- author = "Alexander Lalejini and Charles Ofria",
- title = "Evolving Reactive Agents with {SignalGP}",
- booktitle = "ALIFE 2018: The 2018 Conference on Artificial Life",
- year = "2018",
- editor = "Takashi Ikegami and Nathaniel Virgo and Olaf Witkowski and Mizuki Oka and Reiji Suzuki and Hiroyuki Iizuka",
- pages = "5368--369",
- address = "Tokyo, Japan",
- month = "23-27 " # jul,
- publisher = "The MIT Press",
- keywords = "genetic algorithms, genetic programming",
-
URL = "
https://doi.org/10.1162/isal_a_00069",
-
DOI = "
doi:10.1162/isal_a_00069",
- size = "2 pages",
-
abstract = "We introduce SignalGP, a technique for creating
digital organisms that harnesses the event-driven
programming paradigm. These organisms can evolve to
automatically react to signals from the environment or
from other agents in a biologically-inspired manner. In
addition to introducing SignalGP, we summarize previous
results demonstrating the value of the event-driven
paradigm in environments dominated by agent-agent and
agent-environment interaction. Our full introduction to
SignalGP can be found in (Lalejini and Ofria,
2018).
Digital evolution has its roots in Genetic Programming (GP), wherein computer programs are evolved using natural principles. For example, Avida (Ofria et al., 2009) is a popular digital evolution system that uses self-replicating linear genetic programs as its organisms. The organisms generally follow an imperative programming paradigm where computation is driven procedurally. Program execution starts at the top of the program and proceeds in sequence, instruction-by-instruction, jumping or branching as dictated by executed instructions (McDermott and O'Reilly, 2015).
In contrast to imperative programming, program execution in event-driven computing is directed primarily by signals (events), simplifying the development of programs that dynamically react to events around them. This biologically-realistic mode of execution is often employed when developing software for agents that must frequently interact with each other or the environment (such as robotics or distributed systems). By capturing the event-driven paradigm, SignalGP aims to improve our capacity to evolve computer programs that operate in interaction-heavy environments, expanding our ability to generate complex agent-agent and agent-environment interactions.",
- notes = "Cites \cite{Lalejini:2018:GECCO} ALIFE 2018",
