Tiamat is a bot developed for the 2018 mRTS competition. Tiamat uses an evolutionary procedure to derive action abstractions. Tiamat casts the problem of inducing action abstractions as a problem of selecting a subset of pure strategies from a possibly large set of options. Tiamat uses Strategy Strategy Selection (SSS) to plan in real time in the space defined by the action abstraction thus generated.

Authors:

Julian Mariño - Universidade de São Paulo – [email protected].

Rubens de Oliveira Moraes Filho - Universidade Federal de Viçosa - [email protected].

Levi Lelis - Universidade Federal de Viçosa - [email protected].

Claudio Toledo - Universidade de São Paulo - [email protected]

How to cite it (in bibtex):

@inproceedings{MarinoMoraesToledoLelis, author = {Julian R. H. {Mari~no} and Rubens O. Moraes and Claudio F. M. Toledo and Levi H. S. Lelis}, title = {Evolving Action Abstractionsfor Real-Time Planning in Extensive-Form Games}, booktitle = {Proceedings of the {AAAI} Conference on Artificial Intelligence}, year = {2019} }

@inproceedings{lelis2017, author = {L. H. S. Lelis}, title = {Stratified Strategy Selection for Unit Control in Real-Time Strategy Games}, booktitle = {International Joint Conference on Artificial Intelligence}, year = {2017}, pages = {3735--3741} }