Organizer: A. S. Demirkol, A. Ascoli, R. Tetzlaff Technische Universität Dresden, Germany
Due to the aggressive downscaling pursued in semiconductor technology over the past few decades, the size of the electronic components cannot be shrunk further. As a result, there is a high demand for the development of novel multi-functional devices, enabling to foster progress in integrated circuit design without increasing the number of transistors in the available area. This enables to develop alternative data processing hardware architectures, addressing the Von Neumann bottleneck of standard computers. Neuromorphic computing systems, being inspired from the structure, morphology, and functionality of the human brain, constitute a prominent alternative to conventional data processing machines, enabling a smart and efficient management of vision, auditory, tactile and olfactory sensory data, which is of great interest to the Internet-of-Things industry, nowadays. Recent advances in device technology, circuit design techniques, and computational modeling have gathered scientists from various fields, such as electronics engineering, computer science, computational neuroscience, material science, and device manufacturing, under the same research framework. The common goal of this multidisciplinary research community is to enable the development of electronic systems, serving artificial intelligence (AI) applications more efficiently than traditional technical products, as well as the conception of neuromorphic hardware, mimicking biological neural networks with a higher degree of accuracy than conventional purely-CMOS architectures. The aim of the special session is to enable knowledge sharing, exchange of ideas, and promote future cooperation for the advancement of neuromorphic engineering.
Organizers: A. Buscarino University of Catania, Italy, D. Comminiello, University of Roma, Italy, C. Famoso, University of Catania, Italy, Luca Patanè, University of Messina, Italy
Quaternion and hypercomplex neural networks have recently gained an increased interest from the research
community due to their properties that allow for properly modeling multidimensional and multimodal data while
reducing the number of parameters and consequently gaining faster inference. These models exploit hypercomplex
algebra properties, to design interactions among the imaginary units, thus sensibly reducing the number free
parameters with respect to real-valued models. Hypercomplex models have shown superior performance when
processing data with highly-related dimensions or modalities, in several practical applications ranging from
multichannel audio to multimodal medical images. Indeed, hypercomplex neural networks can learn local
relations in data and thus preserve the correlations among the dimensions of the input.
This Special Session
aims at proposing to the audience a circuits and systems point-of-view over hypercomplex neural networks,
introducing the basic concepts of their algebra and reporting the most recent results related either to the
optimization of the learning algorithms or to their emerging applications.
Organizer: İ. H. Giden, ASELSAN Academy, Türkiye
This special session aims to bring together government, academia and industry to present new fundamental basic research, innovative technologies and supports to solve critical security and defense challenges. This special session seeks to attract diverse participation and collaboration from academia, industry, defense and government agencies that will promote security interests with opportunities to increase technical depth and breadth as well as networking with peers.
Organizers: F. Y. Vural, Middle East Technical University, Türkiye, N. Arıca, Piri Reis University, Türkiye
Nowadays, AI methods, specifically deep neural networks, offer a powerful tool to Neuroscientists to
decipher the secrets of our brain during the cognitive tasks, such as, memory, language and perception. These
models are also used to develop technology for many application areas including, robotics, computer vision,
natural language processing and speech recognition. Furthermore, multimodal brain data can be modeled to
analyze and diagnose various neurological diseases, such as Alzheimer and epilepsy.
In this special
section, the intersection of Artificial Intelligence and Neuroscience will be considered to study and model
the human brain.
Organizers: O. Kızılbey, Scientific and Technological Research Council of Türkiye, Türkiye, T. Nesimoglu Middle East Technical University, Türkiye
Design of active and passive microwave circuits is the major challenge for all communication systems. This special session will unveil the mysterious aspects of microwave communications to today’s portable devices and wearable/implantable wireless sensors. Recently, there is a high demand for cellular communications for 5G & 6G and beyond.
The final deadline for paper submission has been extended to July 15.
Special sessions are announced.
Selected papers from ICECS 2023 will be published in IEEE CAS Transactions II
NEW!
The template is provided by http://graygrids.com
