Deniz Gencaga

Deniz Gencaga
Antalya Bilim University · Department of Electrical and Electronics Engineering

Ph.D

About

24
Publications
4,486
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
184
Citations
Introduction
Deniz Gencaga currently works at the Department of Electrical and Electronics Engineering, Antalya International University, following his research at the Carnegie Mellon University. He is interested in the interdisciplinary applications of statistical signal processing and machine learning. Some of his research topics include "Formant manipulation in voice disguise by mimicry", "Statistical dependency and causal relationship analysis with applications on climate and aerosols" and "Bayesian modeling of signals using Particle Filters and variants of Kalman Filter."
Additional affiliations
September 2017 - present
Antalya Bilim University
Position
  • Faculty Member
October 2016 - August 2017
Carnegie Mellon University
Position
  • Researcher
October 2014 - October 2016
Carnegie Mellon University
Position
  • PostDoc Position
Education
September 2000 - May 2007
Bogazici University
Field of study
  • Electrical and Electronics Engineering-Signal Processing

Publications

Publications (24)
Article
Full-text available
Neuronal noise is a major factor affecting the communication between coupled neurons. In this work, we propose a statistical toolset to infer the coupling between two neurons under noise. We estimate these statistical dependencies from data which are generated by a coupled Hodgkin–Huxley (HH) model with additive noise. To infer the coupling using o...
Article
Full-text available
In this work, we propose an approach to better understand the effects of neuronal noise on neural communication systems. Here, we extend the fundamental Hodgkin-Huxley (HH) model by adding synaptic couplings to represent the statistical dependencies among different neurons under the effect of additional noise. We estimate directional information-th...
Preprint
Full-text available
In this work, we propose an approach to better understand the effects of neuronal noise on neural communication systems. Here, we extend the fundamental Hodgkin-Huxley (HH) model by adding synaptic couplings to represent the statistical dependencies among different neurons under the effect of additional noise. We estimate directional information-th...
Article
Full-text available
Statistical relationships among the variables of a complex system reveal a lot about its physical behavior[...]
Article
Full-text available
nformation-theoretic quantities, such as entropy and mutual information (MI), can be used to quantify the amount of information needed to describe a dataset or the information shared between two datasets. In the case of a dynamical system, the behavior of the relevant variables can be tightly coupled, such that information about one variable at a g...
Conference Paper
Full-text available
In this survey, we present and compare different approaches to estimate Mutual Information (MI) from data to analyse general dependencies between variables of interest in a system. We demonstrate the performance difference of MI versus correlation analysis, which is only optimal in case of linear dependencies. First, we use a piece-wise constant B...
Conference Paper
Full-text available
Air quality information is increasingly becoming a public health concern, since some of the aerosol particles pose harmful effects to peoples health. One widely available metric of aerosol abundance is the aerosol optical depth (AOD). The AOD is the integrated light extinction coefficient over a vertical atmospheric column of unit cross section, wh...
Conference Paper
Full-text available
Measurements made by satellite remote sensing, Moderate Resolution Imaging Spectroradiometer (MODIS), and globally distributed Aerosol Robotic Network (AERONET) are compared. Comparison of the two datasets measurements for aerosol optical depth values show that there are biases between the two data products. In this paper, we present a general f...
Article
Full-text available
We present a novel and general methodology for modeling time-varying vector autoregressive processes which are widely used in many areas such as modeling of chemical processes, mobile communication channels and biomedical signals. In the literature, most work utilize multivariate Gaussian models for the mentioned applications, mainly due to the lac...
Article
In this work, we present a novel method for modeling time-varying autoregressive impulsive signals driven by symmetric alpha stable distributions. The proposed method can be interpreted as a two-stage Gibbs sampler composed of a particle filter, which is capable of estimating the unknown time-varying autoregressive coefficients, and a hybrid Monte...
Conference Paper
Full-text available
Understanding the origins of life has been one of the greatest dreams throughout history. It is now known that star‐forming regions contain complex organic molecules, known as Polycyclic Aromatic Hydrocarbons (PAHs), each of which has particular infrared spectral characteristics. By understanding which PAH species are found in specific star‐forming...
Conference Paper
Full-text available
Information-theoretic quantities, such as entropy, are used to quantify the amount of information a given variable provides. Entropies can be used together to compute the mutual information, which quantifies the amount of information two variables share. However, accurately estimating these quantities from data is extremely challenging. We have dev...
Article
In the literature, impulsive signals are mostly modeled by symmetric alpha-stable processes. To represent their temporal dependencies, usually autoregressive models with time-invariant coefficients are utilized. We propose a general sequential Bayesian modeling methodology where both unknown autoregressive coefficients and distribution parameters c...
Conference Paper
Full-text available
In the last decade alpha-stable distributions have become a standard model for impulsive data. Especially the linear symmetric alpha-stable processes have found applications in various fields. When the process parameters are time-invariant, various techniques are available for estimation. However, time-invariance is an important restriction given t...
Conference Paper
Full-text available
In adaptive signal processing, joint process estimation plays an important role in various estimation problems. It is well known that a joint process estimator consists of two struc-tures, namely the orthogonalizer and the regression filter. In literature, orthogonalization step is performed either by or-thogonal transformations or by linear predic...
Conference Paper
Full-text available
In this work, we propose a novel approach to perform Dependent Component Analysis (DCA). DCA can be thought as the separation of latent, dependent sources from their observed mixtures which is a more realistic model than Independent Component Analysis (ICA) where the sources are assumed to be independent. In general, the sources can be spatio‐tempo...
Conference Paper
Full-text available
In this work, a new method is proposed in order to sequentially estimate the time-varying parameters of a Cauchy distributed process. For this purpose, particle filters, which are used in non-Gaussian and nonlinear Bayesian applications, are utilised. The proposed method forms a basis for the possible future applications of the α-stable distributio...
Article
In this work, a new method is proposed in order to sequentially estimate the time-varying parameters of a Cauchy distributed process. For this purpose, particle filters, which are used in non-Gaussian and nonlinear Bayesian applications, are utilised. The proposed method forms a basis for the possible future applications of the �-stable distributio...

Network

Cited By

Projects

Projects (2)
Project
Neuroscience is based on better understanding the interactions between neurons. In literature, models, such as those proposed by (Hudgkin and Huxley, 1953) and by (Izhikevich, 2003) are two commonly preferred approaches. However, the interactions between the neurons are modeled via differential equations without taking the effect of neuronal noise into account. The objective of this project is to develop a much more generalized model where the interactions of the neuronal noise can also be integrated into the above equations. This modeling is of utmost importance as these signals can change the values of the action potential values above or belove the threshold levels. Here, we will utilize information-theoretic quantities to estimate these relationships from data with their statistical significance. These will include the estimation of quantities such as the Mutual Information(MI) and the Transfer Entropy(TE). In literature, correlation based techniques have been widely preferred to model neuronal interactions. However, correlation based methods have been found to be poorly performing to estimate nonlinear interactions. As neuron interactions are highly nonlinear, MI and TE have been utilized for nonlinear cases (Kantz, H and Schreiber, T., 2003). Mutual information is a measure of information shared between two variables and it is symmetric. In order to model the direction of the interactions between neurons, this is not enough. Therefore, an asymmetrical quantity, known as the Transfer Entropy is preferred. TE can be explained by the information gain obtained by using the past values of another process compared to the future value of the current process, by exluding the effects of its own past. Therefore, it provides direction of the information flow between two Markov processes. In this project, the cause and effect relationships between the neurons and the neuronal noise will be explored from spike trains. Hodgkin-Huxley and Izhikevich models will be generalized to include these noise interactions. Through this approach, we will learn new information on the working mechanism of the highly complicated neuronal system. Additionally, we will develop a new Bayesian piecewise-constant model to estimate the MI and TE by providing a posteriori probabilistic distributions. By this approach, we will provide a much more reliable estimation technique compared to the widely utilized methods, such as the histogram, Kernel Density Estimation and Adaptive Partitioning. In this project, we will propose novel approaches to model highly complex interactions between the neurons and the neuronal noise. Using these models, we will contribute to the scientific community in the area of modeling neuronal interactions with reliable data analysis tools. The proposed techniques will be applied to real neuronal data measured from C. Elegans earthworm, and the cause and effect of neuronal interactions will be tested by experiments and simulations.
Project
Comprehensive theoretical understanding of DNNs using information theory