Doctoral Candidates

Mikhail Ashkerov

Name: Mikhail Ashkerov (scholarship holder)

Email:  Mikhail.Ashkerov(at)b-tu.de

Supervisor:

 Prof. Dr. Douglas W. Cunningham

Title of the dissertation:

Motion detection in image sequences and 3d reconstruction from motion (Working title)

Description:

The work under this project is divided into two major phases. The first concerns extraction of information about moving objects from a set of consecutive images. To date, this problem is usually solved by optical flow methods. These methods have many variations and improvements, but all of them are not without flaws. Our goal is to determine the movement using the methods studied in psychology for human perception. The second phase of our project is the reconstruction of three-dimensional information from the movement. Today, there are many ways to obtain three-dimensional information from the image. Our task is to find out what additional information can be obtained if we extract data from the sequence of images by methods wich we discussed in the first phase.

Short bio:

I graduated from the Reshetnev Aerocosmic University in 2009 as an engineer in aircraft control systems. During the study, I passed practice at the university and participated in some research projects. For two years I worked as an engineer for setting up and repairing electronics. Two years worked as a programmer in the company that produces drones. In 2014, I came in the BTU for writing a Ph.D.

Philipp Hahn

Name: Philipp Hahn (scholarship holder)

Email:  hahnphil(at)b-tu.de

Supervisors:

 Prof. Dr. Douglas W. Cunningham

 Prof. Dr. rer. biol. hum. Erich Schneider

Title of the dissertation:

Social Virtual Agents and the Effects of the Degree of Visual Realism on Socio-Emotional Communication (Working title)

Description:

Nowadays humans try to generate virtual agents - i.e. characters provided with artificial intelligence, looking and acting like human beings - which are able to process and generate socio-emotional information.

Against this background, this research project focuses on methods to design and analyze virtual agents. Of special interest is if and how the appearances of the characters affect the socio-emotional information perceived by unbiased users when interacting with those agents. Thereby, the effect of using different non-photorealistic renderings and virtual eye-movements will be analyzed.

For that reason, an underlying platform for controlling the virtual agents and many of its features will be implemented, including a customizable input- and output-interface. Moreover, an environment for controlled experiments will be designed, allowing participants to interact with the virtual agents and validate their experiences during that process.

Short bio:

Philipp Hahn is a PhD candidate at the Department of Graphic Systems at the Brandenburg University of Technology Cottbus-Senftenberg, under the advice of Prof. Dr. Douglas W. Cunningham and Prof. Dr. Erich Schneider. Previously, he obtained his B.Sc. and M.Sc. in Computer Science at the same university.

His research interests span rendering and the perception of humans and how these fields affect one another, with a focus on human-computer interaction and virtual agents. He aims to find the effects of realistic and non-realistic rendering on the socio-emotional information a human observer or operator perceives. Towards this goal, he already conducted some research on static imagery in his thesis “A Blender based Affective Talking Head with variable degree of visual realism” that he wants to expand and delve into during his PhD.

Toni Schneidereit

Name: Toni Schneidereit (scholarship holder)

Email: Toni.Schneidereit(at)b-tu.de

Supervisor:

Prof. Dr. rer. nat. habil. Michael Breuß

Title of the Dissertation:

PDE-basierte Bildverarbeitungsmodelle für quantenmechanische Informationsgewinnung (Working title)

Description:

Die Verfolgung von Objekten in einer Bildsequenz ist als Korrespondenzproblem bekannt, während
die Bewegung der Objekte selbst als Vektorfeld, dem Optischen Fluss, gesucht wird. Zur Berechnung
existieren bereits zahlreiche Methoden (z.B. Horn & Schunck), welche mit Hilfe von PDE-basierten
numerischen Deep-Learning Ansätzen unter anderem Randbedingungen lernen sollen. Diese Ansätze
sollen später auf Bildverarteitungsmodelle zur quantenmechanischen Informationsgewinnung
übertragen und angewendet werden.

Short bio:

Mit dem Master of Science beendete Toni Schneidereit Anfang 2018 sein Physik-Studium an der BTU
Cottbus-Senftenberg im Bereich der numerischen Hydrodynamik. Seit Februar desselben Jahres
arbeitet er als Stipendiat der Graduate Research School am Lehrstuhl für Angewandte Mathematik
unter Prof. Dr. rer. nat. habil. Michael Breuß.

Martin Schorradt

Name: Martin Schorradt (scholarship holder)

Email: schormar(at)b-tu.de

Supervisor:

Prof. Dr. habil. Douglas W. Cunningham

Title of the Dissertation:

The Semantic and Spatio-Temporal Structure of Audible Speech (Working title)

Description:

Several years ago, personal assistance systems have entered our lives. They are designed to make our daily lives easier by supporting us in the case of burdensome tasks or by completely taking them over. Due to the fact, that people tend to assign personality to those machines, they always use emotions while talking to them and assume the back channel will do as well. Although communication is multimodal - as emotions are conveyed through mimic, gestures, words and voice - voice is the most widely used channel for this kind of human-machine communication. Hence, if the machine does not react emotionally correct, it may lead to miscommunication and confusion on the human side. In order to make communication as natural as possible, such systems MUST be able to correctly recognize and synthesize emotions.
It is assumed that information about the emotional condition of a person that is conveyed through the acoustic portion of the communication is mainly in his speech melody, the prosody. The aim of this project is to examine the influence of different prosodic features on the perception of affective speech. For that, it is first necessary to determine which features are crucial, and secondly, it is required to qualify how those features can be used for accurate humanlike speech synthesis. Therefore, the correlation of physical features and the semantic space of affective speech will be investigated in more detail.

Short bio:

I obtained my M.Sc. in computer science at the Brandenburg University of Technology Cottbus-Senftenberg in 2018. During my studies my interest in human-machine communication, especially in the acoustic channel, was raised. Therefore, I decided to gain a better insight into this part of communication starting with my bachelor thesis about affective speech synthesis by an articulatory synthesis system.  During May of 2018, I started working for my Ph.D. at the Chair of the Graphic Systems Department under supervision of Prof. Dr. Douglas W. Cunningham.

Elham Bahrami Foroutan

Name: Elham Bahrami Foroutan (scholarship holder)

E-Mail: bahraelh(at)b-tu.de

Supervisor:

Prof. Dr.-Ing. Ingo Schmitt

Title of the dissertation:

Quantum-Based Modelling and Querying

Description:

Sometimes an agent or agents confront conditions in which the best result cannot be reached with considering only one type of the world. We are interested in applying hybrid or merged worlds to deal with such problems. Previous research works are based on non-deterministic and deterministic environment separately. However, none of them considers such positions which require combining both approaches. Moreover, possible approaches to reduce the complexity of quantum based algorithms should be investigated.

We will generally try to use a suitable semantic and quantum based method to predict entire possible results and also to reduce its algorithm complexity.

  • Statement of the problem: Using a non-deterministic environment with a fixed set of actions and composing them to form a suitable model for cognitive agents.
  • Objectives: We want to create an efficient and effective plausible world to be able to predict activity results using previous observations based on composed fixed and non-deterministic environments and also decrease the complexity of quantum based algorithm.
  • Hypothesis: Cognitive agent A which is coupled to its environment would be considered where the world U may be non-deterministic or fixed or a composition of both of them.
  • Methods: There are 3 approaches
  1. Quantum based method
  2. Modelling and Querying
  3. Changing environment
  • Conclusions :
  1. A quite accurate and powerful model will be developed because of our proposed semantic method and probabilistic databases.
  2. Prediction of  the activity results will be possible based on previous observations using composed environments.
  3. Complexity of algorithms will be reduced using quantum based approaches.

Short bio:

I hold a master degree in the field of computer engineering. The field of computer engineering (software) is attractive for me concerning its essential role in analyzing and solving problems of different scientific specialties. I have some teaching experiences at universities and technical-vocational organizations and I also succeeded to publish some papers in outstanding journals. My reseach interests are generally based on the following area

  • Data Mining
  • Clustering
  • Ontology
  • Web Mining

The topic of my Ph.D. scholarship program is Quantum-Based Modelling and Querying”. The idea about making attempt to get involved in this topic arises from my educational and technical experiences which help me to be efficient in such areas. In my opinion, the main purpose for a researcher shouldn't be the study itself, but rather discovering ways and applying scientific research works to new requirements of the modern world. The above-stated topic has such features. This research will be useful for industries dealing with huge volume of data in different conditions to make commercial/technical decisions for future positions.

Sanjit Kumar Saha

Name: Sanjit Kumar Saha (scholarship holder)

Email: sahasanj(at)b-tu.de

Supervisor:

Prof. Dr.-Ing. habil. Ingo Schmitt

Title of the dissertation:

Quantum-Based Modelling and Querying

Description:

Nowadays, the more relevant and interesting question is how difficult it is to cluster data sets that can be clustered well. Clustering is the process that groups the sets of objects into mutually exclusive subsets (clusters) in terms of their attributes, so that objects in one cluster are similar to each other and dissimilar to members of other clusters. And the clustering technique takes dissimilarity measures of objects as inputs. Often, clustering algorithms are applied to dissimilarities that violate the usual triangle inequality and therefore, it fail to be metrics and compromises the quality of the resulting clustering.

We propose a theoretically and empirically motivated model based on mathematical formulation of Quantum Mechanics and probability theory that can calculate the similarity measure of objects and perform clustering in an efficient way that the resulting clusters meet the quality and always satisfies the triangle inequality.

Short bio:

Sanjit Kumar Saha has obtained his both Bachelor of Science and Master of Science degree in computer science and engineering from Jahangirnagar University, Bangladesh. He is an assistant professor of department of computer science and engineering at Jahangirnagar University, Bangladesh. He has 8+ years of teaching experience to both undergraduate and graduate students. Saha has published journal and conference papers in the International and National journal and conferences. His current research interest includes quantum computing, deep learning, artificial neural network, fuzzy logic and systems. From February 2019, he has started his PhD at the chair of the Database and Information Systems of Prof. Dr.-Ing. habil. Ingo Schmitt.

This website uses cookies. There are two types of cookies: The first type supports the basic functionality of our website. The second allows us to improve our content for you by saving and analyzing pseudonymised user data. Since this second type is technically not required to run the website, you can withdraw your consent to those cookies at any time. For more information please visit our pages on data protection.

Mandatory

These cookies are needed for a smooth operation of our website.

Statistic

For statistical reasons, we use the platform Matomo to analyse the user flow with the help of website users‘ pseudonymised data. This allows us to optimize website content.

Name Purpose Lifetime Type Provider
_pk_id Used to store a few details about the user such as the unique visitor ID. 13 months HTML Matomo
_pk_ref Used to store the attribution information, the referrer initially used to visit the website. 6 months HTML Matomo
_pk_ses Short lived cookie used to temporarily store data for the visit. 30 minutes HTML Matomo