International Symposium on
Electronics and Telecommunications ETC 2010
Ninth Edition


Timisoara, November 11-12, 2010
link
space
 
  spaceInvited speakers
 
space
link Home
link Call for Papers
link Paper Submission
link Paper Presentation
link Technical Program Committee
link Organizing Committee
link Invited Speakers
link Paper Form
link Previous Conference
link Registration
link Venue
link Accomodation
link Tourist information
link Publication
link Sponsors
link Useful links
link Contact


space

Claude Berrou, Télécom Bretagne, Brest, France

Sparse neural networks with large learning diversity

This talk deals with a new family of neural networks devised from some concepts of graph theory, error correction coding and sparsity. Precisely, coded recurrent neural networks with three levels of sparsity are introduced. The first level is related to the size of messages, much smaller than the number of available neurons. The second one is provided by a particular coding rule, acting as a local constraint in the neural activity. The third one is a characteristic of the low final connection density of the network after the learning phase. Though the proposed network is very simple since it is based on binary neurons and binary connections, it is able to learn a large number of messages and recall them, even in presence of strong erasures.


Ioannis Pitas, Aristotle University of Thessaloniki and CERTH/ITI, Greece

Human centered Video Analysis For Multimedia Postproduction
N. Nikolaidis, I. Pitas

During the last two decades, we have witnessed an increasing research interest towards what one could call *anthropocentric (human centered) video analysis*, namely, algorithms that aim to extract, describe, and organize information regarding the basic element of most videos: humans. This diverse group of algorithms processes videos from various sources and extracts a wealth of useful information related to the state (presence, identity, body posture, emotional state, etc.) and state transitions (body parts movements, activities, etc.) of individuals, interactions or communication modes between two or more humans (dialogues, social signals, etc.) and physical characteristics of humans, such as 3D body models. The interest of the scientific community for anthropocentric video analysis stems from the fact that the extracted information can be utilised in various important applications. One such application domain is film and games postproduction where anthropocentric video analysis results can be used in tasks such as indexing and retrieval or automatic semantic annotation. In this presentation we will review recent research results in a number of diverse areas such as face detection and tracking, eye detection, visual speech recognition, 3D face reconstruction, face clustering, face recognition, facial expression analysis, and activity recognition, either from a single or multiple visual (image/video) sources.
space
UNDER THE AEGIS OF:
space
link
IEEE Romania Section

IEEE ComSoc Germany Chapter
space
link
Romanian Academy of Technical Sciences
space
ORGANIZERS:
space
link
"Politehnica" University of Timisoara
space
link
Faculty of Electronics and Telecommunications
space
 
space