This paper contrasts autoconstructive evolution, where evolving programs determine their own variations, with the PushGP genetic programming system. A problem size scaling analysis is performed on structural problems to study the dynamics of this hybrid approach.
Jul 7, 2012
This study examines the emergence of signaling using the Sir Philip Sidney Game, comparing infinite and finite population models. It finds that honest signaling equilibria are difficult to reach from non-communicative origins and highlights the limitations of finite models in capturing the dynamics of deceptive signaling.
Jul 3, 2012
This study examines the role of social tags in the emergence and maintenance of signaling, using the Sir Philip Sidney Game. The findings suggest that while inclusive fitness theory alone may not explain the evolution of cooperation, tag-based relatedness can achieve cooperativity in some cases.
May 20, 2012
This paper explores the use of emotional communication in a predator-prey game to evaluate the tradeoff between socio-emotional behavior at individual and species scales, suggesting that such dynamics may support the emergence of altruistic behavior.
Jul 31, 2011
We present a technique for evolving modular programs with genetic programming using tags to label and refer to code fragments. The tag-based module technique allows for incremental co-evolution of labels and code, scaling well with problem size and performing effectively in complex environments.
Jul 12, 2011
This paper explores the use of the Belousov-Zhabotinsky (BZ) reaction as a substrate for computation, demonstrating a technique to invert input signals and compute logical operators like NOT and NOR, with experimental progress in implementing these operators using soft lithography.
Aug 11, 2010
This paper proposes a paradigm for enhancing strategic real-time systems by introducing emotions into simulated agents, enabling better decision-making and cooperation that mimics human behavior through internal and external emotional control.
Mar 9, 2008
SuperDuperWalker is a framework for experiments on the evolution of locomotion, using a 3D physical simulation and a genetic algorithm to control agents. It supports an inquiry cycle for educational purposes.
Aug 3, 2005