Information Theoretic Learning

Information Theoretic Learning (ITL) regards the use of criteria derived from Information Theory and that would allow to overcome the limitations of second-order statistics.

The ITL paradigm proposes the adoption of training criteria based on information measures such as entropy, mutual information and the new generalized correlation function, called correntropy, instead of traditional criteria based on second order statistics. Correntropy is a positive definite function which yields a generalized similarity measure between random variables (or between time samples of a stochastic process) and it involves higher-order statistics of input signals, therefore it can be a promising candidate for a diverse set of applications in machine learning and signal processing.

Optimization and Simulation

The goal is to propose metaheuristic methodologies and its integration with simulation techniques in order to tackle complex optimization problems which in turn support complex decision-making in the areas of smart cities, telecommunications systems and transportation and logistics,.

An optimization problem is always associated with a set of decision variables, i.e., a set of variables should be chosen such that the solution is feasible and optimal. An important step when approaching optimization problems is to identify the type of problem we are handling with, since algorithms for solving optimization problems are tailored to a particular type of problem. Although, metaheuristics constitute a powerful tool to tackle complex optimization problems in telecommunication area, some of these methodologies have been developed considering deterministic problems when, real-world communication scenarios are plenty of uncertainty.

Telecommunication Systems

In today’s complex world, there is a visible trend in the implementation of smart technologies to city planning and management, leading to greater optimization of time and resources, and resulting in more efficiency. All these form a series of very challenging problems.

The emergence of trends such of new Internet-based applications, video streaming, and content distribution have generated a great demand for the design and development of the network infrastructure. The continued growth of problems in term of size has led researchers to propose alternatives to traditional exact methods to solve complex problems in "real time". The main concern when thinking on telecommunication problems is to provide new fast and efficient methodologies or improve the methodologies already existing that allow decision makers and engineers to find approximate solutions to big problems in a short period of time. Thus, new heuristics need to be developed to solve associated problems in telecommunication systems. Especially, those that require an effective solution approach to handle a larger problem in practice in a considerable small amount of time.

I am really interested in all of these topics, if you fell like we can collaborate feel free to contact me.