Search

Your search keyword '"Hasan Sildir"' showing total 3 results
Start Over Author "Hasan Sildir" Language english
3 results on '"Hasan Sildir"'

1. Design of Feedforward Neural Networks in the Classification of Hyperspectral Imagery Using Superstructural Optimization

Author

Hasan Sildir, Erdal Aydin, and Taskin Kavzoglu

Subjects
artificial neural networks, classification, superstructure optimization, mixed-inter nonlinear programming, hyperspectral images, Science
Abstract

Artificial Neural Networks (ANNs) have been used in a wide range of applications for complex datasets with their flexible mathematical architecture. The flexibility is favored by the introduction of a higher number of connections and variables, in general. However, over-parameterization of the ANN equations and the existence of redundant input variables usually result in poor test performance. This paper proposes a superstructure-based mixed-integer nonlinear programming method for optimal structural design including neuron number selection, pruning, and input selection for multilayer perceptron (MLP) ANNs. In addition, this method uses statistical measures such as the parameter covariance matrix in order to increase the test performance while permitting reduced training performance. The suggested approach was implemented on two public hyperspectral datasets (with 10% and 50% sampling ratios), namely Indian Pines and Pavia University, for the classification problem. The test results revealed promising performances compared to the standard fully connected neural networks in terms of the estimated overall and individual class accuracies. With the application of the proposed superstructural optimization, fully connected networks were pruned by over 60% in terms of the total number of connections, resulting in an increase of 4% for the 10% sampling ratio and a 1% decrease for the 50% sampling ratio. Moreover, over 20% of the spectral bands in the Indian Pines data and 30% in the Pavia University data were found statistically insignificant, and they were thus removed from the MLP networks. As a result, the proposed method was found effective in optimizing the architectural design with high generalization capabilities, particularly for fewer numbers of samples. The analysis of the eliminated spectral bands revealed that the proposed algorithm mostly removed the bands adjacent to the pre-eliminated noisy bands and highly correlated bands carrying similar information.

Published
2020
Full Text
View/download PDF

2. Optimal Design and Operation of Integrated Microgrids under Intermittent Renewable Energy Sources Coupled with Green Hydrogen and Demand Scenarios

Author

Su Meyra Tatar, Handan Akulker, Hasan Sildir, Erdal Aydin, Aydın, Erdal, Tatar, Su Hümeyra, Akülker, H., Sildir, H., Koç University Tüpraş Energy Center (KUTEM) / Koç Üniversitesi Tüpraş Enerji Merkezi (KÜTEM), College of Engineering, Graduate School of Sciences and Engineering, and Department of Chemical and Biological Engineering

Subjects
Fuel Technology, Optimization and Control (math.OC), Renewable Energy, Sustainability and the Environment, Green hydrogen, Mixed-integer linear programming, Optimal renewable energy integration, Renewable energy management, Scenario generation, FOS: Mathematics, Energy Engineering and Power Technology, Condensed Matter Physics, Mathematics - Optimization and Control, Chemistry, Electrochemistry, Energy and fuels
Abstract

Renewable energy integration into existing or new energy hubs together with Green technologies such as Power to Gas and Green Hydrogen has become essential because of the aim of keeping the average global temperature rise within 2 °C with regard to the Paris Agreement. Hence, all energy markets are expected to face substantial transitions worldwide. On the other hand, investigation of renewable energy systems integrated with green chemical conversion, and in particular combination of green hydrogen and synthetic methanation, is still a scarce subject in the literature in terms of optimal and simultaneous design and operation for integrated energy grids under weather intermittency and demand uncertainty. In fact, the integration of such promising new technologies has been studied mainly in the operational phase, without considering design and management simultaneously. Thus, in this work, a multi-period mixed-integer linear programming (MILP) model is formulated to deal with the aforementioned challenges. Under current carbon dioxide limitations dictated by the Paris Agreement, this model computes the best configuration of the renewable and non-renewable-based generators, their optimal rated powers, capacities and scheduling sequences from a large candidate pool containing thirty-nine different equipment simultaneously. Moreover, the effect of the intermittent nature of renewable resources is analyzed comprehensively under three different scenarios for a specific location. Accordingly, a practical scenario generation method is proposed in this work. It is observed that photovoltaic, oil co-generator, reciprocating ICE, micro turbine, and bio-gasifier are the equipment that is commonly chosen under the three different scenarios. Results also show that concepts such as green hydrogen and power-to-gas are currently not preferable for the investigated location. On the other hand, analysis shows that if the emission limits are getting tightened, it is expected that constructing renewable resource-based grids will be economically more feasible., Scientific and Technological Research Council of Turkey (TÜBİTAK); 2322-International Fellowship for Outstanding Researchers Program

Published
2022

3. A mixed-integer linear programming based training and feature selection method for artificial neural networks using piece-wise linear approximations

Author

Hasan Sildir, Erdal Aydin, Aydın, Erdal, Şıldır, Hasan, Koç University Tüpraş Energy Center (KUTEM) / Koç Üniversitesi Tüpraş Enerji Merkezi (KÜTEM), College of Engineering, and Department of Chemical and Biological Engineering

Subjects
Mathematical optimization, Mixed integer nonlinear programming, Global optimization, Relaxation, Artificial neural network, Linear programming, Computer science, Applied Mathematics, General Chemical Engineering, Feature vector, Feature selection, General Chemistry, Industrial and Manufacturing Engineering, Convexity, Reduction (complexity), Machine learning, Artificial neural networks, Piece-wise linear artificial neural networks, Mixed-integer programming, Linearization, Integer programming
Abstract

Artificial Neural Networks (ANNs) may suffer from suboptimal training and test performance related issues not only because of the presence of high number of features with low statistical contributions but also due to their non-convex nature. This study develops piecewise-linear formulations for the efficient approximation of the non-convex activation and objective functions in artificial neural networks for optimal, global and simultaneous training and feature selection in regression problems. Such formulations include binary variables to account for the existence of the features and piecewise-linear approximations, which in turn, after one exact linearization step, calls for solving a mixed-integer linear programming problem with a global optimum guarantee because of convexity. Suggested formulation is implemented on two industrial case studies. Results show that efficient approximations are obtained through the usage of the method with only a few number of breakpoints. Significant feature space reduction is observed bringing about notable improvement in test accuracy., 2232 International Fellowship for Outstanding Researchers Program; Scientific and Technological Research Council of Turkey (TÜBİTAK)

Published
2022

Catalog

Books, media, physical & digital resources
See catalog results