Banca de QUALIFICAÇÃO: THIAGO HENRIQUE FREIRE DE OLIVEIRA

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
DISCENTE : THIAGO HENRIQUE FREIRE DE OLIVEIRA
DATA : 04/07/2018
HORA: 09:00
LOCAL: Núcleo de Pesquisas em Inovação em Tecnolgia da Informação - NPITI
TÍTULO:

Reinforcement Learning Algorithms for Multiobjective Optimization Problems

PALAVRAS-CHAVES:

Multiobjective optimization, Q-Learning, Scalarization , Pareto Front.

PÁGINAS: 69
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Elétrica
RESUMO:

Multiobjective optimization problems portray real situations and so this class of problems
is extremely important. However, it lacks techniques that can overcome its limitations
that are imposed by the class and not by a specific problem. Reinforcement learning
algorithms and techniques are in line with this type of problem and some frameworks have
been proposed, they are based on one of the most popular algorithms used in this field,
Q-Learning. Thus, our proposal is to develop algorithms that can overcome the limitations
of the class of multiobjective optimization problems. They are used in benchmark problems
that simulate situations that occur in real problems and serve to validate techniques. For
the initial tests of the algorithm the benchmark Deep Sea Treasure was used, which is
extremely important and offers all the possible limitations that a multiobjective problem
can offer. Our algorithms can overcome some limitations, however, they can and should be refined
in order to fully solve these limitations, especially those that refer to the diversity of
solutions found under the Pareto Front and the possibility of choosing a posteriori optimal
policy

MEMBROS DA BANCA:
Presidente - 347628 - ADRIAO DUARTE DORIA NETO
Interno - 1837240 - MARCELO AUGUSTO COSTA FERNANDES
Externo ao Programa - 350241 - JORGE DANTAS DE MELO
Externo à Instituição - FRANCISCO CHAGAS DE LIMA JUNIOR - UERN