Iterative Specification as a Modeling and Simulation Formalism for I/O General Systems

Muzy, Alexandre; Zeigler, Bernard P.; Grammont, Franck

doi:10.1109/jsyst.2017.2728861

Cited by 8 publications

(6 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DEVS is a system theoretic characterization of discrete event simulations based on abstraction of events and time intervals from continuous data streams [15][16][17]. Such abstractions carry information that can be efficiently employed, not only in simulation, but also in accounting for the real-world constraints that shape cognitive information processes [18,19].…”

Section: Review Of Devs Abstractions For Brain Architecturesmentioning

confidence: 99%

DEVS-Based Building Blocks and Architectural Patterns for Intelligent Hybrid Cyberphysical System Design

Zeigler

2021

Information

View full text Add to dashboard Cite

The DEVS formalism has been recognized to support generic open architectures that allow incorporating multiple engineering domains within integrated simulation models. What is missing for accelerated adoption of DEVS-based methodology for intelligent cyberphysical system design is a set of building blocks and architectural patterns that can be replicated and reused in system development. As a start in this direction, this paper offers a notional architecture for intelligent hybrid cyberphysical system design and proceeds to focus on the decision layer to consider DEVS models for basic behaviors such as choice of alternatives, perception of temporal event relations, and recognition and generation of finite state languages cast into DEVS time segments. We proceed to describe a methodology to define DEVS-based building blocks and architectural patterns for design of systems employing fast, frugal, and accurate heuristics. We identify some elements of this kind and establish their status as minimal realizations of their defined behaviors. As minimal realizations such designs must ipso facto underlie any implementation of the same cognitive behaviors. We discuss architectures drawn from the cognitive science literature to show that the fundamental elements drawn from the fast, frugal, and accurate paradigm provide insights into intelligent hybrid cyberphysical system design. We close with open questions and research needed to confirm the proposed concepts.

show abstract

Section: Review Of Devs Abstractions For Brain Architecturesmentioning

confidence: 99%

DEVS-Based Building Blocks and Architectural Patterns for Intelligent Hybrid Cyberphysical System Design

Zeigler

2021

Information

View full text Add to dashboard Cite

show abstract

“…However, the spikes must be transmitted between pre and post synaptic neurons, between two time steps. With more advanced techniques such as hybrid parallel computation [19], the spikes are computed in parallel for each neuron thanks to discrete event programming, which allows a large precision, while the exchange at the level of the synapses can be done at a less fine scale in a discretized way [37,3,45,50]. Using hybrid parallel computations, in [24] the authors were able to simulate a network of order 10 6 neurons and 10 10 synapses with parallel supercomputers capable to run in parallel tens of millions of threads.…”

Section: Introductionmentioning

confidence: 99%

Scalability of Large Neural Network Simulations via Activity Tracking With Time Asynchrony and Procedural Connectivity

2022

Self Cite

View full text Add to dashboard Cite

We present a new algorithm to efficiently simulate random models of large neural networks satisfying the property of time asynchrony. The model parameters (average firing rate, number of neurons, synaptic connection probability, and postsynaptic duration) are of the order of magnitude of a small mammalian brain or of human brain areas. Through the use of activity tracking and procedural connectivity (dynamical regeneration of synapses), computational and memory complexities of this algorithm are proved to be theoretically linear with the number of neurons. These results are experimentally validated by sequential simulations of millions of neurons and billions of synapses running in a few minutes using a single thread of an equivalent desktop computer.

show abstract

“…At each time step, each process corresponding to a different neuron has to wait for the calculations of all the other processes to know what needs to be updated before computing the next step. With time asynchrony, we can leverage discrete-event programming (25)(26)(27)(28) to track the whole system in time by jumps: from one spike in the network to another spike in the network. Since a very small percentage of a brain is firing during a given unit of time (29), the gain we have is tremendous in terms of computations.…”

Section: Introductionmentioning

confidence: 99%

“…However, the spikes must be transmitted between pre and post synaptic neurons, between two time steps. With more advanced techniques such as hybrid parallel computation [19], the spikes are computed in parallel for each neuron thanks to discrete event programming, while the exchange at the level of the synapses can be done at a less fine scale in a discretized way [36, 3, 44, 49]. Nevertheless, in both cases (classical or hybrid parallel computation), the synchronization between neurons is necessary to propagate the spikes at the synapses, even if it is not done at the same time scale.…”

Section: Introductionmentioning

confidence: 99%

“…However, the spikes must be transmitted between pre and post synaptic neurons, between two time steps. With more advanced techniques such as hybrid parallel computation [19], the spikes are computed in parallel for each neuron thanks to discrete event programming, which allows a large precision, while the exchange at the level of the synapses can be done at a less fine scale in a discretized way [37, 3, 45, 50]. Using hybrid parallel computations, in [24] the authors were able to simulate a network of order 10 6 neurons and 10 10 synapses with parallel supercomputers capable to run in parallel tens of millions of threads.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Scalability of large neural network simulations via activity tracking with time asynchrony and procedural connectivity

Mascart

Reynaud-Bouret²,

Muzy

2021

Preprint

Self Cite

View full text Add to dashboard Cite

We present here a new algorithm based on a random model for simulating efficiently large brain neuronal networks. Model parameters (mean firing rate, number of neurons, synaptic connection probability and postsynaptic duration) are easy to calibrate further on real data experiments. Based on time asynchrony assumption, both computational and memory complexities are proved to be theoretically linear with the number of neurons. These results are experimentally validated by sequential simulations of millions of neurons and billions of synapses in few minutes on a single processor desktop computer.Significance StatementTime asynchrony refers to models that prevent any two neurons to fire at the exact same time (up to the numerical precision). Brain models that are time asynchronous have a huge numerical advantage on other ones: they can be easily simulated without huge parallelization or use of GPU. Indeed, this work proposes to exploit the time asynchrony property to derive new simulation algorithms, whose computational and memory complexity can be estimated beforehand. Application on realistic set of parameters show that they can indeed easily reach the size of a small monkey brain, on a usual single processor desktop computer.

show abstract

Iterative Specification as a Modeling and Simulation Formalism for I/O General Systems

Cited by 8 publications

References 19 publications

DEVS-Based Building Blocks and Architectural Patterns for Intelligent Hybrid Cyberphysical System Design

DEVS-Based Building Blocks and Architectural Patterns for Intelligent Hybrid Cyberphysical System Design

Scalability of Large Neural Network Simulations via Activity Tracking With Time Asynchrony and Procedural Connectivity

Scalability of large neural network simulations via activity tracking with time asynchrony and procedural connectivity

Contact Info

Product

Resources

About