General Purpose Artificial Intelligence

Áreas

A General-Purpose Artificial Intelligence System (GPAIS) refers to an advanced AI system capable of effectively performing a range of distinct tasks. Its degree of autonomy and ability is determined by several key characteristics, including the capacity to adapt or perform well on new tasks that arise at a future time, the demonstration of competence in domains for which it was not intentionally and specifically trained, the ability to learn from limited data, and the proactive acknowledgment of its own limitations in order to enhance its performance.

Current DaSCI research lines:

AutoML aims to find the best AI model for a new problem, allowing automatic adaptation of AI to new tasks.
Evolutionary Deep Learning. Bio-inspired algorithms can be used to determine the best hyper-parameters and models of an AI model. Evolutionary DL is a research project to design DL models.
Few-shot learning refers to classifying new data having seen only a few training examples, based on how humans learn.

Computational intelligence approaches for GPAIS. Computational Intelligence (CI) is the theory, design, application and development of biologically and linguistically motivated computational paradigms. Traditionally the three main pillars of CI have been Neural Networks, Fuzzy Systems and Evolutionary Computation. The technologies based CI are useful to tackle the GPAIS open issues and challenges.