I am a First year phd student. I like to work with algorithms and theoretical stuff, as well as Machine Learning and all its experimental applications. I consider Artificial Life as one of those things you will have to be familiar with in the future, so it is best to start early on.
Dedicated page to the project here.
03-Nov-2014 01:46:49PM-0700 Abstract Reworked. Possible Titles:
Abstract: In the Movable Feast Machine, direct interaction is limited to small regions of space, but many computations could benefit from larger scale structure. This paper presents a simple aggregation and localization strategy allowing individuals to form communicating groups and find their own absolute positions within the group. We demonstrate X and Y which are great and also explore Z which is bad but we mitigate that with W. Thank you.
This paper explores large scale communication in an artificial life setting. We propose and analyse a simple communication protocol that allows for individuals to be aware of regions beyond their own neighborhood. We claim that by using our protocol, the overhead in the elements is small and the individuals gain awareness of the universe with some confidence level. Moreover, We found out complexity and unexpected patterns can arise when communication channels outside certain specified neighborhood are provided, which demonstrates that non-local interactions have a large impact on the overall state of the simulation.
Abstract: In this paper we explore the impact of large scale communication in an artificial life setting, by engineering a set of procedures that takes interaction between agents to a whole new level. We propose and analyse a communication protocol which is simple, yet it allows for individuals to be aware of a portion of the universe that is much larger than their neighborhood. We claim that by using our protocol, the overhead in the elements is small and the individuals gain awareness of the universe with some confidence level. Moreover, We found out complexity and unexpected patterns can arise when communication channels outside certain specified neighborhood are provided, which demonstrates that non-local interactions have a large impact on the overall state of the simulation.
Tentative title: Engineering large scale communication in a virtual environment: From individual behavior to complex patterns.
Abstract: In this paper, we explore the impact of one of the most influential components for complexity to be achieved: communication. By engineering a set of procedures that allow for entities to communicate with each other, we found out complexity and unexpected patterns can arise. We carry out our experiments by using an alife simulator, which allows for a more controlled environment which still clearly maps out to different circumstances of the real world. Our experiments involve the analysis of a communication protocol which is simple enough to be carried out by each individual agent, yet correct enough that achieves large scale communication, while maintaining a low overhead on the design of the individuals. This experiments show that it is possible to get interesting unanticipated complex patterns from the overall system that do not depend entirely on the initial conditions and/or the isolated behavior of the individuals.
Conclusions: We have shown that it is possible to achieve a simple protocol that achieves global communication, by exploiting interaction patterns on the mfm platform. However, lots of testing still need to be done, when extremely complex patterns do not flourish because of some of limitations imposed by the protocol.
You can see both videos I took of my element, the first one only shows the simplest behavior I was trying to achieve, the second one, shows how a line of elements disperses itself to form a very interesting structure. I will continue to play with it and keep posting interesting videos or show demonstrations in class.
Here are the two videos