Learning without neuronsTo what extent can living and non-living things be reshaped by their history so as to acquire new function? (a) Neural computation without neural networks
Learning is usually associated with neural networks; but many of these properties come from them being highly interconnected many-body systems with collective dynamics. To what extent can inevitable processes in disordered non-equilibrium physical systems show similar neural network-like behavior? Recent collaborators: Schulman, Nagel, Winfree Recent representative work: Associative memory through molecular nucleation (arXiv), Supervised physical learning in mechanics (PNAS), A review: Learning without neurons (Ann. Rev. Cond. Mat.) Recent publications
Click here for full list + short summariesPattern recognition in the nucleation kinetics of non-equilibrium self-assembly C.G. Evans, J.O. Brien, E. Winfree, A. Murugan arXiv (2022) Learning without neurons in physical systems N. Stern, A. Murugan Annual Reviews of Condensed Matter Physics (to appear) Standardized excitable elements for scalable engineering of far-from-equilibrium chemical networks S W. Schaffter, K-L Chen, J O’Brien, M Noble, A Murugan, R Schulman Nature Chemistry (2022) Learning to self-fold at a bifurcation Arinze C, Stern M, Nagel SR, Murugan A. Physical Review E (2023) Learning to control active matter M. Falk, V. Alizadehyazdi, H. Jaeger, A. Murugan Physical Review Research 2021, arxiv Continual learning of multiple memories in mechanical networks M. Stern, M. Pinson, A. Murugan Physical Review X (Aug 2020) (arxiv version) Supervised learning through physical changes in a mechanical system M. Stern, C. Arinze, L. Perez, S. Palmer, A. Murugan PNAS (2020) Temporal pattern recognition through analog molecular computation J O'Brien, A. Murugan ACS Synthetic Biology (March 2019) Popular summary by MIT Tech Review Bioinspired nonequilibrium search for novel materials A. Murugan, H. Jaeger MRS Bulletin 44(2):96-105 pdf here Shaping the topology of folding pathways in mechanical systems with: M. Stern, V. Jayaram, Nature Communications 9:4303 (2018) The difficulty of folding self-folding origami with: M. Stern, M. Pinson Physical Review X, arXiv link (2017) Self-folding origami at any energy scale with: M. Pinson*, M. Stern*, A Ferrero, T.Witten, E. Chen Nature Communications 8:15477 (2017) Associative pattern recognition through macro-molecular self-assembly with: W. Zhong, D.J. Schwab Journal of Statistical Physics, Volume 167, Issue 3–4, May 2017 Biological implications of dynamical phases in non-equilibrium reaction networks, with: S. Vaikuntanathan invited contribution, Journal of Statistical Physics (2016, 162 (5)) Undesired usage and the robust self-assembly of heterogeneous structures, with: J. Zou, and M. Brenner Nature Communications 6, 6203 (Jan 2015) Multifarious Assembly Mixtures: Systems Allowing Retrieval of Diverse Stored Structures, with: Z. Zeravcic, S. Leibler and M. Brenner Proceedings of the National Academy of Sciences 112(1) 54-59 (Dec 2014) (b) Biological adaptation and evolution in changing environments
In physics, we tend to distinguish parameters and physical degrees of freedom. But in biology, what seem like parameters are often themselves dynamic and tuned by the system on long enough timescales. We study how biological systems can reshape their architecture and learn from their environment in a range of problems: evolution and ecology in time-varying environments, evolution of genetic variation (kinetic proofreading, standing variation), circadian biology, promiscuity and molecular specificity. Recent collaborators: Elowitz, Hwa, Pincus, Rust, Ranganathan, Szostak, Wang Recent representative work: Non-equilibrium evolution of antibodies can protect against future viruses (PNAS), Kinetic proofreading by exploiting space (eLife), Physical soft modes constrain epistasis (MBE) Recent publications
Click here for full list + short summariesNon-Convex Optimization by Hamiltonian Alternation A Apte, K Marwaha, A. Murugan, arXiv (2022) Ligand-receptor promiscuity enables cellular addressing C Su, A Murugan, J Linton, A Yeluri, J Bois, H Klumpe, Y Antebi, M Elowitz Cell Systems 2021 Roadmap on biology in time varying environments A. Murugan et al Physical Biology 2021 Proofreading through spatial gradients Vahe Galstyan, Kabir Husain, Fangzhou Xiao, Arvind Murugan+, Rob Phillips+ eLife 2020; 9:e60415 NF-κB responds to absolute differences in cytokine concentrations M Son, A G Wang, H Tu, M O Metzig, P Patel, K Husain, J Lin, A Murugan, A Hoffmann, S Tay Science Signaling 2021 Jan 19;14(666):eaaz4382 Physical constraints on epistasis K. Husain, A. Murugan Molecular Biology and Evolution (MBE) (2020) , arxiv Tuning environmental timescales to evolve and maintain generalists V. Sachdeva*, K. Husain*, J. Sheng, S. Wang+, A. Murugan+ PNAS (April 2020) Non-equilibrium statistical mechanics of continuous attractors W. Zhong, Z. Lu, D.J.Schwab+, A. Murugan+ Neural Computation (2020) pdf Kalman-like Self-Tuned Sensitivity in Biophysical Sensing K Husain, W Pittayakanchit, G Pattanayak, M J Rust, A. Murugan Cell Systems 2019, 459–465.e6 Information content of downwelling skylight for non-imaging visual systems with: R. Thiermann, A. Sweeney bioRxiv (Sep 2018) Biophysical clocks face a trade-off between internal and external noise resistance with: W Pittayakanchit*, Z Lu*, J Chew, M J Rust eLife (2018);7:e37624 High Protein Copy Number Is Required to Suppress Stochasticity in the Cyanobacterial Circadian Clock with: J. Chew, E. Leypunskiy, J. Lin, M. Rust Nature Communications 9:3004 (2018) Topologically protected modes in non-equilibrium stochastic systems with: S. Vaikuntanathan Nature Communications (Jan 2017) The Information Capacity of Specific Interactions with: M. Huntley, M. Brenner Proceedings of the National Academy of Sciences (May 2016) Receptor arrays optimized for natural odor statistics, with: D. Zwicker, M. Brenner Proceedings of the National Academy of Sciences (Apr 2016) Openings: If you are interested in working on such themes as a postdoc, graduate student or undergrad, contact amurugan@uchicago.edu. Besides physics, our work often requires combining ideas from quantitative biology, non-equilibrium dynamics and theoretical computer science. Openings Dec 2022: We have one opening for theory of learning in physical systems and one opening for theory+experiments of molecular evolution in non-equilibrium regimes. Our work is supported by the NSF and the Simons, Sloan and Moore Foundations. Lab members have been supported by the James S. McDonnell Foundation, Hertz and NSF fellowships. |
|
