Prof. Enrico Simetti

Prof. José María Maestre Torreblanca Talk: "Predictive control of water systems with agents in the loop" This talk will focus on the integration of heterogeneous agents in the control loop of model predictive controllers (MPC) in the context of water systems. The aim of the talk is to introduce different frameworks on how to effectively incorporate agents into MPC systems and to explore the challenges and opportunities that arise in this type of applications.

Dr. Sergio Toral Talk: ‘Intelligent Management of a Fleet of Unmanned Aquatic Vehicles for the Monitoring of Surface Water Bodies (AQUATRONIC)’ The AQUATRONIC project proposes the development of a flexible, robust, and reactive fleet of Autonomous Surface Vehicles (ASVs) to perform collaborative, real-time monitoring of surface water bodies. The ASVs will be equipped with sensors to measure the main physicochemical characteristics of the water (e.g. temperature, pH, conductivity, turbidity, redox potential, dissolved oxygen or nitrates) and withcommunication technology to transmit this data to the cloud (4G or 5G). Based on this and on model-based learning techniques, new algorithms will produce real-time maps of the variables of interest. The key characteristics of the proposed platform are real-timemonitoring, flexibility, robustness, reactiveness and collaboration. The goal is for the fleet to learn from the environment as it is deployed, determining the optimal laws of motion to detect and monitor with sufficient resolution the phenomena that affect the quality of aquatic ecosystems, thus enabling better management of water resources.

Dr. Thanh Phuong Nguyen Talk: ‘Towards an efficient embedded vision through neural network compression’ Nowadays, deep learning plays a key role in many areas of computer science and related fields such as computer vision, speech recognition, natural language processing, robotics, and so on. In general, deep models require a large amount of training data, have a significant memory footprint, and often involve high computational complexity during the inference phase. To democratize deep learning on embedded devices with limited computational resources, it is essential to design lightweight models that are effective in energy consommation and computational complexity while maintaining strong performance, making them suitable for deployment on edge devices. This talk presents our recent efforts in designing lightweight deep models or compressing pre-trained deep networks without significantly degrading their performance. We introduce efficient neural architectures specifically designed for deployment on embedded systems. Furthermore, we present various methods for effectively compressing pre-trained convolutional models, relying on neural network pruning techniques or tensor decomposition methods to represent convolutional layers with low-rank approximations.