1 / 31

Artificial Life and Evolving Intelligence

Artificial Life and Evolving Intelligence. Laura M. Grabowski, Ph.D. Department of Computer Science The University of Texas-Pan American. What is Life?. Alife and Evolution. Avida. Evolving Intelligence. Conclusion. Intro to Artificial Life. What is Artificial Life?. Computer Science.

robert
Download Presentation

Artificial Life and Evolving Intelligence

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Artificial Life and Evolving Intelligence Laura M. Grabowski, Ph.D. Department of Computer Science The University of Texas-Pan American

  2. What is Life? Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  3. What is Artificial Life? Computer Science Biology Artificial Life Engineering Philosophy Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  4. Situating Artificial Life (Alife) in Computer Science Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  5. AI Run Amok! Or, Gratuitous Pop Culture References Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  6. Evolutionary Computation • Subfield of Artificial Intelligence (AI) • Methods apply principles of Darwinian evolution to problem-solving • EC methods can produce patentable, human-competitive solutions • EC systems contain • One or more populations of individuals • Competition for resources Image source: http://sci2s.ugr.es/keel/links.php Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  7. EC Overview • Genetic algorithm • Iterative search • Individuals are encodings of candidate solutions • Optimization problems • Genetic programming • Individuals are computer programs • Digital evolution • Evolutionary robotics Evolved satellite antenna design Image source: www.egr.msu.edu/~goodman/GECSummitIntroToGA_Tutorial-goodman.pdf Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  8. What is Alife? • Common thread: understanding the “general prinicples that govern the living state” 2 • Broad range of disciplines • Computer science • Engineering • Biochemistry • Physics • … and more … 2 Adami, C. (1998). Introduction to Artificial Life. Telos, Santa Clara, CA, p. 34. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  9. “What makes living systems alive?”1 • How do we approach this question? • Life is complex and diverse • Deconstructing living system (analysis)? • Constructing artificial living systems (synthesis)! Image source: http://www.fws.gov 1 Adami, C. (1998). Introduction to Artificial Life. Telos, Santa Clara, CA, p. vii. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  10. AlifeThemes • Construct life in artificial medium • Compare with terrestrial life • Desire to learn about living world (emulation, simulation) • Alife may give results that falsify theories about life • Alife may provide novel or improved solutions to engineering problems Image source: http://www.karlsims.com/evolved-virtual-creatures.html Image source: http://www.digitalspace.com/avatars/book/fullbook/papers/ngarden.htm Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  11. Alife and Evolution • Alife is a part of the field of Evolutionary Computation • Alife approaches often leverage evolutionary processes Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  12. Fundamental Principles of Evolution* • Replication • Descent with modification • Competition *D. C. Dennett. (2002). The new replicators. In M. Pagel, editor, Encyclopedia of Evolution. Oxford University Press, New York, E83-E92. Charles Darwin, age 31 Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  13. Brief Highlights from Alife Work • Karl Sims Creatures, 1994 • Hod Lipson, self-copying robots • Risto Mikkulainen’s group, 2008 • Big Dog robot Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  14. Digital Evolution • Population of self-replicating “digital organisms” • Organisms replicate, mutate, compete Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  15. The Avida* Digital Evolution Platform • Virtual environment (“vitrual Petri dish), but real evolution** • Used for research in biology and computer science • * C. Ofria and C. O. Wilke. (2004). Avida: a software platform for research in computational evolutionary biology. In Artificial Life 10, 191-229. • ** Pennock, R.T. (2007). Models, simulations, instantiations, and evidence: the case of digital evolution. Journal of Experimental and Theoretical Artificial Intelligence, 19(1):29-42. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  16. Avida Overview Figure from: Lenski, R. E., Ofria, C., Pennock, R.T., & Adami, C. (2003). The evolutionary origin of complex features. Nature, 423, 139-144. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  17. Evolving Intelligence • Top-down vs. bottom-up approaches • Goals • Behavioral flexibility • Emergence • “Intelligence” is not one big ability, but many smaller ones • Building blocks Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  18. Building Blocks of Intelligence • Common threads • Differential behavior based on current environment • Using past experience • Requires such capabilities as • Sensing • Memory • Decision-making Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  19. Purpose: Proof-of-concept, evolving simple navigation Evolving Motility and Taxis Using Avida • Inspired by chemical gradient-following behavior of E. coli • Evolution of chemotaxis-like response* *Grabowski, L.M., Elsberry, W.R., Ofria, C., and Pennock, R.T. (2008). On the evolution of motility and intelligent tactic response. GECCO '08: Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation, pp. 209-216. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  20. Experimental Environment • Organisms evolved to traverse idealized gradient • Two treatments • Implicit memory: provided prior information • Without implicit memory: did not provide prior information Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  21. Representative Trajectory Plots of Evolved Organisms Implicit Memory 26/100 replicate populations* Evolved Memory 7/100 replicate populations* Organism Trajectory Peak Location Initial Location *Closest approach within 10% of initial distance to peak Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  22. Key result: Evolution of rudimentary memory mechanism Conclusions • Proof-of-concept for motility in Avida • Evolution of fundamental navigation • Tactic behavior easier to evolve when memory is provided Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  23. Memory: Critical Component of Intelligence Memory is a hurdle in evolving intelligent behavior. Primary hippocampal neuron Image source: http://www.unm.edu/~neurohsc/ZhaoLab/links/index.html Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  24. Evolving Memory Use • Bees: An example of behavioral intelligence • Variety of navigation strategies • Show use of different memory capabilities Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  25. Path-following Experiments in Avida • Path-following experiments with honey bees* • Similar path environment in Avida • User-defined set of environmental “cues” *Zhang, S.W., Bartsch, K. & Srinivasan, M. V. (1996). Maze learning by honeybees. Neurobiology of Learning and Memory, 66:267-282. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  26. Evolving Short-Term Memory • Intermittent updating of information • “New” turn direction cued by specific right or left cue • Other turns have different cue, repeat-last Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  27. Results: Paths Experienced During Evolution Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  28. Results: Novel Path Grabowski, L. M., Bryson, D. M., Pennock, R. T., Dyer, F. C., & Ofria, C. (2010). Early evolution of memory use in digital organisms. Proceedings of the 12th International Conference on the Synthesis and Simulation of Living Systems (ALife XII). MIT Press, pp. 224-231. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  29. Conclusions • Memory and flexible behavior evolve in even simple environments. • Evolution capitalizes on both environmental change and environmental regularity to construct solutions. • Evolved memory mechanisms use both organization of genome and volatile states in the CPU. Alife and Evolution Avida Evolving Intelligence Conclusion Intro to Artificial Life

  30. Future Directions • Evolving Navigation Behaviors • Path Integration • Landmark Navigation • Evolution of complexity • Transfer from Avida to robotic platforms • Global domination Rudimentary Memory Precursors to Memory One-bit Memory Conclusion Introduction

  31. Thank you! • For more information: • Laura M. Grabowski’s Homepage • http://www.cs.panam.edu/~lmgrabowski • Avida Digital Life Platform • http://avida.devosoft.org/ • Michigan State University Digital Evolution Laboratory • http://devolab.msu.edu/ • Michigan State University Evolving Intelligence Project • https://www.msu.edu/~pennock5/research/EI.html

More Related