Filters








654 Hits in 4.0 sec

Gibbs Sampling with Low-Power Spiking Digital Neurons [article]

Srinjoy Das, Bruno Umbria Pedroni, Paul Merolla, John Arthur, Andrew S. Cassidy, Bryan L. Jackson, Dharmendra Modha, Gert Cauwenberghs, Ken Kreutz-Delgado
2015 arXiv   pre-print
This paper demonstrates such an implementation on an array of digital spiking neurons with stochastic leak and threshold properties for inference tasks and presents some key performance metrics for such  ...  Inference and learning in these algorithms uses a Markov Chain Monte Carlo procedure called Gibbs sampling.  ...  Our proposed method of realization of the sigmoidal function with low-power, digital integrate-and-fire neurons is well suited for Gibbs sampling in RBMs and DBNs with parallel arrays of visible and hidden  ... 
arXiv:1503.07793v2 fatcat:cioz5qxolfgo5ew5vr5rypxvim

A Nonparametric Framework for Quantifying Generative Inference on Neuromorphic Systems [article]

Ojash Neopane, Srinjoy Das, Ery Arias-Castro, Kenneth Kreutz-Delgado
2016 arXiv   pre-print
Generative inference in such algorithms can be performed very efficiently on hardware using a Markov Chain Monte Carlo procedure called Gibbs sampling, where stochastic samples are drawn from noisy integrate  ...  and fire neurons implemented on neuromorphic substrates.  ...  on such substrates for ultra low-power, realtime applications.  ... 
arXiv:1602.05996v1 fatcat:mf6up2zhrjapxdymbdrssmcgsq

Mapping Generative Models onto a Network of Digital Spiking Neurons

Bruno U. Pedroni, Srinjoy Das, John V. Arthur, Paul A. Merolla, Bryan L. Jackson, Dharmendra S. Modha, Kenneth Kreutz-Delgado, Gert Cauwenberghs
2016 IEEE Transactions on Biomedical Circuits and Systems  
For this, we first propose a method of producing the Gibbs sampler using bio-inspired digital noisy integrate-and-fire neurons.  ...  Next, we describe the process of mapping generative RBMs trained offline onto the IBM TrueNorth neurosynaptic processor -- a low-power digital neuromorphic VLSI substrate.  ...  THE TRUENORTH DIGITAL NEUROSYNAPTIC PROCESSOR IBM's TrueNorth is a very low-power, brain-inspired digital neurosynaptic processor [8] , with 4096 cores, totaling 1 million programmable spiking neurons  ... 
doi:10.1109/tbcas.2016.2539352 pmid:27214915 fatcat:pfygt6j6tbc5rgpv2yuoyfgvbm

Event-driven contrastive divergence for spiking neuromorphic systems

Emre Neftci, Srinjoy Das, Bruno Pedroni, Kenneth Kreutz-Delgado, Gert Cauwenberghs
2014 Frontiers in Neuroscience  
Their implementation on neuromorphic hardware platforms emulating large-scale networks of spiking neurons can have significant advantages from the perspectives of scalability, power dissipation and real-time  ...  We demonstrate our approach by training an RBM composed of leaky I&F neurons with STDP synapses to learn a generative model of the MNIST hand-written digit dataset, and by testing it in recognition, generation  ...  drive the rest of the network without tending to induce spike-to-spike correlations (e.g., synchrony), which is incompatible with the assumptions made for sampling with I&F neurons and event-driven CD  ... 
doi:10.3389/fnins.2013.00272 pmid:24574952 pmcid:PMC3922083 fatcat:erazltqd55ghffe2ge4db4ynny

Spiking neurons with short-term synaptic plasticity form superior generative networks

Luziwei Leng, Roman Martel, Oliver Breitwieser, Ilja Bytschok, Walter Senn, Johannes Schemmel, Karlheinz Meier, Mihai A. Petrovici
2018 Scientific Reports  
We thereby uncover a powerful computational property of the biologically inspired, local, spike-triggered synaptic dynamics based simply on a limited pool of synaptic resources, which enables them to deal  ...  We demonstrate how similar results can be achieved in spiking networks endowed with local short-term synaptic plasticity.  ...  After training, we compared the generative output of a Gibbs sampler, an AST sampler and a spiking network with STP.  ... 
doi:10.1038/s41598-018-28999-2 pmid:30006554 pmcid:PMC6045624 fatcat:kc6emgbx45ddtk4aeruonlaw64

Stochastic Synapses Enable Efficient Brain-Inspired Learning Machines

Emre O. Neftci, Bruno U. Pedroni, Siddharth Joshi, Maruan Al-Shedivat, Gert Cauwenberghs
2016 Frontiers in Neuroscience  
The spiking neuron-based synaptic sampling machines outperform existing spike-based unsupervised learners, while potentially offering substantial advantages in terms of power and complexity, and are thus  ...  Synaptic sampling machines perform equally well using discrete-timed artificial units (as in Hopfield networks) or continuous-timed leaky integrate & fire neurons.  ...  As in RBMs, the dSSM was sampled using Gibbs sampling. The continuous-time, spiking SSM consisted of a network of deterministic I&F spiking neurons, connected through stochastic (blank-out) synapses.  ... 
doi:10.3389/fnins.2016.00241 pmid:27445650 pmcid:PMC4925698 fatcat:x5jt5klkxfctbmjnlvyfmpel64

Dual stochasticity of neurons and synapses provides a biologically plausible learning [article]

Jun-nosuke Teramae
2019 bioRxiv   pre-print
framework and provide an efficient and biologically-plausible learning algorithm that consistently explains various experimental findings of the brain, which includes statistics of cortical circuit and the power-low  ...  scaling of population activity of cortical neurons.  ...  circuit [31] , and response properties of cortical neurons including the nearly optimal power-low scaling of population activity [9] .  ... 
doi:10.1101/811646 fatcat:n2hhlu6xpbhl7enumprmkuky6u

Building fast Bayesian computing machines out of intentionally stochastic, digital parts [article]

Vikash Mansinghka, Eric Jonas
2014 arXiv   pre-print
compatible with the Poisson firing statistics of cortical neurons.  ...  We find that by connecting stochastic digital components according to simple mathematical rules, one can build massively parallel, low precision circuits that solve Bayesian inference problems and are  ...  acknowledge Tomaso Poggio, Thomas Knight, Gerald Sussman, Rakesh Kumar and Joshua Tenenbaum for numerous helpful discussions and comments on early drafts, and Tejas Kulkarni for contributions to the spiking  ... 
arXiv:1402.4914v1 fatcat:mnjmxywzyrgo5avrttcvsxosri

Brain-Inspired Hardware Solutions for Inference in Bayesian Networks

Leila Bagheriye, Johan Kwisthout
2021 Frontiers in Neuroscience  
These efforts lead to several implementations ranging from digital circuits, mixed-signal circuits, to analog circuits by leveraging new emerging nonvolatile devices.  ...  Hence, developing probabilistic spiking neurons with low area and power consumption is highly required.  ...  SNNs are implemented on low-power event-driven hardware, and the time and energy consumption are proportional to the number of spike events.  ... 
doi:10.3389/fnins.2021.728086 pmid:34924925 pmcid:PMC8677599 fatcat:tihogzl6tfbpjdybwpggllwd5u

Robustness of spiking Deep Belief Networks to noise and reduced bit precision of neuro-inspired hardware platforms

Evangelos Stromatias, Daniel Neil, Michael Pfeiffer, Francesco Galluppi, Steve B. Furber, Shih-Chii Liu
2015 Frontiers in Neuroscience  
The on-going work on design and construction of spike-based hardware platforms offers an alternative for running deep neural networks with significantly lower power consumption, but has to overcome hardware  ...  Furthermore, the network performance of spiking DBNs is characterized with regard to noise in the spiking input signal.  ...  Once trained, the single bit precision analog network can potentially be implemented on a digital neuron platform such as TrueNorth (Merolla et al., 2014b) , using the Gibbs sampling methods recently  ... 
doi:10.3389/fnins.2015.00222 pmid:26217169 pmcid:PMC4496577 fatcat:f3bjob43irhg5owjbmpix5rmam

Noise as a Resource for Computation and Learning in Networks of Spiking Neurons

Wolfgang Maass
2014 Proceedings of the IEEE  
I will also describe why these results are paving the way for a qualitative jump in the computational capability and learning performance of neuromorphic networks of spiking neurons with noise, and for  ...  Recent theoretical results have provided insight into how this can be achieved: how noise enables networks of spiking neurons to carry out probabilistic inference through sampling and also enables creative  ...  Rueckert for scientific advice and help with the figures. He would also like to thank three anonymous reviewers for helpful suggestions.  ... 
doi:10.1109/jproc.2014.2310593 fatcat:54mgt3scqje5flvjqnad45okfi

Nyquist interpolation improves neuron yield in multiunit recordings

Timothy J. Blanche, Nicholas V. Swindale
2006 Journal of Neuroscience Methods  
For most purposes, including spike sorting, sample rates below 25 kHz with bandlimited interpolation to 50 kHz were ideal, with negligible gains above this rate.  ...  Here we show that bandlimited interpolation with sample-and-hold delay correction reduces waveform variability, leading to improved reliability of threshold-based event detection and improved spike sorting  ...  (B) Uncentred spike amplitude clusters from another neuron (n = 1253). Underestimation of spike amplitudes produced the drift in cluster centres at low sample rates.  ... 
doi:10.1016/j.jneumeth.2005.12.031 pmid:16481043 fatcat:nnt65jbwirbrjnmxsxynr5jfzy

Solving Constraint Satisfaction Problems with Networks of Spiking Neurons

Zeno Jonke, Stefan Habenschuss, Wolfgang Maass
2016 Frontiers in Neuroscience  
machines) and Gibbs sampling.  ...  Such spike-based computations promise to be substantially more power-efficient than traditional clocked processing schemes.  ...  Our theoretical analysis implies that this effect enhances exploration in spike-based networks, compared with Boltzmann machines (Gibbs sampling).  ... 
doi:10.3389/fnins.2016.00118 pmid:27065785 pmcid:PMC4811945 fatcat:ecdskaa5f5h2hbjhiu3wr5rqrq

A theoretical basis for efficient computations with noisy spiking neurons [article]

Zeno Jonke, Stefan Habenschuss, Wolfgang Maass
2014 arXiv   pre-print
Furthermore, one can demonstrate for the Traveling Salesman Problem a surprising computational advantage of networks of spiking neurons compared with traditional artificial neural networks and Gibbs sampling  ...  Such spike-based computations promise to be substantially more power-efficient than traditional clocked processing schemes.  ...  Powerful computations with spiking neurons have previously been demonstrated in [9] .  ... 
arXiv:1412.5862v1 fatcat:qfg7wp4b75hllkoye2x45zt4ei

Neuromorphic Processing and Sensing: Evolutionary Progression of AI to Spiking [article]

Philippe Reiter, Geet Rose Jose, Spyridon Bizmpikis, Ionela-Ancuţa Cîrjilă
2020 arXiv   pre-print
Neuromorphic technologies based on Spiking Neural Network algorithms hold the promise to implement advanced artificial intelligence using a fraction of the computations and power requirements by modeling  ...  the functioning, and spiking, of the human brain.  ...  to also leverage low-power, low-latency and highly parallel SNN models, where appropriate.  ... 
arXiv:2007.05606v1 fatcat:mw7nczubnzao3l73kyibxyvjpy
« Previous Showing results 1 — 15 out of 654 results