Hybrid genetic algorithms for feature selection

Oh, Il‐Seok; Lee, Jinseon; Moon, Byung-Ro

doi:10.1109/tpami.2004.105

Cited by 712 publications

(92 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Thus, they propose hybridizing GA with a steepest descent algorithm and argue that this can positively influence the overall convergence rate. Similar proposals were made in [7], [8].…”

supporting

confidence: 69%

See 1 more Smart Citation

Convergence measurement in evolutionary computation using Price's theorem

Bashir

Neville

2012

2012 IEEE Congress on Evolutionary Computation

View full text Add to dashboard Cite

Abstract-Evolutionary computations are naturally inspired stochastic algorithms that are capable of running perpetually. When deployed as optimization tools, it is imperative to prescribe a set of definitive stopping criteria that if satisfied, the evolutionary process could be brought to a halt. User specified limits on maximum evaluations or generations are the common measures used to stop the evolution due to resource constraints that might directly/indirectly be imposed on the system. Conversely, we propose a novel convergence detection mechanism that monitors the contribution of the genetic operators on the fitness progress and the diversity profile of the population via the ±σ crossover envelope. This adaptively terminates the evolution as convergence sets in. Extended Price's theorem is utilized to estimate the dynamical contributions of the individual genetic operators. Experimental results show that under standard parameter settings with binary tournament selection, the proposed technique is robust and could be a promising alternative to the conventional similarity measure-based methods for convergence detection.

show abstract

“…Thus, they propose hybridizing GA with a steepest descent algorithm and argue that this can positively influence the overall convergence rate. Similar proposals were made in [7], [8].…”

supporting

confidence: 69%

“…Therefore, the following extension (8) to the Price's equation contains only three terms; a term for the selection, crossover and mutation respectively. Each of these terms estimates the changes in the mean of the population's fitness (∆q) due to one of the three genetic operators.…”

Section: A Extension Of Price's Equationmentioning

confidence: 99%

Convergence measurement in evolutionary computation using Price's theorem

Bashir

Neville

2012

2012 IEEE Congress on Evolutionary Computation

View full text Add to dashboard Cite

show abstract

“…Sebelum masuk ke proses GA, proporsi data latih dan data uji harus ditetapkan. Lalu merujuk pada [4] dan [1] maka dilakukan tahap-tahap GA yang telah disesuaikan dengan kebutuhan penelitian ini sebagai berikut.…”

Section: Seleksi Fitur Gaunclassified

Deteksi Kanker berdasarkan Klasifikasi Data Microarray menggunakan Functional Link Neural Network dengan Seleksi Fitur Genetic Algorithm

Ramadhani

2017

IndoJC

View full text Add to dashboard Cite

According to Data and Information Center, Ministry of Health Republic of Indonesia in 2015, cancer is one of the major causes of death in the world. In fact, in 2012 there were 8.2 million people death caused by cancer. The number of death is estimated to increase every year even, in the two next decades will increase 15 million to 22 million. The cancer can be known and detected early have possibilty to get better handling, and microarray technology has a role in it. In recent years, microarray technology has a great impact for determining genes informative causes cancer. Microarray has capability to determine the expression of thousands genes simultaneously monitoring biological processes. By analyzing microarray data, the thousand genes expression that represent a human, would be classified as a cancer or not. In this final project, the author implements Functional Link Neural Network with Legendre Polynomial basis functions for the accurate classification of data and uses Genetic Algorithm as a feature selection to reduce the high dimensional does exist on microarray data. Thus after all the processes have been conducted, then we retrieved the highest performance against the classification of microarray data of Colon Tumors of 92.3% and classification for Leukemia of 87.5%. The differences of performance is caused by the difference of the acquired characteristics of data respectively.Keywords: cancer detection, microarray, Functional Link Neural Network, Genetic Algorithm Abstrak Berdasarkan Pusat Data dan Informasi Kementerian Kesehatan RI Tahun 2015, penyakit kanker merupakan salah satu penyebab kematian utama di seluruh dunia, dan di tahun 2012, sekitar 8,2 juta kematian disebabkan oleh kanker. Diperkirakan penderita kanker terus meningkat setiap tahunnya bahkan dalam dua dekade berikutnya akan meningkat dari 14 juta pada 2012 menjadi 22 juta. Kanker yang dapat diketahui dan dideteksi lebih awal memiliki probabilitas untuk mendapatkan penanganan yang lebih baik, dan teknologi microarray memiliki peran dalam hal itu. Di beberapa tahun terakhir, pemanfaatan teknologi microarray memiliki pengaruh besar dalam menentukan gen informatif yang menyebabkan kanker. Micorarray mampu menentukan ekspresi ribuan gen dan secara simultan memantau proses bilogis yang sedang berlangsung. Dengan melakukan analisa terhadap data micorarray, selanjutnya ekspresi dari ribuan gen yang merepresentasikan suatu jaringan pada manusia, akan diklasifikasikan sebagai jaringan kanker atau bukan. Dalam penelitian ini, penulis mengimplementasikan Functional Link Neural Network dengan fungsi basis Legendre Polynomial untuk klasifikasi data yang akurat dan menggunakan Genetic Algorithm sebagai seleksi fitur untuk mereduksi data berdimensi tinggi yang sering ditemukan pada data microarray. Dengan serangkaian proses yang telah dilakukan, maka diperoleh kinerja tertinggi terhadap klasifikasi data microarray OPEN ACCESS

show abstract

“…The disadvantages are different degrees of similarity cause information loss and algorithm to compute reduction is slow. [17] proposed an unique hybrid genetic algorithm based approach for feature selection. The hybridization technique produces important effects such as a considerable improvement in the final performance and the acquisition of subset-size control.…”

Section: Dsyeung Et Almentioning

confidence: 99%