Filters








1,096 Hits in 9.2 sec

Fast maximum likelihood estimation using continuous-time neural point process models

Kyle Q. Lepage, Christopher J. MacDonald
2015 Journal of Computational Neuroscience  
By using continuous-time models of neural activity and the optimally efficient Gaussian quadrature, memory requirements and computation times are dramatically decreased in the commonly encountered situation  ...  A commonly employed statistical paradigm using discretetime point process models of neural activity involves the computation of a maximum-likelihood estimate.  ...  Kass for a discussion regarding the content of this paper and on the use of Gaussian quadrature in statistics, to Mikio Aoi for a useful comment regarding the scope of the paper, and to Sujith Vijayan  ... 
doi:10.1007/s10827-015-0551-y pmid:25788412 fatcat:tsj2hnvzezhw7gpi3barjh2gxm

Horseshoe Regularization for Machine Learning in Complex and Deep Models [article]

Anindya Bhadra, Jyotishka Datta, Yunfan Li, Nicholas G. Polson
2019 arXiv   pre-print
Specifically, we focus on methodological challenges in horseshoe regularization in nonlinear and non-Gaussian models; multivariate models; and deep neural networks.  ...  Most of the existing literature has focused on the linear Gaussian case; see Bhadra et al. (2019b) for a systematic survey.  ...  Acknowledgements We thank the AE and two anonymous referees for many helpful comments. Bhadra and Polson are supported by Grant No. DMS-1613063 by the US National Science Foundation.  ... 
arXiv:1904.10939v2 fatcat:zduypevpdfdrvglrqz4eim646q

Bayesian Clustering of Neural Activity with a Mixture of Dynamic Poisson Factor Analyzers [article]

Ganchao Wei, Ian H. Stevenson, Xiaojing Wang
2022 arXiv   pre-print
, and may, thus, be a useful tool for neural data analysis.  ...  To do the analysis of DPFA model, we propose a novel Markov chain Monte Carlo (MCMC) algorithm to efficiently sample its posterior distribution.  ...  The LDS model is built on the state-space model and assumes latent factors evolve with linear dynamics. On the other hand, GPFA models the latent vectors by non-parametric Gaussian processes.  ... 
arXiv:2205.10639v1 fatcat:7c7ta7sxzfg2hb7yldeltziq5q

Probabilistic Tensor Decomposition of Neural Population Spiking Activity

Hugo Soulat, Sepiedeh Keshavarzi, Troy W. Margrie, Maneesh Sahani
2021 Neural Information Processing Systems  
spike-count data.  ...  Here, we extend the Pólya-Gamma (PG) augmentation, previously used in sampling-based Bayesian inference, to implement scalable variational inference in non-conjugate spike-count models.  ...  Acknowledgments and Disclosure of Funding We would like to thank Joaquin Rapela for early contributions to this project and Marc Deisenroth and Céline Marié for helpful comments on the manuscript.  ... 
dblp:conf/nips/SoulatKMS21 fatcat:royylexisref3lt5bzudizetxa

Estimation of neuronal firing rate using Bayesian Adaptive Kernel Smoother (BAKS) [article]

Nur Ahmadi, Timothy G. Constandinou, Christos-Savvas Bouganis
2017 biorxiv/medrxiv   pre-print
We evaluate the performance of BAKS using synthetic spike train data generated by biologically plausible models: inhomogeneous Gamma (IG) and inhomogeneous inverse Gaussian (IIG).  ...  We also apply BAKS to real spike train data from non-human primate (NHP) motor and visual cortex. We benchmark the proposed method against the established and previously reported methods.  ...  We tune the parameter of BAKS using synthetic spike train data stochastically sampled from 3 rate functions (as representation of non-stationary underlying processes).  ... 
doi:10.1101/204818 fatcat:gltjrnyburhddgwjv6nfkwumdq

Estimation of neuronal firing rate using Bayesian Adaptive Kernel Smoother (BAKS)

Nur Ahmadi, Timothy G Constandinou, Christos-Savvas Bouganis
2018 PLoS ONE  
We evaluate the performance of BAKS using synthetic spike train data generated by biologically plausible models: inhomogeneous Gamma (IG) and inhomogeneous inverse Gaussian (IIG).  ...  We also apply BAKS to real spike train data from non-human primate (NHP) motor and visual cortex. We benchmark the proposed method against established and previously reported methods.  ...  We tune the parameter of BAKS using synthetic spike train data stochastically sampled from 3 rate functions (as representation of non-stationary underlying processes).  ... 
doi:10.1371/journal.pone.0206794 pmid:30462665 pmcid:PMC6248928 fatcat:tqsl4n6einayxmccjfiaubfyay

Convolutional spike-triggered covariance analysis for neural subunit models

Anqi Wu, Il Memming Park, Jonathan W. Pillow
2015 Neural Information Processing Systems  
Specifically, we show that a "convolutional" decomposition of a spike-triggered average (STA) and covariance (STC) matrix provides an asymptotically efficient estimator for class of quadratic subunit models  ...  Here we address this problem by providing a theoretical connection between spike-triggered covariance analysis and nonlinear subunit models.  ...  Simple LN models with Gaussian or Poisson noise can be fit very efficiently with spiketriggered-moment based estimators [6] [7] [8] , but there is no equivalent theory for LN-LN or subunit models.  ... 
dblp:conf/nips/WuPP15 fatcat:xjiqzg6tc5epdp5a4xviijttrq

Computing loss of efficiency in optimal Bayesian decoders given noisy or incomplete spike trains

Carl Smith, Liam Paninski
2013 Network  
Such models are sufficiently flexible that we may incorporate realistic stimulus encoding and spiking dynamics, but nonetheless permit exact computation via efficient hidden Markov model forward-backward  ...  In particular, decoders that model the probability distribution of spike-neuron assignments significantly outperform decoders that use only the most likely spike assignments, and are ignorant of the posterior  ...  We also gratefully acknowledge the use of the Hotfoot shared cluster computer at Columbia University.  ... 
doi:10.3109/0954898x.2013.789568 pmid:23742213 fatcat:ayrb2hhdnratbizqotlihlt5de

Prior-preconditioned conjugate gradient method for accelerated Gibbs sampling in "large n large p" Bayesian sparse regression [article]

Akihiko Nishimura, Marc A. Suchard
2022 arXiv   pre-print
We can then solve the linear system by the conjugate gradient (CG) algorithm through matrix-vector multiplications by Φ; this involves no explicit factorization or calculation of Φ itself.  ...  to compute and factorize.  ...  Outside the Gaussian process literature, Zhou & Guan (2019) use an iterative method to address the bottleneck of having to solve large linear systems when computing Bayes factors in a model selection problem  ... 
arXiv:1810.12437v6 fatcat:4cuwdp4sgrbo7gddihu4i2y3ey

Scalable Bayesian GPFA with automatic relevance determination and discrete noise models [article]

Kristopher T. Jensen, Ta-Chu Kao, Jasmine Talia Stone, Guillaume Hennequin
2021 bioRxiv   pre-print
Here, we bridge this gap by developing a fully Bayesian yet scalable version of Gaussian process factor analysis (bGPFA) which models neural data as arising from a set of inferred latent processes with  ...  To enable the analysis of continuous recordings without trial structure, we introduce a novel variational inference strategy that scales near-linearly in time and also allows for non-Gaussian noise models  ...  Acknowledgements We are grateful to O'Doherty et al. (2017) for making their data publicly available and to Marine Schimel and David Liu for insightful discussions.  ... 
doi:10.1101/2021.06.03.446788 fatcat:u4rjygyxafcppouycnmrynhzda

Segmenting sign language into motor primitives with Bayesian binning

Dominik Endres, Yaron Meirovitch, Tamar Flash, Martin A. Giese
2013 Frontiers in Computational Neuroscience  
For the example of sign language we investigate whether such segments can be identified by Bayesian binning (BB), using a Gaussian observation model whose mean has a polynomial time dependence.  ...  are also not adequately represented by low order polynomials and require higher order polynomials for a good approximation.  ...  Meir Etedgy for very helpful discussions and for their critical contributions to the data acquisition of the sign language data. We also thank the reviewers for their constructive comments.  ... 
doi:10.3389/fncom.2013.00068 pmid:23750135 pmcid:PMC3664315 fatcat:miv6izptlbdobf6t4dyyi2h6va

Space Alternating Variational Estimation Based Sparse Bayesian Learning for Complex-value Sparse Signal Recovery Using Adaptive Laplace Priors [article]

Zonglong Bai, Liming Shi, Jinwei Sun, Mads Græsbøll Christensen
2022 arXiv   pre-print
In experiments, the proposed algorithm is studied for complex Gaussian random dictionaries and different types of complex signals.  ...  However, most existing methods are based on the real-value signal model, with the complex-value signal model rarely considered.  ...  Acknowledge This work is supported by the Fundamental Research Funds for the Central Universities (No.2022MS077).  ... 
arXiv:2006.16720v3 fatcat:okx7qkj5bbdq7isfczch7d7jta

Seeing into Darkness: Scotopic Visual Recognition

Bo Chen, Pietro Perona
2017 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
For non-spiking hardwares, the number of spikes also translate to the number of floating point multiplications required for the layers above.  ...  We also define the first layer feature as the activity prior to non-linearity. 6 We use a Gamma prior because it is the conjugate prior of the Poisson likelihood. regime.  ... 
doi:10.1109/cvpr.2017.771 dblp:conf/cvpr/ChenP17 fatcat:qc4dgs3euva2jgpyhookrjzhfm

Neuromorphic Visual Scene Understanding with Resonator Networks [article]

Alpha Renner, Lazar Supic, Andreea Danielescu, Giacomo Indiveri, Bruno A. Olshausen, Yulia Sandamirskaya, Friedrich T. Sommer, E. Paxon Frady
2022 arXiv   pre-print
The spiking neuron model allows to map the resonator network onto efficient and low-power neuromorphic hardware.  ...  ; (3) the design of a multi-compartment spiking phasor neuron model for implementing complex-valued vector binding on neuromorphic hardware.  ...  We thank Intel Neuromorphic Computing Lab for providing access to the Loihi hardware and related software. We thank Elvin Hajizada for running CPU power measurements.  ... 
arXiv:2208.12880v1 fatcat:3nc46ya5cbdp5e6sxvy27plft4

The case of negative day-ahead electricity prices

Enzo Fanone, Andrea Gamba, Marcel Prokopczuk
2013 Energy Economics  
In this paper, we present a non-Gaussian process to model German intra-day electricity prices and propose an estimation procedure for this model.  ...  Most importantly, our model is able to generate extreme positive and negative spikes. A simulation study demonstrates the ability of our model to capture the characteristics of the data.  ...  Benth et al. (2007) and Klüppelberg et al. (2010) consider a continuoustime three-factor non-Gaussian OU process for modeling electricity spot prices.  ... 
doi:10.1016/j.eneco.2011.12.006 fatcat:bh7fqcevbfai5lhw74vlvilf3m
« Previous Showing results 1 — 15 out of 1,096 results