Reservoir computing with swarms

Lymburn, Thomas; Algar, Shannon D.; Small, Michael; Jüngling, Thomas

doi:10.1063/5.0039745

Cited by 11 publications

(7 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, it has been demonstrated that the computational capacity of RC is maximized at "the edge of chaos (stability)" [23,25,26], namely the critical state at the transition from the ordered to chaotic states, and, to achieve the optimal performance, the machine should be tuned to the vicinity of this critical state. This finding is consistent with the findings in cellular automata [27,28], and is widely regarded as one of the important principles for designing RC [10,[29][30][31]. This principle, however, has been questioned by some recent studies [20], which show that for some types of nodal dynamics the prediction performance might be deteriorated when the reservoir is close to the critical state.…”

supporting

confidence: 87%

Criticality in Reservoir Computer of Coupled Phase Oscillators

Wang,

Fan,

Xiao

et al. 2021

Preprint

View full text Add to dashboard Cite

Accumulating evidences show that the cerebral cortex is operating near a critical state featured by powerlaw size distribution of neural avalanche activities, yet evidence of this critical state in artificial neural networks mimicking the cerebral cortex is lacking. Here we design an artificial neural network of coupled phase oscillators and, by the technique of reservoir computing in machine learning, train it for predicting chaos. It is found that when the machine is properly trained, oscillators in the reservoir are synchronized into clusters whose sizes follow a power-law distribution. This feature, however, is absent when the machine is poorly trained. Additionally, it is found that despite the synchronization degree of the original network, once properly trained, the reservoir network is always developed to the same critical state, exemplifying the "attractor" nature of this state in machine learning. The generality of the results is verified in different reservoir models and by different target systems, and it is found that the scaling exponent of the distribution is independent on the reservoir details and the bifurcation parameter of the target system, but is modified when the dynamics of the target system is changed to a different type. The findings shed lights on the nature of machine learning, and are helpful to the design of high-performance machine in physical systems.

show abstract

supporting

confidence: 87%

Criticality in Reservoir Computer of Coupled Phase Oscillators

Wang,

Fan,

Xiao

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…It also provides the coupling of the active particle dynamics to its past allowing to implement virtual nodes living on a single physical active particle system via time-multiplexing [12,44]. The information processing that is provided by such a single node may therefore extend the rare simulation work on reservoir computing with active particle swarms with interparticle coupling [37]. Additionally, the nonlinearity that is required for computations is an intrinsic physical property of our active particle system and requires no extra treatment of the output signal [52].…”

Section: Discussionmentioning

confidence: 99%

“…Information processing [28,29] and learning [30] in experiment [31] and simulation [32][33][34][35][36] have also entered the field of synthetic active matter. However, studies that extend the use of synthetic microparticles as computational substrates are still rare or purely computational [37].…”

Section: Introductionmentioning

confidence: 99%

Harnessing Synthetic Active Particles for Physical Reservoir Computing

Cichos,

Wang

2023

Preprint

View full text Add to dashboard Cite

The processing of information is an indispensable property of living systems realized by networks of active processes with enormous complexity. They have inspired many variants of modern machine learning one of them being reservoir computing, in which stimulating a network of nodes with fading memory enables computations and complex predictions. Reservoirs are implemented on computer hardware, but also on unconventional physical substrates such as mechanical oscillators, spins, or bacteria often summarized as physical reservoir computing. Here we demonstrate physical reservoir computing with a synthetic active microparticle system that self-organizes from an active and passive component into inherently noisy nonlinear dynamical units. The self-organization and dynamical response of the unit is the result of a delayed propulsion of the microswimmer to a passive target. A reservoir of such units with a self-coupling via the delayed response can perform predictive tasks despite the strong noise resulting from the Brownian motion of the microswimmers. To achieve efficient noise suppression, we introduce a special architecture that uses historical reservoir states for output. Our results pave the way for the study of information processing in synthetic self-organized active particle systems.

show abstract

“…The intuition behind the power of nonlinear dynamics in an organism arises from a reservoirs' ability to project the incoming dynamics up onto a high dimensional space through which a linear classifier can draw a plane to separate classes. These techniques have recently been extended to study the ca-pacity of flocking 'boids' [84].…”

Section: The Tissue As a Physical Reservoirmentioning

confidence: 99%

Ciliary flocking and emergent instabilities enable collective agility in a non-neuromuscular animal

Bull,

Prakash,

Prakash

2021

Preprint

View full text Add to dashboard Cite

Effective organismal behavior is characterized by the ability to respond appropriately to changes in the surrounding environment. Attaining this delicate balance of sensitivity and stability is a hallmark of the animal kingdom. By studying the locomotory behavior of a simple animal (Trichoplax adhaerens) without muscles or neurons, here we demonstrate how monociliated epithelial cells work collectively to give rise to an agile non-neuromuscular organism. Via direct visualization of large ciliary arrays, we report the discovery of sub-second ciliary reorientations under a rotational torque that is mediated by collective tissue mechanics and the adhesion of cilia to the underlying substrate.In an illuminating toy model, we show a mapping of this system onto an "active-elastic resonator". This framework explains how perturbations propagate information in this array as linear speed traveling waves in response to mechanical stimulus. Next, we explore the implications of periodic and noisy parametric driving in this active-elastic resonator and show that such driving can excite mechanical 'spikes'. These spikes in collective mode amplitudes are consistent with a system driven by parametric amplification and a saturating nonlinearity. We conduct extensive numerical experiments to corroborate these findings within a polarized active-elastic sheet. These results indicate that periodic and stochastic forcing are critical for increasing the sensitivity of collective ciliary flocking. We support these theoretical predictions via direct experimental observation of linear speed traveling waves which are made possible through the hybridization of spin and overdamped density waves. We map how these ciliary flocking dynamics result in agile motility via coupling between an amplified resonator and a tuning (Goldstone-like) mode of the system. This sets the stage for how activity and elasticity can self-organize into behavior which benefits the organism as a whole.

show abstract

Reservoir computing with swarms

Cited by 11 publications

References 31 publications

Criticality in Reservoir Computer of Coupled Phase Oscillators

Criticality in Reservoir Computer of Coupled Phase Oscillators

Harnessing Synthetic Active Particles for Physical Reservoir Computing

Ciliary flocking and emergent instabilities enable collective agility in a non-neuromuscular animal

Contact Info

Product

Resources

About