Arif Selcuk Ogrenci

Channel estimation for visible light communications using neural networks

Visible light communications (VLC) is an emerging field in technology and research. Estimating the channel taps is a major requirement for designing reliable communication systems. Due to the nonlinear characteristics of the VLC channel those parameters cannot be derived easily. They can be calculated by means of software simulation. In this work, a novel methodology is proposed for the prediction of channel parameters using neural networks. Measurements conducted in a controlled experimental setup are used to train neural networks for channel tap prediction. Our experiment results indicate that neural networks can be effectively trained to predict channel taps under different environmental conditions.

Epidemic models for phase transitions: application to a physical gel

ABSTRACT Carrageenan gels are characterized by reversible sol–gel and gel–sol transitions under cooling and heating processes and these transitions are approximated by generalized logistic growth curves. We express the transitions of carrageenan-water system, as a representative of reversible physical gels, in terms of a modified Susceptible-Infected-Susceptible epidemic model, as opposed to the Susceptible-Infected-Removed model used to represent the (irreversible) chemical gel formation in the previous work. We locate the gel point Tc of sol–gel and gel–sol transitions and we find that, for the sol–gel transition (cooling), Tc > Tsg (transition temperature), i.e. Tc is earlier in time for all carrageenan contents and moves forward in time and gets closer to Tsg as the carrageenan content increases. For the gel–sol transition (heating), Tc is relatively closer to Tgs; it is greater than Tgs, i.e. later in time for low carrageenan contents and moves backward as carrageenan content increases.

Empirical results about efforts for effective teaching to y-generation freshman students

New techniques are deployed to teach the new generation of students effectively. This work tries to share our experience in a blended course for over eleven years. It has been observed that the online portion of the course has to be adjusted carefully in order to obtain a high level of student satisfaction and overall throughput from the course.

Mathematical Characterization of Thermo-reversible Phase Transitions of Agarose Gels

ABSTRACT The thermal phase transition temperatures of high (HMP) and low melting point (LMP) agarose gels were investigated by using UV–vis spectroscopy techniques. Transmitted light intensities from the gel samples with different agarose concentrations were monitored during the heating (gel-sol) and cooling (sol–gel) processes. It was observed that the transition temperatures, Tm, defined as the location of the maximum of the first derivative of the sigmoidal transition paths obtained from the UV–vis technique, slightly increased by increasing the agarose concentration in both the HMP and LMP samples. Here, we express the phase transitions of the agar-water system, as a representative of reversible physical gels, in terms of a modified Susceptible-Infected-Susceptible epidemic model whose solutions are the well-known 5-point sigmoidal curves. The gel point is hard to determine experimentally and various computational techniques are used for its characterization. Based on previous work, we locate the gel point, T0, of sol-gel and gel-sol transitions in terms of the horizontal shift in the sigmoidal transition curve. For the gel-sol transition (heating), T0 is greater than Tm, i.e. later in time, and the difference between T0 and Tm is reduced as the agarose content increases. For the sol-gel transition (cooling), T0 is again greater than Tm, but it is earlier in time for all agarose contents and moves forward in time and gets closer to Tm as the agarose content increases.

Transmitter source location estimation using crowd data

Abstract The problem of transmitter source localization in a dense urban area has been investigated where a supervised learning approach utilizing neural networks has been adopted. The cellular phone network cells and signals have been used as the test bed where data are collected by means of a smart phone. Location and signal strength data are obtained by random navigation and this information is used to develop a learning system for cells with known base station location. The model is applied to data collected in other cells to predict their base station locations. Results are consistent and indicating a potential for effective use of this methodology. The performance increases by increasing the training set size. Several shortcomings and future research topics are discussed.

Weight exchange in distributed learning.

Neural networks may allow different organisations to extract knowledge from the data they collect about a similar problem domain. Moreover learning algorithms usually benefit from being able to use more training instances. But the parties owning the data are not always keen on sharing it. We propose a way to implement distributed learning to improve the performance of neural networks without sharing the actual data among different organisations. This paper deals with the alternative methods of determining the weight exchange mechanisms among nodes. The key is to implement the epochs of learning separately at each node, and then to select the best weight set among the different neural networks and to publish them to each node. The results show that an increase in performance can be achieved by deploying simple methods for weight exchange.

A Systems Software Architecture for Training Neural, Fuzzy Neural and Genetic Computational Intelligent Networks

A systems software architecture for training distributed neural, fuzzy neural and genetic networks and their relevant information models have been developed. Principles of on-line architecture building, training, managing, and optimization guidelines are provided and extensively discussed. Qualitative comparisons of neural training strategies have been provided.