Paper “Multi-Agent Actor-Critic Multitask Reinforcement Learning Based on Consensus,” by Milos S. Stankovic, Marko Beko, Nemanja Ilic and Srdjan S. Stankovic, has been accepted for presentation at the renowned IEEE Conference on Decision and Control (CDC) to be held in December 2022.

Paper “Convergent Distributed Actor-Critic Algorithm Based on Gradient Temporal Difference”, by M.S. Stanković, M. Beko and S.S. Stanković, has been accepted for presentation at 30th European Signal Processing Conference, EUSIPCO 2022, to be held in Belgrade, from August 29th, 2022 to September 2nd, 2022.

Paper “Distributed Actor-Critic Learning Using Emphatic Weightings”, by M.S. Stanković, M. Beko, and S.S. Stankovic, has been presented at the conference CoDIT 2022, Istanbul, Turkey, May 2022.

Paper “Deep Learning Based SWIR Object Detection in Long-Range Surveillance Systems: An Automated Cross-Spectral Approach“, by Pavlović, M.S.; Milanović, P.D.; Stanković, M.S.; Perić, D.B.; Popadić, I.V.; Perić, M.V., has been published in Sensors!

By using a multi-spectral imaging setting, the paper proposes a new cross-spectral automatic data annotation methodology for SWIR channel training dataset creation, in which the visible-light channel provides a source for detecting object types and bounding boxes which are then transformed to the SWIR channel. With the proposed cross-spectral methodology, the goal of the paper is to improve object detection in SWIR images captured in challenging outdoor scenes. Experimental tests using a state-of-the-art deep neural network-based YOLOX model demonstrate that retraining with the created SWIR image dataset significantly improves average detection precision.

The following four papers have been presented at the special session of Conference Sinteza 2022 dedicated to the project DECIDE:

1. A. Ćuk, J. Gavrilović, M. Tanasković, M. Stanković, “Mobile Robot Path Planning Optimization by Artificial Bee Colony,” in Sinteza 2022 – International Scientific Conference on Information Technology and Data Related Research, Belgrade, Singidunum University, Serbia, 2022, pp. 399-403.
2. M. Stanković, M. Beko, M. Pavlović, I. Popadić, S. Stanković, “Distributed On-Policy Actor-Critic Reinforcement Learning,” in Sinteza 2022 – International Scientific Conference on Information Technology and Data Related Research, Belgrade, Singidunum University, Serbia, 2022, pp. 389-393.
3. U. Dragović, M. Tanasković, M. Stanković, A. Ćuk, “Autonomous Drone Control for Visual Search Based on Deep Reinforcement Learning,” in Sinteza 2022 – International Scientific Conference on Information Technology and Data Related Research, Belgrade, Singidunum University, Serbia, 2022, pp. 382-388.
4. I. Walter, M. Tanasković, “Image Segmentation Processing for Thermographic Analysis,” in Sinteza 2022 – International Scientific Conference on Information Technology and Data Related Research, Belgrade, Singidunum University, Serbia, 2022, pp. 394-398.

Paper “Adaptive Consensus-based Distributed System for Multisensor Multitarget Tracking“, by S.S. Stanković, N. Ilić and M.S. Stanković, has been published in IEEE Transactions on Aerospace and Electronic Systems  (IEEE TAES)!

The paper proposes a new comprehensive system for distributed multisensor multitarget tracking, with all of its functions, including track initiation, confirmation, maintenance and termination, as well as track-to-track association and fusion, built around the concept of the probability of target existence and an adaptive consensus scheme. Stability of the proposed system is studied for the steady state and time-varying regimes. The system as a whole achieves high performance close to the centralized solution, outperforming all the comparable existing state-of-the-art approaches, keeping much lower communication and computation requirements.

Paper “Forensic analysis of commercial drones”, by J. Gavrilović, A. Ćuk, M. Tanasković and M.S. Stanković, has been presented at the conference INFOTEH 2022, March 2022.

Paper “Distributed Consensus-Based Multi-Agent Temporal-Difference Learning,” by Milos S. Stankovic, Marko Beko and Srdjan S. Stankovic , has been presented at the 60th IEEE Conference on Decision and Control (CDC), December 2021.

Paper “Application of Deep Learning Algorithms and Architectures in the New Generation of Mobile Networks“, by Dejan Dašić, Miljan Vučetić, Nemanja Ilić, Miloš Stanković and Marko Beko, has been published in Serbian Journal of Electrical Engineering.

The paper presents a detailed overview of state-of-the-art applications and services related to the new generation of mobile networks that employ deep learning methods. Modern architectures used for their deployment have also been discussed. The paper also presents a practical use case of modulation classification using deep learning in an application essential for modern spectrum management.

Paper “Distributed Value Function Approximation for Collaborative Multi-Agent Reinforcement Learning“, by M.S. Stanković, M. Beko and S.S. Stanković, has been published in IEEE Transactions on Control of Network Systems (IEEE TCNS)!

In the paper, several new distributed gradient-based temporal difference algorithms for decentralized multi-agent off-policy learning of the value function in Markov decision processes were proposed, rigorously theoretically analyzed and verified using extensive simulations.