Page 9 - Fister jr., Iztok, and Andrej Brodnik (eds.). StuCoSReC. Proceedings of the 2017 4th Student Computer Science Research Conference. Koper: University of Primorska Press, 2017
P. 9
le 8: Comparison and ranking based on Computational 31–34, 2015.
Time (CT), Mean Fitness value (Fitm), Standard Deviation
((Fitstd) and PSNR for 4-level multi-thresholding. [4] Iztok Fister Jr, Uroˇs Mlakar, Xin-She Yang, and Iztok
Fister. Parameterless bat algorithm and its
Alg. CT Fitm Fitstd PSNR performance study. In Nature-Inspired Computation in
27.7275(1) 0 (1) Engineering, pages 267–276. Springer, 2016.
PL-1 21.22 (7) 27.7275(1) 0 (1) 19.05 (1)
PL-2 21.41 (8) 27.7275(1) 0 (1) 19.05 (1) [5] Iztok Fister Jr., Xin-She Yang, Iztok Fister, Janez
PL-3 21.02 (5) 27.7275(1) 0 (1) 19.05 (1) Brest, and Duˇsan Fister. A brief review of
PL-4 20.82 (4) 27.7256(6) 2.0100e-12 (6) 19.05 (1) nature-inspired algorithms for optimization.
PL-5 21.04 (6) 27.7259(5) 2.0093e-12 (5) 18.85 (6) Elektrotehniˇski vestnik, 80(3):116–122, 2013.
PL-6 20.56 (3) 27.7239(7) 3.0303e-12 (7) 18.87 (5)
PL-7 20.54 (2) 27.7194(8) 1.0001e-11 (8) 18.83 (7) [6] Teo J. and M.Y. Hamid. A parameterless differential
PL-8 19.91 (1) 18.83 (7) evolution optimizer. In 5th International Conference
on Systems Theory and Scientific Computation
Table 9: Comparison and ranking based on Computational (ISTASC’05), Malta, pages 330–335, 2005.
Time (CT), Mean Fitness value (Fitm), Standard Deviation
(Fitstd) and PSNR for 5-level multi-thresholding. [7] Fernando G Lobo and David E Goldberg. An overview
of the parameter-less genetic algorithm. Urbana,
Alg. CT Fitm Fitstd PSNR 51:61801, 2003.
31.6975 (1) 0 (1)
PL-1 21.93 (7) 31.6975 (1) 0 (1) 20.40 (1) [8] Diego Oliva, Erik Cuevas, Gonzalo Pajares, Daniel
PL-2 20.88 (2) 31.6959 (3) 2.1044e-15 (3) 20.40 (1) Zaldivar, and Marco Perez-Cisneros. Multilevel
PL-3 21.73 (6) 31.6957 (4) 1.0030e-13 (5) 20.34 (3) thresholding segmentation based on harmony search
PL-4 21.55 (5) 31.6952 (5) 1.9001e-13 (7) 20.34 (3) optimization. Journal of Applied Mathematics, 2013,
PL-5 21.46 (4) 31.6937 (6) 1.7703e-13 (6) 20.29 (5) 2013.
PL-6 21.02 (3) 31.6930 (7) 2.0030e-13 (8) 20.24 (6)
PL-7 21.94 (8) 31.6862 (8) 2.0011e-14 (4) 20.23 (7) [9] Soham Sarkar, Sujoy Paul, Ritambhar Burman,
PL-8 20.45 (1) 20.20 (8) Swagatam Das, and Sheli Sinha Chaudhuri. A fuzzy
entropy based multi-level image thresholding using
criteria have been used here for analysis the efficiency of differential evolution. In International Conference on
the PLHSs. Analysis of the experimental results prove that Swarm, Evolutionary, and Memetic Computing, pages
PLHSs with lower population size are better for maximiz- 386–395. Springer, 2014.
ing the Shannon’s entropy based objective function with less
standard deviation but with more computational time when [10] Ruhul Amin Sarker and MF Azam Kazi. Population
Iteration based stopping criterion is used. Larger popula- size, search space and quality of solution: An
tion size helps to reduce the computational time, but may experimental study. In Evolutionary Computation,
performs premature convergence. In the case of M AX F E 2003. CEC’03. The 2003 Congress on, volume 3,
based stopping criterion, HS with population size ∈ [40, pages 2011–2018. IEEE, 2003.
160] gives best and consistent output. Here large popu-
lation size also affect the stability issue. But, M AX F E [11] LA Silva, PB Ribeiro, GH Rosa, KAP Costa, and
based stopping condition assists to reduce the stability issue Joa˜o Paulo Papa. Parameter setting-free harmony
with large computational time compare to iteration based search optimization of restricted boltzmann machines
termination condition. Therefore, development of robust and its applications to spam detection. In 12th
adaptive nature-inspired optimization algorithms algorithms International Conference Applied Computing, pages
where all parameters including population size and stopping 142–150, 2015.
criterion for a set of problems are automatically adapted is
still a big problem in this optimization field. In the future, [12] Xin-She Yang. Nature-inspired optimization
an extensive study and systematic analysis of the parame- algorithms. Elsevier, 2014.
ters of different nature-inspired optimization algorithms are
needed over a different set of problems.
6. REFERENCES
[1] Thomas Back, David B. Fogel, and Zbigniew
Michalewicz, editors. Handbook of Evolutionary
Computation. IOP Publishing Ltd., Bristol, UK, UK,
1st edition, 1997.
[2] Ashish Kumar Bhandari, Anil Kumar, S Chaudhary,
and Girish Kumar Singh. A novel color image
multilevel thresholding based segmentation using
nature inspired optimization algorithms. Expert
Systems with Applications, 63:112–133, 2016.
[3] Iztok Fister Jr, Iztok Fister, and Xin-She Yang.
Towards the development of a parameter-free bat
algorithm. In StuCoSReC: Proceedings of the 2015 2nd
Student Computer Science Research Conference, pages
StuCoSReC Proceedings of the 2017 4th Student Computer Science Research Conference 9
Ljubljana, Slovenia, 11 October
Time (CT), Mean Fitness value (Fitm), Standard Deviation
((Fitstd) and PSNR for 4-level multi-thresholding. [4] Iztok Fister Jr, Uroˇs Mlakar, Xin-She Yang, and Iztok
Fister. Parameterless bat algorithm and its
Alg. CT Fitm Fitstd PSNR performance study. In Nature-Inspired Computation in
27.7275(1) 0 (1) Engineering, pages 267–276. Springer, 2016.
PL-1 21.22 (7) 27.7275(1) 0 (1) 19.05 (1)
PL-2 21.41 (8) 27.7275(1) 0 (1) 19.05 (1) [5] Iztok Fister Jr., Xin-She Yang, Iztok Fister, Janez
PL-3 21.02 (5) 27.7275(1) 0 (1) 19.05 (1) Brest, and Duˇsan Fister. A brief review of
PL-4 20.82 (4) 27.7256(6) 2.0100e-12 (6) 19.05 (1) nature-inspired algorithms for optimization.
PL-5 21.04 (6) 27.7259(5) 2.0093e-12 (5) 18.85 (6) Elektrotehniˇski vestnik, 80(3):116–122, 2013.
PL-6 20.56 (3) 27.7239(7) 3.0303e-12 (7) 18.87 (5)
PL-7 20.54 (2) 27.7194(8) 1.0001e-11 (8) 18.83 (7) [6] Teo J. and M.Y. Hamid. A parameterless differential
PL-8 19.91 (1) 18.83 (7) evolution optimizer. In 5th International Conference
on Systems Theory and Scientific Computation
Table 9: Comparison and ranking based on Computational (ISTASC’05), Malta, pages 330–335, 2005.
Time (CT), Mean Fitness value (Fitm), Standard Deviation
(Fitstd) and PSNR for 5-level multi-thresholding. [7] Fernando G Lobo and David E Goldberg. An overview
of the parameter-less genetic algorithm. Urbana,
Alg. CT Fitm Fitstd PSNR 51:61801, 2003.
31.6975 (1) 0 (1)
PL-1 21.93 (7) 31.6975 (1) 0 (1) 20.40 (1) [8] Diego Oliva, Erik Cuevas, Gonzalo Pajares, Daniel
PL-2 20.88 (2) 31.6959 (3) 2.1044e-15 (3) 20.40 (1) Zaldivar, and Marco Perez-Cisneros. Multilevel
PL-3 21.73 (6) 31.6957 (4) 1.0030e-13 (5) 20.34 (3) thresholding segmentation based on harmony search
PL-4 21.55 (5) 31.6952 (5) 1.9001e-13 (7) 20.34 (3) optimization. Journal of Applied Mathematics, 2013,
PL-5 21.46 (4) 31.6937 (6) 1.7703e-13 (6) 20.29 (5) 2013.
PL-6 21.02 (3) 31.6930 (7) 2.0030e-13 (8) 20.24 (6)
PL-7 21.94 (8) 31.6862 (8) 2.0011e-14 (4) 20.23 (7) [9] Soham Sarkar, Sujoy Paul, Ritambhar Burman,
PL-8 20.45 (1) 20.20 (8) Swagatam Das, and Sheli Sinha Chaudhuri. A fuzzy
entropy based multi-level image thresholding using
criteria have been used here for analysis the efficiency of differential evolution. In International Conference on
the PLHSs. Analysis of the experimental results prove that Swarm, Evolutionary, and Memetic Computing, pages
PLHSs with lower population size are better for maximiz- 386–395. Springer, 2014.
ing the Shannon’s entropy based objective function with less
standard deviation but with more computational time when [10] Ruhul Amin Sarker and MF Azam Kazi. Population
Iteration based stopping criterion is used. Larger popula- size, search space and quality of solution: An
tion size helps to reduce the computational time, but may experimental study. In Evolutionary Computation,
performs premature convergence. In the case of M AX F E 2003. CEC’03. The 2003 Congress on, volume 3,
based stopping criterion, HS with population size ∈ [40, pages 2011–2018. IEEE, 2003.
160] gives best and consistent output. Here large popu-
lation size also affect the stability issue. But, M AX F E [11] LA Silva, PB Ribeiro, GH Rosa, KAP Costa, and
based stopping condition assists to reduce the stability issue Joa˜o Paulo Papa. Parameter setting-free harmony
with large computational time compare to iteration based search optimization of restricted boltzmann machines
termination condition. Therefore, development of robust and its applications to spam detection. In 12th
adaptive nature-inspired optimization algorithms algorithms International Conference Applied Computing, pages
where all parameters including population size and stopping 142–150, 2015.
criterion for a set of problems are automatically adapted is
still a big problem in this optimization field. In the future, [12] Xin-She Yang. Nature-inspired optimization
an extensive study and systematic analysis of the parame- algorithms. Elsevier, 2014.
ters of different nature-inspired optimization algorithms are
needed over a different set of problems.
6. REFERENCES
[1] Thomas Back, David B. Fogel, and Zbigniew
Michalewicz, editors. Handbook of Evolutionary
Computation. IOP Publishing Ltd., Bristol, UK, UK,
1st edition, 1997.
[2] Ashish Kumar Bhandari, Anil Kumar, S Chaudhary,
and Girish Kumar Singh. A novel color image
multilevel thresholding based segmentation using
nature inspired optimization algorithms. Expert
Systems with Applications, 63:112–133, 2016.
[3] Iztok Fister Jr, Iztok Fister, and Xin-She Yang.
Towards the development of a parameter-free bat
algorithm. In StuCoSReC: Proceedings of the 2015 2nd
Student Computer Science Research Conference, pages
StuCoSReC Proceedings of the 2017 4th Student Computer Science Research Conference 9
Ljubljana, Slovenia, 11 October
