No

Authors

Title

Publication Info

Published

Files

International Publications:

1

S. Halim,
R.H.C. Yap,
F. Halim

Engineering Stochastic Local Search for the
Low Autocorrelation Binary Sequence Problem

Citation info by Google Scholar (N/A yet)

Principles and Practice of Constraint Programming (CP 2008)

Sydney, Australia
September 14-18, 2008

pp ?-?

LNCS 5xxx?
The book N/A

The paper N/A
 

pdf

poster
N/A

Summary: This is our newest paper, we use the integrated white+black-box approach (Halim et al., 2007) and SLS engineering tool Viz (Halim and Yap, 2007) to analyze the fitness landscape of a hard combinatorial problem: LABS. We improve the implementation of the existing Tabu Search Algorithm by Dotu and van Hentenryck, CP 2006, use a new search strategy, and managed to obtain state-of-the-art results for this COP.

Browse the Viz result page for the details...
 

2

S. Halim,
R.H.C. Yap,
H.C. Lau

An Integrated White+Black Box Approach for Designing and Tuning Stochastic Local Search

Citation info by Google Scholar (N/A yet)

Principles and Practice of Constraint Programming (CP 2007)

Providence, Rhode Island, United States of America
September 23-27, 2007

pp 332-347

LNCS 4741
The book

The paper
 

pdf

ppt

Summary: In this paper, we described that combining white-box approach: visualization (via the tool Viz) - as described in (Halim and Yap, 2007) below - and black-box approach: full factorial design on the narrowed configuration space is effective for dealing with the SLS design and tuning problem.

Browse through these special Viz pages for the latest updates...
 

3

S. Halim,
R.H.C. Yap

Designing and Tuning SLS through Animation and Graphics: an Extended Walk-through

Citation info by Google Scholar

Engineering Stochastic Local Search Algorithms (SLS 2007)

Brussels, Belgium
September 6-8, 2007

pp 16-30

LNCS 4638
The book

The paper
 

pdf

ppt

Summary: In this paper, we describe the ideas and implementation of Fitness Landscape and Search Trajectory (FLST) visualization in visualization tool Viz. Next, we show how to use visualization tool Viz to understand what is happening when a Robust Tabu Search algorithm is exploring the fitness landscape of two different types of Quadratic Assignment Problem instances and then to design and tune the algorithm accordingly. The quality of the visualization tool Viz presented in this paper is improved substantially over our past work in (Halim et al., 2006b).

Browse through these special Viz pages for the latest updates...
 

4

S. Halim,
H.C. Lau

Tuning Tabu Search Strategies via Visual Diagnosis

Citation info by Google Scholar

Metaheuristics International Conference (MIC 2005) post conference volume:

Metaheuristics: Progress in Complex Systems Optimization

pp 365-388

Springer OR-CS
The book

The paper
 

pdf

N/A

Summary: This is a `book chapter' version of the previous conference paper (Lau et al., 2005). It contains the in-depth explanation of our new classification of tuning problem (where we look tuning problem in a broader sense, introducing the term `tuning search strategies'), visual diagnosis tuning methodology (a relatively novel approach for using human-computer interaction for addressing tuning problem), and our preliminary visualization/statistical tool to support the methodology: V-MDF.

We have since update our visualization tool from V-MDF to Viz, see (Halim and Yap, 2007), and refine our 'visual diagnosis tuning' methodology into 'integrated white+black box approach', see (Halim et al., 2007).
 

5

H.C. Lau,
W.C. Wan,
S. Halim,
K. Toh

A Software Framework for Fast Prototyping of Meta-heuristics Hybridization

Citation info by Google Scholar

International Transactions in Operational Research (ITOR)

Volume 14 Issue 2
March, 2007

pp 123-141

Blackwell-Synergy
 

pdf

N/A

Summary: This is the much more in-depth journal version of MDF (Lau et al., 2004). This is my first journal publication, but most credits go to Wan Wee Chong as he is the one who did most of the work for this paper.
 

6

S. Halim,
R.H.C. Yap,
H.C. Lau

Viz: A Visual Analysis Suite for Explaining Local Search Behavior

Citation info by Google Scholar

User Interface System and Technology (UIST 2006)

Montreux, Switzerland
October 15-18, 2006

pp 57-66

ACM
digital library
 

pdf

ppt

video

Summary: We improved our visualization Viz from (Halim et al., 2006a) below into a more complete visualization tool. Acceptance in a premiere UI forum tells something about the usefulness of this information visualization tool.

Browse through these special Viz pages for the latest updates...
 

7

S. Halim,
R.H.C. Yap,
H.C. Lau

Search Trajectory Visualization for Analyzing Trajectory-Based Metaheuristics

Citation info by Google Scholar

European Conference on Artificial Intelligence (ECAI 2006)

Riva del Garda, Italy
August 28-September 6, 2006

pp 703-704 (poster)
 

IOS Press
Books Online

pdf

pdf

Summary: This visualization idea is the major improvement compared to V-MDF and its distance radar, see (Lau et al., 2005) below. We update the meaning of anchor points and now present them in an abstract 2-D space for easier comprehension. The search trajectory will be replayed back with respect to the anchor points in the abstract 2-D space.

Browse through these special Viz pages for the latest updates...
 

8

H.C. Lau,
W.C. Wan,
S. Halim

Tuning Tabu Search Strategies via Visual Diagnosis

Citation info by Google Scholar

Metaheuristics International Conference (MIC 2005)

Vienna, Austria
August 22-26, 2005

pp 630-636
 

CD-ROM only

pdf

ppt

Summary: The main idea of this paper is that while implementing metaheuristics applications is generally easy, tuning its search strategies is difficult... (there is a distinction between tuning metaheuristic parameters and components with tuning search strategies). Especially for industry related problems where there is no or few papers in literature that explains how to attack these problems. Many people who used metaheuristics have headache in finding the best search strategies to attack these problems...

Our idea is to give human a visual diagnostic tool to see what is 'wrong' within the search (its 'run-time dynamics'), and do remedial actions to their implementation, and immediately see the impact of his actions. This formed {cause-action-outcome} rules which can be stored in Knowledge Base. We call this "visual diagnosis tuning". After that, the Knowledge Base will filter the rules collected by human using a certain statistical test (we are expanding this section). The rules that are deemed useful will form the final set of the metaheuristics algorithm.

However, in order for human to diagnose the search on a non-polynomial search space, we need a special visualization tool. We have developed a tool called "Distance Radar" and in this paper, we describe the details and the usage of this radar in determining the effectiveness of the search.

We have since update our visualization tool from V-MDF to Viz, see (Halim and Yap, 2007), and refine our 'visual diagnosis tuning' methodology into 'integrated white+black box approach', see (Halim et al., 2007).
 

9

H.C. Lau,
W.C. Wan,
M.K. Lim,
S. Halim

A Development Framework for Rapid Meta-Heuristics Hybridization

Citation info by Google Scholar

28th International Computer Software and Applications Conference (COMPSAC 2004)

Hong Kong
September 28-30, 2004

pp 362-367
 

IEEExplore

pdf

N/A

Summary: This paper describes the architecture of our metaheuristics software framework (MDF). This framework can be used to build metaheuristic applications rapidly and also able to support hybridization between four supported metaheuristics (Tabu Search, Ants Colony Optimization, Simulated Annealing, and Genetic Algorithm). We illustrate this rapid development and hybridization capability by showing various possible models of hybrids and quickly deduce which one is more effective for solving Traveling Salesman Problem (TSP). For a more detailed information, you may be interested to browse through my team mate (Wan Wee Chong) Master thesis presentation slides.

ERRATA!!!
For those who have read our paper... There are two typo errors:
1. Section 3.7. last two lines:
    "Initial experimentation of has shown..." should be
    "Initial experimentation has shown..."
2. Section 4. Figure 4. Typo...
     Column header "A, B, C, D, E, F, G, H" corresponds to:
    "Pure TS, Pure ACO, HASTS-CC, HASTS-ED, HASTS-CCED, HASTS-IE,HASTS-EA,& HASTS-IEEA",
    respectively...
 

Local Publications:

1

T.D. Trung,
S. Halim,
W.C. Wan, H.C. Lau
 

Solving the 0-1 multidimensional knapsack problem using Tabu Search and Visualization.

17th Singapore Science Research Congress

N/A

pdf

N/A

Summary: This work is part of Science Research Programme conducted by Ministry of Education (MOE) Singapore, where some Junior College (JC) students are attached to faculty members in various tertiary institutions throughout Singapore. My supervisor decided to try this programme. We asked the student (Tuan Dung), to solve an interesting industry problem (MDKP) using our MDF. He managed to understand how MDF works (which show that this MDF framework is understandable by a JC student who has computer programming background), solved MDKP, and obtain a good results.

Trung Tuan Dung managed to obtain gold medal during Singapore science and engineering fair 2005 for this work =).