Yazar "Avcu, Fatih Mehmet" seçeneğine göre listele
Listeleniyor 1 - 5 / 5
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe Classification of some chemical drugs by genetic algorithm and deep neural network hybrid method(Wiley, 2021) Karakaplan, Mustafa; Avcu, Fatih MehmetDeep neural networks (DNN) and genetic algorithm (GA) are gaining importance quickly with many successful applications in the field of science and technology. They are indispensable tool for the numerical solution of difficult problems. It is possible to optimize DNNs using the GA and this combination can be used to classify data. In this article, some drugs are classified by Monte Carlo sampling with combination of GA and DNN due to stochastic nature of the domain, exponential number of variables and small number of chemical species. In addition to the values obtained from the databases of selected drugs, molecular dynamic and ab initio molecular mechanical calculation results were also used. The aim of this study is to generalize the molecular classification with the data obtained from chemical databases as well as molecular docking results by using the combination of deep learning and GA and its usability in drug design. The selected drugs are some agonist and antagonist drugs that bind to dopamine receptors, which are widely studied and well known in the literature. To train the DNN, input datasets were chosen by the GA framework written in pure Python named PyEvolve. Classification of drugs has been analyzed with the focus on orbital energies and docking results. It is possible to use this algorithm in many in silico calculations such as affinity and separation processes. The reliability of the algorithm was tested with the results given in the literature and the expected values were estimated at 93.8%.Öğe Deconvolution of Gaussian peaks with mixed real and discrete-integer optimization based on evolutionary computing(Wiley, 2020) Karakaplan, Mustafa; Avcu, Fatih MehmetThis study describes an alternative method for deconvolution of overlapping characteristic Gauss peaks with the help of optimization of a mixed variable genetic algorithm. Continuous and discrete variables and nonlinear discrete variables in optimization problems cause solution complexity. The processing and analysis of complex analytical signals is important not only in analytical chemistry but also in other fields of science. As the amount of data increases and linearity decreases, high-performance computations are needed to solve analytical signals. It takes a long time to perform these calculations with traditional processor systems and algorithms. We have used NVIDIA graphical processing units (GPUs) to shorten the duration of these calculations. Solving such analytical signals with genetic algorithms is widely used in computational sciences. In this study, we present a new curve-fitting method using a genetic algorithm based on Gauss functions used to deconvolve overlapping peaks and find the exact peak number in absorption spectroscopy. The deconvolution of individual bands in the UV-VIS region is a complex task, because the absorption bands are broad and often strongly overlap. Useful information about the molecular structure and environment can only be obtained by appropriate and truthful separation of these peaks.Öğe Finding Exact Number Of Peaks in Broadband UV-Vis Spectra Using Curve Fitting Method Based On Evolutionary Computing(2020) Avcu, Fatih Mehmet; Karakaplan, MustafaAbstract: High performance calculations are needed in order to resolve analytic signals of the day. However, it requires very long periods of time to perform these calculations with single processor systems. In order to reduce these calculation times, there is a need to turn to parallel programming algorithms that share more than one processor. Recently, solving complex problems with genetic algorithms has been widely used in computational sciences. In this work, we show a new method of curve fitting via genetic algorithm based on Gaussian functions, for deconvolution of the overlapping peaks and find the exact number of peaks in UV-VIS absorption spectroscopy. UV-VIS spectra are different than other instrumental analysis data. The resolution of UV-VIS spectra are complicated since the absorption bands are strongly overlapped. Useful information about molecular structure and environment can often be obtained by resolving these peaks properly. The algorithm was parallelized with the island model in which each processor computes a different population. This method has been used for resolving of the UV-VIS overlapping spectrum. The method particular algorithm is robust against bad resolution or noise. The results clearly show the effectiveness of the proposed method.Öğe Kompleks karışımların spektroskopik sinyallerinin paralel genetik algoritma ile analizi ve yorumlanması(İnönü Üniversitesi, 2012) Avcu, Fatih MehmetOptimizasyon diğer bilim dallarında olduğu gibi analitik kimya araştırmalarında da kullanılan önemli bir hesaplama işlemidir. Yapay zeka kullanımına başlanmadan önce araştırmacılar bir problemdeki parametreleri optimize etmek, aralarındaki ilişkiyi bulmak için deneme yanılma yolunu kullanırlardı. Problemdeki parametre sayısı arttığında çözümsüzlük veya elde edilen çözümden bir sonuç çıkarılamama durumları ortaya çıkmıştır. Yapay zeka tekniklerinden biri olan genetik algoritma son zamanlarda hesaplamalı bilimlerde sıklıkla kullanılır. Bir optimizasyon tekniği olan genetik algoritma, araştırma tekniklerinin alışılmamış bir türü olarak tanımlanmaktadır. Bu çalışmada amaç kompleks analitik sinyallerin değerlendirilmesinde genetik algoritmanın güçlü bir teknik olduğunu vurgulamak ve özellikle kimya alanında ne derece faydalı bir araç olduğunu göstermektir. Basitçe bir uzayda doğru sonucu arama işi olarak kullandığımız genetik algoritma, aynı zamanda paralel programlama aracılığı ile birden fazla noktada çalıştırarak (ada modeli) sonucun bulunma süreside kısaltılmıştır. Bu tezde farklı ada modelleri kullanıldı ve program en etkili olanı seçildi. Geliştirilen program literatürdeki test fonksiyonları ile denendi, paralel ve paralel olmayan test fonksiyonları karşılaştırılmıştır. Bu testlerden sonra geliştirilen yazılım önceki çalışmalarımızdan, İnönü Üniversitesi Merkezi Araştırma Laboratuvarından ve kimya veri tabanlarından elde edilen veriler kullanılmak suretiyle de test edildi. NMR, UV ve IR verilerinin çözümlemesi esnasında bulunan standart hata, ticari programlarla karşılaştırılmıştır. Geliştirilen yazılımın ticari yazılmla kısaylandığında daha iyi sonuçlar verdiği görülür. Sonuç olarak, sonuçlarımız deneysel veriye çok yakın olduğundan geliştirilen yazılımın analitik sinyaller için güvenli bir şekilde kullanılabileceği kanaatine varılmıştır.Öğe A parallel and non-parallel genetic algorithm for deconvolution of NMR spectra peaks(Elsevier, 2013) Karakaplan, Mustafa; Avcu, Fatih MehmetInterpreting high resolution nuclear magnetic resonance (NMR) spectra of complex samples is investigated by the use of real valued parallel and nonparallel genetic algorithm based on stochastic search procedure. The population-centric crossover operators were used in real coded genetic algorithm (RCGA) based on some probability distributions. This paper also presents parallel genetic algorithms computations with different genetic immigration operators. Different results were found with respect to the problems, even at the different stages of the genetic process in the same problem. It is observed that the grid and centralized type genetic immigration (island models) were effective on the global optimization. The parallel and non-parallel algorithms were also applied for solving some multi-modal test problems. It is found that the island models achieve superior performance on multi-modal test problems and on deconvolution of complex NMR spectra. (C) 2013 Elsevier B.V. All rights reserved.
