Efe Soyman (7326125), David S. Vicario (7326128) · Figshare · 2019
<p>(A) Neural decoding accuracies were analyzed as the probability of correct decoding across time points along the stimulus duration and stimulus presentations. The black arrows schematically indicate the subsequent time point (left) and probability level (right) analyses. (B) Correct neural decoding probabilities at all selected time points increased with stimulus presentations. (C) Correct neural decoding latencies at all selected probability levels decreased with stimulus presentations. (D) The slopes of correct neural decoding probabilities were higher than zero at all time points for presentations 1–6 and at the 250 and 500-ms time points for presentations 6–25. (E) The slopes of correct neural decoding latencies were lower than zero at all probability levels for presentations 1–6 and at the 0.875 probability level for presentations 6–25. Shadings and error bars indicate SEMs. Asterisks denote significant differences from zero.</p>
Cited by 0Open access
Encoding And Decoding · 2007
timulus s appears and that response r is recorded, respectively. To study neural encoding and decoding, we need, in addition, two conditional probabilities: the probability of evoking a response r given that the stimulus s was presented, denoted by P[r|s]; and the probability that a given stimulus s was presented given that a particular response r was recorded, P[s|r]. The encoding map is characterized by the probabilities P[r|s] which reflect features of the encoding process as discussed in chapter 1. The decoding map is defined by the probabil- Draft: March 13, 1999 Theoretical Neuroscience 2 Neural Decoding ities, P[s|r]. When the respo
Cited by 0
John Hertz, Barry J. Richmond, B. G. Hertz et al. · 1991
Cited by 1
2018
Cited by 0
Islam Badreldin, Karim Oweiss · Encyclopedia of Computational Neuroscience · 2015
Cited by 0
Rezwan Firuzi, Ayub Bokani, Jahan Hassan et al. · 2025
Human civilization has witnessed transformative technological milestones, from ancient fire-lighting to the internet era. This chapter delves into the invasive brain–machine interface (BMI), a pioneering technology poised to be a defining chapter in our progress. Beyond aiding medical conditions, invasive BMI promises far-reaching impacts across diverse technologies and aspects of life. The exploration begins by unraveling the biological and engineering principles essential for BMI implementation. The chapter comprehensively analyzes potential applications, methodologies for detecting and decoding brain signals, and options for stimulating signals within the human brain. It concludes with a discussion on the multifaceted challenges and opportunities for the continued development of invasive BMI. This chapter not only provides a profound understanding of the foundational elements of invasive BMI but also serves as a guide through its applications, intricacies, and potential societal implications. Navigating neurobiology, engineering innovations, and the evolving landscape of human-AI symbiosis, the chapter sheds light on the promises and hurdles that define the future of invasive BMI.
Cited by 0
Behtash Babadi · Handbook of Neuroengineering · 2023
Cited by 1
Donald Geman, Peter Dayan · 2008
We propose adaptive testing as a general mechanism for extracting information about stimuli from spike trains. Each test or question corresponds to choosing a neuron and a time interval and checking for a given number of spikes. No assumptions are made about the distribution of spikes or any other aspect of neural encoding. The chosen questions are those which most reduce the uncertainty about the stimulus, as measured by entropy and estimated from stimulus-response data. Our experiments are based on accurate simulations of responses to pure tones in the auditory nerve and are meant to illustrate the ideas rather than investigate the auditory system. The results cohere nicely with well-understood encoding of amplitude and frequency in the auditory nerve, suggesting that adaptive testing might provide a powerful tool for investigating complex and poorly understood neural structures.
Cited by 0
Pascal Wallisch, Michael Lusignan, Marc Benayoun et al. · Elsevier eBooks · 2009
Cited by 0
Matthew R. Whiteway (11467528), Dan Biderman (8835644), Yoni Friedman (11467531) et al. · Figshare · 2021
<p><b>Top left</b>: frames from the original behavioral video. <b>Top center</b>: reconstructed frames from the PS-VAE. <b>Top right</b>: frames reconstructed by the neural activity. <b>Bottom left</b>: residual between the PS-VAE and neural reconstructions. <b>Bottom right</b>: PS-VAE latents (gray traces) and their predictions from neural activity (colored traces; colors are consistent with those in the main figures). Reconstructions are shown for several batches of test data, which were not used for the training or model selection of either the PS-VAE or the neural decoders. Each test batch is separated by black frames in the frame reconstructions, and breaks in the traces.</p> <p>(PNG)</p>
Cited by 0Open access
Keith A. Bush (5472590), Clinton D. Kilts (228932) · Figshare · 2022
Cited by 0Open access
Christian d&apos;Avignon, Donald Geman · 2003
We propose adaptive testing as a general mechanism for extracting information about stimuli from spike trains. Each test or question corresponds to choosing a neuron and a time interval and checking for a given number of spikes. No assumptions are made about the distribution of spikes or any other aspect of neural encoding. The chosen questions are those which most reduce the uncertainty about the stimulus, as measured by entropy and estimated from stimulus-response data. Our experiments are based on accurate simulations of responses to pure tones in the auditory nerve and are meant to illustrate the ideas rather than investigate the auditory system. The results cohere nicely with well-understood encoding of amplitude and frequency in the auditory nerve, suggesting that adaptive testing might provide a powerful tool for investigating complex and poorly understood neural structures.
Cited by 0
Ethan M. Meyers · Frontiers in Neuroinformatics · 2013
Population decoding is a powerful way to analyze neural data, however, currently only a small percentage of systems neuroscience researchers use this method. In order to increase the use of population decoding, we have created the Neural Decoding Toolbox (NDT) which is a Matlab package that makes it easy to apply population decoding analyses to neural activity. The design of the toolbox revolves around four abstract object classes which enables users to interchange particular modules in order to try different analyses while keeping the rest of the processing stream intact. The toolbox is capable of analyzing data from many different types of recording modalities, and we give examples of how it can be used to decode basic visual information from neural spiking activity and how it can be used to examine how invariant the activity of a neural population is to stimulus transformations. Overall this toolbox will make it much easier for neuroscientists to apply population decoding analyses to their data, which should help increase the pace of discovery in neuroscience.
Cited by 187Open access
Yi Yu, Shaomin Zhang, Huaijian Zhang et al. · Journal of Zhejiang University SCIENCE B · 2010
Brain-machine interface (BMI) has been developed due to its possibility to cure severe body paralysis. This technology has been used to realize the direct control of prosthetic devices, such as robot arms, computer cursors, and paralyzed muscles. A variety of neural decoding algorithms have been designed to explore relationships between neural activities and movements of the limbs. In this paper, two novel neural decoding methods based on probabilistic neural network (PNN) in rats were introduced, the PNN decoder and the modified PNN (MPNN) decoder. In the experiment, rats were trained to obtain water by pressing a lever over a pressure threshold. Microelectrode array was implanted in the motor cortex to record neural activity, and pressure was recorded by a pressure sensor synchronously. After training, the pressure values were estimated from the neural signals by PNN and MPNN decoders. Their performances were evaluated by a correlation coefficient (CC) and a mean square error (MSE). The results show that the MPNN decoder, with a CC of 0.8657 and an MSE of 0.2563, outperformed the traditionally-used Wiener filter (WF) and Kalman filter (KF) decoders. It was also observed that the discretization level did not affect the MPNN performance, indicating that the MPNN decoder can handle different tasks in BMI system, including the detection of movement states and estimation of continuous kinematic parameters.
Cited by 8Open access
Yaroslav O. Halchenko · 2009
Cited by 0
Sonia Todorova, Valérie Ventura · Neural Computation · 2017
Decoding in the context of brain-machine interface is a prediction problem, with the aim of retrieving the most accurate kinematic predictions attainable from the available neural signals. While selecting models that reduce the prediction error is done to various degrees, decoding has not received the attention that the fields of statistics and machine learning have lavished on the prediction problem in the past two decades. Here, we take a more systematic approach to the decoding prediction problem and search for risk-optimized reverse regression, optimal linear estimation (OLE), and Kalman filter models within a large model space composed of several nonlinear transformations of neural spike counts at multiple temporal lags. The reverse regression decoding framework is a standard prediction problem, where penalized methods such as ridge regression or Lasso are routinely used to find minimum risk models. We argue that minimum risk reverse regression is always more efficient than OLE and also happens to be 44% more efficient than a standard Kalman filter in a particular application of offline reconstruction of arm reaches of a rhesus macaque monkey. Yet model selection for tuning curves-based decoding models such as OLE and Kalman filtering is not a standard statistical prediction problem, and no efficient method exists to identify minimum risk models. We apply several methods to build low-risk models and show that in our application, a Kalman filter that includes multiple carefully chosen observation equations per neural unit is 67% more efficient than a standard Kalman filter, but with the drawback that finding such a model is computationally very costly.
Cited by 1Open access
Dado, T.M., Güçlütürk, Y., Ambrogioni, L. et al. · Data Archiving and Networked Services (DANS) · 2021
We show how generative modeling approximates the neural (face) manifold by obtaining state-of-the-art reconstructions of perceived face images from fMRI activations.
Cited by 0
Kang Li · OSF Preprints (OSF Preprints) · 2019
Repository for the paper: Kang Li and Susanne Ditlevsen. (2019). Neural decoding with visual attention using sequential Monte Carlo for leaky integrate-and-fire neurons
Cited by 0
John Platt, J. J. Hopfield · AIP conference proceedings · 1986
This paper develops a particular error correction code which can be effectively decoded by a relatively simple neural network. In high noise situations, this code is comparable to that used at present in deep space communications. The neural decoder has N! stable states with only N2 neurons, and can quickly extract information from analog noise. This example illustrates the effectiveness of neural networks in solving real problems when the problem can be cast in such a fashion that it fits gracefully on the network.
Cited by 7
Daniel Valencia, Amirhossein Alimohammad · Biomedical Engineering Letters · 2022
Cited by 9Open access