algorithm designer) visual perception ability in finding interesting patterns in SLS behavior on COP fitness landscape. However, we do not guarantee that the user will always manage to gain insights about his SLS behavior. However, as what we have shown in this page, gaining such insights using Viz is possible. One can get such insights with higher probability when using Viz than when analyzing SLS behavior using text-based output only. If you still fail to spot interesting behavior, try again, do not give up.

On the other hand, you may get insights to debug, tune, and improve your SLS algorithm after analyzing it using Viz. If that happens, congratulations =). You may want to write a scientific paper and publish it in good conferences or journals. However, please do not forget to cite Viz! (the appropriate article can be found in our publication page). Also, please let us know about your findings too, we will be very happy to know your success stories.

PS: Other than white-box analysis, you can also use the black-box tool in Viz to help you find the best working configuration for your SLS.

This document, results.html, has been accessed 8760 times since 08-May-06 13:23:45 SGT. This is the 1st time it has been accessed today.

A total of 3373 different hosts have accessed this document in the last 4820 days; your host, ec2-54-172-234-236.compute-1.amazonaws.com, has accessed it 1 times.

If you're interested, complete statistics for this document are also available, including breakdowns by top-level domain, host name, and date.