Paper “Evolving Structures in Complex Systems”, authored by Hugo Cisneros, Josef Sivic and Tomas Mikolov received the Best Student Paper Award at the IEEE Symposium on Artificial Life as a part of the IEEE Symposium Series on Computational Intelligence (SSCI) 2019 that took place in Xiamen, China from December 6th to December 9th 2019.
The current success stories of artificial intelligence (AI), such as automatic translation or autonomous driving, rely on developing specialized algorithms trained from large amounts of carefully annotated data. This paper makes a step towards systems that are adaptable and capable of developing and evolving on their own, which is a major challenge that might be crucial for eventually developing AI with general learning capabilities.
The main idea of the paper is to measure the rate of complexity growth in cellular automata by measuring how many new patterns emerge during the evolution of the system. For example, a system that produces only chaotic, unpredictable noise does not score well on this metric, as there is no growth of identifiable patterns. On the other hand, systems that achieve the highest values of the emerging complexity growth exhibit a wide range of interesting behaviors. Some of them are similar to the famous Game of life, but others exhibit complex, interesting behaviors that has not been studied before – and look like some sort of artificial, mathematical form of life.
Hugo Cisneros is a PhD student in the IMPACT project, studying at ENS Paris, and splitting his time between CIIRC in Prague and Inria Paris. Josef Sivic is a distinguished researcher at CIIRC CTU and senior researcher at Inria in Paris. Tomas Mikolov is currently a research scientist at Facebook AI and will be joining CIIRC CTU next year.
IEEE SSCI is an annual international conference on computational intelligence sponsored by the IEEE Computational Intelligence Society. It features a range of symposia covering a variety of topics related to computational intelligence.
IEEE Symposium on Artificial Life brings together researchers working on the emerging areas of Artificial Life and Complex Adaptive Systems, aiming to understand and synthesize life-likesystems and applying bio-inspired synthetic methods to other science/engineering disciplines, including Biology, Robotics, Social Sciences, among others. The Best Student Paper Award is sponsored by Wolfram Research, Inc. and was awarded to the paper with the highest cumulative reviewers’ score.
