Brian: a simulator for spiking neural networks in Python

Goodman, Dan F. M.; Brette, Romain

doi:10.3389/neuro.11.005.2008

Cited by 396 publications

(219 citation statements)

References 13 publications

(16 reference statements)

Supporting

Mentioning

208

Contrasting

Order By: Relevance

“…In particular we implemented the same network in Neuron (Carnevale and Hines, 2006), NEST (Gewaltig and Diesmann, 2007), Brian (Goodman and Brette, 2008) and our own simulator Auryn (Methods). The network was tuned initially to a parameter regime in which the model exhibits stable asynchronous irregular activity over extended periods of time (Figure 2A).…”

Section: Resultsmentioning

confidence: 99%

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Zenke

Gerstner

2014

Front. Neuroinform.

View full text Add to dashboard Cite

To understand how the central nervous system performs computations using recurrent neuronal circuitry, simulations have become an indispensable tool for theoretical neuroscience. To study neuronal circuits and their ability to self-organize, increasing attention has been directed toward synaptic plasticity. In particular spike-timing-dependent plasticity (STDP) creates specific demands for simulations of spiking neural networks. On the one hand a high temporal resolution is required to capture the millisecond timescale of typical STDP windows. On the other hand network simulations have to evolve over hours up to days, to capture the timescale of long-term plasticity. To do this efficiently, fast simulation speed is the crucial ingredient rather than large neuron numbers. Using different medium-sized network models consisting of several thousands of neurons and off-the-shelf hardware, we compare the simulation speed of the simulators: Brian, NEST and Neuron as well as our own simulator Auryn. Our results show that real-time simulations of different plastic network models are possible in parallel simulations in which numerical precision is not a primary concern. Even so, the speed-up margin of parallelism is limited and boosting simulation speeds beyond one tenth of real-time is difficult. By profiling simulation code we show that the run times of typical plastic network simulations encounter a hard boundary. This limit is partly due to latencies in the inter-process communications and thus cannot be overcome by increased parallelism. Overall, these results show that to study plasticity in medium-sized spiking neural networks, adequate simulation tools are readily available which run efficiently on small clusters. However, to run simulations substantially faster than real-time, special hardware is a prerequisite.

show abstract

Section: Resultsmentioning

confidence: 99%

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Zenke

Gerstner

2014

Front. Neuroinform.

View full text Add to dashboard Cite

show abstract

“…RNN learning is used for context-sensitive languages recognition and is a difficult and often increasing problem for standard recurrent neural networks (RNNs), because it requires unlimited memory resources. Authors (Goodman, Brette 2008;Schmidhuber et al 2005;Schmidhuber et al 2006) found that Evolino based LSTM learns on average faster and it is able to generalize substantially better that gradient-based LSTM. It is possible using Evolino to learn functions composed of multiple superimposed oscillators such as double sine and triple sine.…”

Section: Description Of Evolino Neural Networkmentioning

confidence: 94%

Application of Neural Network for Forecasting of Exchange Rates and Forex Trading

Maknickienė

Maknickas

2012

The 7th International Scientific Conference "Business and Management 2012". Selected Papers

View full text Add to dashboard Cite

Expert methods, which widely applied for human decision making, were employed for neural networks. It was developed an exchange rates prediction and trading algorithm with using of experts information processing techniques -Delphi method and prediction compatibility. Proposed algorithm limited to eight experts. Each of experts represented recurrent neural network, Evolino-based Long Short-Term Memory (LSTM) by using of genetic learning algorithm, EVOlution of recurrent systems with LINear Outputs (EVOLINO). Statistical investigation of offered algorithm shows the significantly increase of the reliability of prediction. Developed algorithm was applied for trading of historical forex exchange rates. Obtained test trading results were presented

show abstract

“…The Brian simulator (Goodman and Brette 2008) has become very popular among neuroscientists in just a few years and it has been ranked 2nd most popular neural simulator in a recent survey (Hanke and Halchenko 2011), just after the well known Neuron software (Carnevale and Hines 2006). The reason behind this success is certainly the intuitive and highly flexible interface offered by Brian as well as efficient vectorized computations that make Brian quite competitive with regular C code for large networks.…”

Section: Brian Limitationsmentioning

confidence: 97%

DANA: Distributed numerical and adaptive modelling framework

Rougier

Fix

2012

Network: Computation in Neural Systems

View full text Add to dashboard Cite

DANA is a python framework ( http://dana.loria.fr ) whose computational paradigm is grounded on the notion of a unit that is essentially a set of time dependent values varying under the influence of other units via adaptive weighted connections. The evolution of a unit's value are defined by a set of differential equations expressed in standard mathematical notation which greatly ease their definition. The units are organized into groups that form a model. Each unit can be connected to any other unit (including itself) using a weighted connection. The DANA framework offers a set of core objects needed to design and run such models. The modeler only has to define the equations of a unit as well as the equations governing the training of the connections. The simulation is completely transparent to the modeler and is handled by DANA. This allows DANA to be used for a wide range of numerical and distributed models as long as they fit the proposed framework (e.g. cellular automata, reaction-diffusion system, decentralized neural networks, recurrent neural networks, kernel-based image processing, etc.).

show abstract

Brian: a simulator for spiking neural networks in Python

Cited by 396 publications

References 13 publications

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Limits to high-speed simulations of spiking neural networks using general-purpose computers

Application of Neural Network for Forecasting of Exchange Rates and Forex Trading

DANA: Distributed numerical and adaptive modelling framework

Contact Info

Product

Resources

About