Science and Technology Production
Congress
Authorship
RUBIO SCOLA, IGNACIO EDUARDO JESUS
;
Luis Rodolfo Garcia Carrillo
;
Joao Hespanha
Date
2020
Publishing House and Editing Place
IEEE
ISSN
978-1-5386-8266-1
Summary
Information provided by the agent in
SIGEVA
In recent years diverse computational models ofemotional learning observed in the mammalian brain haveinspired a number of self-learning control approaches. Thesearchitectures are promising in terms of their learning abilityand low computational cost. However, the lack of rigorousstability analysis and mathematical proofs of stability andperformance has limited the proliferation of these controllers.To address this drawback, this paper proposes a modifiedbrain emotional neural network structure...
In recent years diverse computational models ofemotional learning observed in the mammalian brain haveinspired a number of self-learning control approaches. Thesearchitectures are promising in terms of their learning abilityand low computational cost. However, the lack of rigorousstability analysis and mathematical proofs of stability andperformance has limited the proliferation of these controllers.To address this drawback, this paper proposes a modifiedbrain emotional neural network structure using a radial basisfunction inside the Thalamus and an emotional signal based onan integral action structure to increase performance. Mathe-matical stability proofs are provided, together with numericalsimulations, demonstrating the superior performance obtainedwith the new modifications proposed to the emoional learning-inspired control.
Show more
Show less
Key Words
Robust controlLyapunov stabilityEmotional learning