Artificial Intelligence is impacting industry and the business world in deep ways. We are also working on some industrial and business applications together with partner companies.
We have recently started a collaboration with INFINEON on the use of reinforcement learning for automatic layout and routing design of analogic circuits of microprocessors.
We are also collaborating with Idrostudi to analyse data and build predictive models of water networks. We have designed an anomaly detection scheme capable of identifying sensor and physical anomalies, and we are now planning to extend these approaches further ideally to build a digital twin of the water network system.
We are also working with General Invest to design more effective risk aware models to improve tactical asset allocation of investments portfolios.
Keywords: Automatic circuit layout and routing design, anomaly detection on water networks, risk-aware tactical asset allocation.
In the following, there is a reasoned bibliography, in which each class of contributions is briefly described and the relevant bibliography is cited.
Automatic circuit design
Automated analog circuit floorplanning and routing is a highly specialized area of electronics design automation (EDA). The goal of floorplanning is to determine the optimal placement of components on a chip to minimize area and improve performance, while routing involves creating the electrical connections between these components. Traditionally, these tasks have been quite challenging due to the combinatorial nature of the problem and sensitivity of analog circuits to parasitic effects. Only in recent years, application of Reinforcement Learning and Neural Networks has led to disruptive advancements in the field.
Anomaly detection of water networks
The growth of urban centers, both in terms of extension and population size, has brought increased stress on water distribution and sewage systems. These systems are essential for local communities but unfortunately rely on ageing networks which often go under stress during extreme weather events. The project, developed in collaboration with Idrostudi srl, has a multiple objectives in mind which leverage AI and ML methods on time series generated by both networks. A few examples are the inpainting of missing data with generative models, anomaly detection and prediction of extreme events in the network as a consequence of meteorological or external events.