Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
167 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

When to be critical? Performance and evolvability in different regimes of neural Ising agents (2303.16195v4)

Published 28 Mar 2023 in cs.NE

Abstract: It has long been hypothesized that operating close to the critical state is beneficial for natural, artificial and their evolutionary systems. We put this hypothesis to test in a system of evolving foraging agents controlled by neural networks that can adapt agents' dynamical regime throughout evolution. Surprisingly, we find that all populations that discover solutions, evolve to be subcritical. By a resilience analysis, we find that there are still benefits of starting the evolution in the critical regime. Namely, initially critical agents maintain their fitness level under environmental changes (for example, in the lifespan) and degrade gracefully when their genome is perturbed. At the same time, initially subcritical agents, even when evolved to the same fitness, are often inadequate to withstand the changes in the lifespan and degrade catastrophically with genetic perturbations. Furthermore, we find the optimal distance to criticality depends on the task complexity. To test it we introduce a hard and simple task: for the hard task, agents evolve closer to criticality whereas more subcritical solutions are found for the simple task. We verify that our results are independent of the selected evolutionary mechanisms by testing them on two principally different approaches: a genetic algorithm and an evolutionary strategy. In summary, our study suggests that although optimal behaviour in the simple task is obtained in a subcritical regime, initializing near criticality is important to be efficient at finding optimal solutions for new tasks of unknown complexity.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (42)
  1. Miguel Aguilera and Manuel G Bedia “Criticality as It Could Be: organizational invariance as self-organized criticality in embodied agents” In Artificial Life Conference Proceedings 14, 2017, pp. 21–28 MIT Press
  2. “Robustness and evolvability in genetic regulatory networks” In Journal of theoretical biology 245.3 Elsevier, 2007, pp. 433–448
  3. “Critical dynamics in genetic regulatory networks: examples from four kingdoms” In PLoS One 3.6 Public Library of Science, 2008, pp. e2456
  4. J. Beggs “The criticality hypothesis: how local cortical networks might optimize information processing” In Proceedings of the Royal Society A., 2007
  5. “Neuronal avalanches are diverse and precise activity patterns that are stable for many hours in cortical slice cultures” In J. Neurosci 24.22, 2004, pp. 5216–5229
  6. “Real-time computation at the edge of chaos in recurrent neural networks” In Neural computation 16.7, 2004, pp. 1413–1436 URL: http://www.mitpressjournals.org/doi/abs/10.1162/089976604323057443
  7. “Information processing in echo state networks at the edge of chaos” In Theory in Biosciences 131.3, 2012, pp. 205–213 URL: http://link.springer.com/article/10.1007/s12064-011-0146-8
  8. “Feedback Mechanisms for Self-Organization to the Edge of a Phase Transition” In Frontiers in Physics 8, 2020, pp. 333 DOI: 10.3389/fphy.2020.00333
  9. “Scale-free correlations in starling flocks” In Proceedings of the National Academy of Sciences 107.26 National Acad Sciences, 2010, pp. 11865–11870
  10. Hugues Chaté and Miguel A Muñoz “Insect swarms go critical” In Physics 7 APS, 2014, pp. 120
  11. “Control of criticality and computation in spiking neuromorphic networks with plasticity” In Nature Communications 11, 2020, pp. 2853 DOI: 10.1038/s41467-020-16548-3
  12. K De Jong “Evolutionary Computation: A Unified Approach” MIT Press, 2006 URL: https://ieeexplore.ieee.org/servlet/opac?bknumber=6267245
  13. Giovanna De Palo, Darvin Yi and Robert G Endres “A critical-like collective state leads to long-range cell communication in Dictyostelium discoideum aggregation” In PLoS biology 15.4 Public Library of Science San Francisco, CA USA, 2017, pp. e1002602
  14. Julianne D Halley, Frank R Burden and David A Winkler “Stem cell decision making and critical-like exploratory networks” In Stem Cell Research 2.3 Elsevier, 2009, pp. 165–177
  15. “Information-based fitness and the emergence of criticality in living systems” In Proceedings of the National Academy of Sciences 111.28 National Acad Sciences, 2014, pp. 10095–10100
  16. “Towards a general theory of adaptive walks on rugged landscapes” In Journal of theoretical Biology 128.1 Elsevier, 1987, pp. 11–45
  17. Stuart A Kauffman “The origins of order: Self-organization and selection in evolution” Oxford University Press, 1993
  18. “Evolution Towards Criticality in Ising Neural Agents” In Artificial Life 26.1 MIT Press, 2020, pp. 112–129
  19. “Optimal dynamical range of excitable networks at criticality.” In Nature Physics 2, 2006, pp. 348–352
  20. Osame Kinouchi, Renata Pazzini and Mauro Copelli “Mechanisms of Self-Organized Quasicriticality in Neuronal Network Models” In Frontiers in Physics 8, 2020, pp. 530 DOI: 10.3389/fphy.2020.583213
  21. “Equation of state calculations by fast computing machines” In The journal of chemical physics 21.6 American Institute of Physics, 1953, pp. 1087–1092
  22. “Are biological systems poised at criticality?” In Journal of Statistical Physics 144.2 Springer, 2011, pp. 268–302
  23. Miguel A Munoz “Colloquium: Criticality and dynamical scaling in living systems” In Reviews of Modern Physics 90.3 APS, 2018, pp. 031001
  24. “Meta-Learning through Hebbian Plasticity in Random Networks” In Advances in Neural Information Processing Systems 33, 2020, pp. 20719–20731 URL: https://proceedings.neurips.cc/paper/2020/file/ee23e7ad9b473ad072d57aaa9b2a5222-Paper.pdf
  25. “Subcritical escape waves in schooling fish” In arXiv preprint arXiv:2108.05537, 2021
  26. “Spike avalanches in vivo suggest a driven, slightly subcritical brain state” In Frontiers in Systems Neuroscience 8 Frontiers, 2014, pp. 108
  27. “The dynamical regime and its importance for evolvability, task performance and generalization” In ALIFE 2021: The 2021 Conference on Artificial Life, 2021, pp. 1–9 MIT Press DOI: 10.1162/isal˙a˙00412
  28. P Rämö, J Kesseli and O Yli-Harja “Perturbation avalanches and criticality in gene regulatory networks” In Journal of Theoretical Biology 242.1 Elsevier, 2006, pp. 164–170
  29. “Measures for information propagation in Boolean networks” In Physica D: Nonlinear Phenomena 227.1 Elsevier, 2007, pp. 100–104
  30. “Dynamical criticality: overview and open questions” In Journal of Systems Science and Complexity 31.3 Springer, 2018, pp. 647–663
  31. “Evolution strategies as a scalable alternative to reinforcement learning” In arXiv preprint arXiv:1703.03864, 2017
  32. “Weak pairwise correlations imply strongly correlated network states in a neural population” In Nature 440.7087 Nature Publishing Group, 2006, pp. 1007–1012
  33. “Parameter-exploring policy gradients” The 18th International Conference on Artificial Neural Networks, ICANN 2008 In Neural Networks 23.4, 2010, pp. 551–559 DOI: https://doi.org/10.1016/j.neunet.2009.12.004
  34. “Thermodynamics and signatures of criticality in a network of neurons” In PNAS 112.37, 2015 DOI: 10.1073/pnas.1514188112
  35. Nergis Tomen, David Rotermund and Udo Ernst “Marginally subcritical dynamics explain enhanced stimulus discriminability under attention” In Frontiers in systems neuroscience 8 Frontiers, 2014, pp. 151
  36. “Intrinsic noise and deviations from criticality in Boolean gene-regulatory networks” In Scientific Reports 6 Nature Publishing Group, 2016, pp. 34743
  37. “Evolving generalists in switching rugged landscapes” In PLoS computational biology 15.10 Public Library of Science San Francisco, CA USA, 2019, pp. e1007320
  38. “Natural evolution strategies” In Proceedings of the 2008 Congress on Evolutionary Computation. CEC’08, 2008 IEEE
  39. “Natural evolution strategies” In The Journal of Machine Learning Research 15.1 JMLR. org, 2014, pp. 949–980
  40. “Between perfectly critical and fully irregular: A reverberating model captures and predicts cortical spike propagation” In Cerebral Cortex 29.6 Oxford University Press, 2019, pp. 2759–2770
  41. Roxana Zeraati, Viola Priesemann and Anna Levina “Self-organization toward criticality by synaptic plasticity” In Frontiers in Physics 9, 2021, pp. 103 DOI: 10.3389/fphy.2021.619661
  42. “Tailored ensembles of neural networks optimize sensitivity to stimulus statistics” In Physical Review Research 2.1 American Physical Society, 2020, pp. 013115
Citations (7)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now