You are on page 1of 17

國立台灣科技大學

National Taiwan University of Science and Technology

Paper Review

Advisor: Prof. Min-Yuan Cheng


Graduate student: Billy Adhi Poetra
Student ID: M11005842

Department of Civil and Construction Engineering


9/04/2023
Contents

1. Paper Review 1
2. Proposed Idea
3. Research Timeline
4. References

2
Paper Review 1

3
Paper Information

Topic K-means Particle Swarm Optimization with Embedded


Chaotic Search for Solving Multidimensional Problems
Author Min-Yuan Cheng, Kuo-Yu Huang, Hung-Ming Chen
Keyword Particle Swarm Optimization, Chaotic Search, K-
Means Clustering.

4
Research Background & Objective

BACKGROUND:
Introduction • Introduced in 1995, Particle Swarm Optimization have been one of many
successful optimization algorithm until this day. Using pbest & gbest
method, PSO evolution can help user finding the best result for various
optimization problem.
Literature Review • Chaotic Search is defined as deterministic, random-like process use in
non-linear, dynamical system. This system have a non-period, non-
converging and bounded characteristic that helps chaos carry out and
overall search at higher velocities.
Research
Methodology • K-means Clustering is a well-known algorithm for clustering which is
one of the most important and challenging classifying algorithm.
Clustering is the process of dispatching a set of data into groups
according to its similarities.
Model Evaluation
and
Implementation

Conclusion and
Recommendation
5
OBJECTIVES:
• In this paper, K-means Clustering and Chaotic Search are incorporated
Introduction
into PSO algorithm to solve the increasing complexity and
dimensionality of upcoming applications.

Literature Review

Research
Methodology

Model Evaluation
and
Implementation

Conclusion and
Recommendation
6
Methodology

Introduction

Literature Review

Research
Methodology

Model Evaluation
and
Implementation
Figure 1 Methodology diagram (Cheng. 2012)
• K-means clustering procedure helps PSO to converge while the chaotic
Conclusion and
search contribute to refine PSO gbest.
Recommendation
7
Methodology

Introduction

Literature Review

Research
Methodology

Table 1 Configuration for benchmark function


Model Evaluation
and • However, there are some obstacles such as inoperative CPSO algorithm
Implementation
using Rosenbrock function and poor performance of KPSO in solving
Rastrigin and Griewank functions. Besides those problem, KCPSO
manage to solve Rastrigrin function.
Conclusion and
Recommendation
8
Result & Discussion

• The complementary characteristic of K-means, Chaotic search, and PSO are


Introduction
compatible with KCPSO and their contributions had evidently improve the
performance of PSO (Marked in the red box). The result of each benchmark are
shown in tables below:

Literature Review

Research
Methodology

Table 2 Sphere Function

Model Evaluation
and
Implementation

Conclusion and
Recommendation

Table 3 Rosenbrock Function 9


Result & Discussion

Introduction

Literature Review

Table 4 Rastrigin Function

Research
Methodology

Model Evaluation
and
Implementation

Table 5 Griewank Function


Conclusion and
Recommendation
10
Result & Discussion

• Comparison between KCPSO and PSO is presented in table 6. In the


Introduction
previous tables, it shows that the proposed KCPSO is more efficient
based on the media iterations.

Literature Review

Research
Methodology

Table 6 KCPSO vs PSO Comparison


Model Evaluation
and • In conclusion, the novel approach of incorporating K-means and
Implementation
Chaotic Search into PS resulted in the efficiency and effectiveness of
the proposed new algorithm.

Conclusion and
Recommendation
11
2. Proposed Idea

12
Proposed Idea

“Data Feature Selection Using Cluster Integrated Algorithm”


Introduction
• Integrating k-means after data training using prediction inference
model.
• Integrating K-means Clustering method with inference model to
help with feature selection of collected data.
Literature Review • After the running the cluster based feature selection, the data will
be recalculated using the same inference model to evaluate the
improvement of the accuracy.
Research Common Clustering method:
Methodology

DATA CLUSTER TRAINING EVALUATION

Model Evaluation
and
Implementation Proposed Clustering Method:
K-MEANS
HYPOTHESIS
DATA TRAINING TESTING
FEATURE EVALUATION
SELECTION
Conclusion and
Recommendation 13
4. Research Timeline

14
Research timeline

Introduction

Literature Review

Research
Methodology

Model Evaluation
and
Implementation

Conclusion and
Recommendation 15
Research timeline

Introduction

Literature Review

Research
Methodology

Model Evaluation
and
Implementation

Conclusion and
Recommendation 16
References
Cheng, M.Y., Huang, K.Y, Chen, H.M. (2012) K-means particle swarm optimization with
embedded chaotic search for solving multidimensional problems. Applied Mathematics
and Computation. Volume 219, Issue 6. Pages 3091-3099.
ISSN 0096-3003,
https://doi.org/10.1016/j.amc.2012.09.039.

17

You might also like