Banca de QUALIFICAÇÃO: RICHARDSON SANTIAGO TELES DE MENEZES

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : RICHARDSON SANTIAGO TELES DE MENEZES
DATE: 10/11/2023
TIME: 09:00
LOCAL: Sala Virtual
TITLE:

Deep Q-Managed: A New Framework for Multi-Objective Deep Reinforcement Learning

KEY WORDS:

Multiobjective reinforcement learning, Deep Q-Learning, Double Q-Learning, Dueling Networks

PAGES: 42
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUMMARY:

The Deep Q-Managed algorithm, proposed in this work, represents a significant advancement in the field of multi-objective reinforcement learning. This novel strategy employs an updated technique for hybrid multi-objective optimization, which offers a mathematical guarantee that all policies belonging to the Pareto Front can be found, excelling in the acquisition of non-dominated multi-objective policies within environments characterized by deterministic transition functions. Its flexibility extends to scenarios where the Pareto Front exhibits convex, concave, or mixed geometric complexities, making it a versatile solution for a wide array of real-world applications. Our proposal is validated using the traditional MORL benchmarks and different configurations of the Pareto front. The quality of the policies found by our algorithm was compared with prominent approaches in the literature using the hypervolume metric. The outcomes of the proposed strategy establish the Deep Q-Managed algorithm as a worthy contender for tackling challenging, multi-objective problems.

COMMITTEE MEMBERS:
Presidente - 347628 - ADRIAO DUARTE DORIA NETO
Externo ao Programa - 2836616 - HELTON MAIA PEIXOTO - UFRNExterno à Instituição - JORGE DANTAS DE MELO - UFRN
Externo à Instituição - THIAGO HENRIQUE FREIRE DE OLIVEIRA - IFRN