Como pode o algoritmo ajudar na gestão de operações num futuro próximo, em áreas como a saúde, retalho online ou no sector energético? é a questão que norteia o próximo webinário no âmbito do Ciclo Research Webinars on Artificial Intelligence & Human Behaviour, com o orador Fábio Neves-Moreira, investigador no INESC TEC e professor auxiliar na Faculdade de Economia da Universidade do Porto.
For many years, operations research practitioners have been discussing about typical complaints heard when a solution of their algorithm is first shown to the planners – “The solution is not feasible!”, “I do not believe that this solution is better than mine!”, “What is your algorithm doing inside?” – they say. The truth is that it is very difficult to convince them about the quality and feasibility of the solutions, without actually going through the whole methodology. However, explaining very complex optimization processes may not be possible in every case. Therefore, we need to find better ways to quickly provide convincing explanations, so that humans can trust and understand every method, even if it is a black-box one. TRUST-AI aims at developing Explainable Artificial Intelligence (XAI) algorithms to tackle a variety of complex management challenges. The use of symbolic learning methods is to be particularly explored, as we envision the discovery of powerful and self-explainable predictive and prescriptive policies. To validate the proposed XAI approaches in a broad range of problems, three use cases will be addressed in healthcare, online retail, and energy-related fields.
Fábio Neves-Moreira
A sessão é aberta, com transmissão via zoom, mediante inscrição prévia.
Este ciclo de webinários é organizado pelo CSG – Research in Social Sciences and Management e REM – Research in Economics and Mathematics, os dois consórcios de investigação do ISEG.