Barak A. Pearlmutter

Reverse-Mode AD in a Functional Framework: Lambda the Ultimate Backpropagator (2009)

Abstract. We discuss the implications of the desire to augment a functionalprogramming language with a derivative-taking operator using forward-mode automatic differentiation (AD). The primary...

Fiachra Matthews, Barak A. Pearlmutter, Tomas E. Ward, Christopher Soraghan, Charles Markham

We show how reverse-mode AD (automatic differentiation)—a generalized gradient-calculation operator—can be incorporated as a first-class function in a functional-programming language. An...

Interfaces (2008)

for Brain-Computer

Reverse-Mode AD in a Functional Framework: Lambda the Ultimate Backpropagator (2008)

doi:10.1006/nimg.2002.1320 Independent Components of Magnetoencephalography: Single-Trial Response Onset Times (2008)

We show that reverse-mode AD (Automatic Differentiation)—a generalized gradient-calculation operator—can be incorporated as a first-class function in an augmented lambda calculus, and therefore...

Akaysha C. Tang, Barak A. Pearlmutter, Natalie A. Malaszenko, Dan B. Phung

We recently demonstrated that second-order blind identification (SOBI), an independent component analysis (ICA) method, can separate the mixture of neuronal and noise signals in...

Brightness Illusions as Optimal Percepts (2008)

Tuning for Criticality: A New Hypothesis for Sleep (2008)

We show that Mach bands and a number of other low-level brightness illusions can be accounted for by assuming that the perceptual system performs simple Bayesian inference using a Gaussian image...

Barak A. Pearlmutter, Conor J. Houghton

We propose that the critical function of sleep [3] is to prevent uncontrolled neuronal feedback while allowing rapid responses and prolonged retention of short-term memories. The goal of learning is...

USE YOUR POWERS WISELY: RESOURCE ALLOCATION IN PARALLEL CHANNELS (2008)

Reverse-Mode AD in a Functional Framework: Lambda the Ultimate Backpropagator (2008)

This study evaluates various resource allocation strategies for simultaneous estimation of two independent signals from noisy observations. We focus on strategies that make use of the underlying...

Nesting forward-mode AD in a functional framework (2008)

We show that reverse-mode AD (Automatic Differentiation)—a generalized gradient-calculation operator—can be incorporated as a first-class function in an augmented lambda calculus, and therefore...

Perturbation Confusion and Referential Transparency Correct Functional Implementation of Forward-Mode AD (2008)

Abstract. We discuss the implications of the desire to augment a functionalprogramming language with a derivative-taking operator using forward-mode automatic differentiation (AD). The primary...

Contents 14 Sparsification for Monaural Source Separation........... 1 (2008)

Abstract. It is tempting to incorporate differentiation operators into functional-programming languages. Making them first-class citizens, however, is an enterprise fraught with danger. We discuss a...

London Milan Paris, Hiroki Asari, Rasmus K. Olsson, Barak A. Pearlmutter, Anthony M. Zador

Perturbation Confusion and Referential Transparency Correct Functional Implementation of Forward-Mode AD (2008)

LETTER Communicated by Michael Lewicki Blind Source Separation by Sparse Decomposition in a Signal Dictionary (2008)

Abstract. It is tempting to incorporate differentiation operators into functional-programming languages. Making them first-class citizens, however, is an enterprise fraught with danger. We discuss a...

USE YOUR POWERS WISELY: RESOURCE ALLOCATION IN PARALLEL CHANNELS (2008)

The blind source separation problem is to extract the underlying source signals from a set of linear mixtures, where the mixing matrix is unknown. This situation is common in acoustics, radio,...

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING (IN SUBMISSION) 1 Blind Source Separation for the Superconducting (2008)

This study evaluates various resource allocation strategies for simultaneous estimation of two independent signals from noisy observations. We focus on strategies that make use of the underlying...

Keeran Maharajh, Sung Chan Jun, Barak A. Pearlmutter, Robert H. Kraus, Petr L. Volegov

Abstract—Magnetic signals from brain responses are very weak compared to those from its surroundings. A system based on SQUIDs (Superconducting Quantum Interference Devices) was constructed to...

Nesting forward-mode AD in a functional framework (2008)

Neuronal Predictions of Sparse Linear Representations (2008)

Abstract. We discuss the implications of the desire to augment a functionalprogramming language with a derivative-taking operator using forward-mode automatic differentiation (AD). The primary...

Barak A. Pearlmutter, Hiroki Asari, Anthony M. Zador

A striking feature of many sensory processing problems is that there appear to be many more neurons engaged in the internal representations of the signal than in its transduction. For example, humans...

Differentiating Functions of the Jacobian with Respect to the Weights (2008)

Gary William Flake, Barak A. Pearlmutter

For many problems, the correct behavior of a model depends not only on its input-output mapping but also on properties of its Jacobian matrix, the matrix of partial derivatives of the model’s...

Abstract (2008)

Sung C. Jun, Barak A. Pearlmutter

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Behavioral/Systems/Cognitive Sparse Representations for the Cocktail Party Problem (2008)

Hiroki Asari, Barak A. Pearlmutter, Anthony M. Zador

A striking feature of many sensory processing problems is that there appear to be many more neurons engaged in the internal representations of the signal than in its transduction. For example, humans...

Tuning for Criticality: A New Hypothesis for Sleep (2008)

Barak A. Pearlmutter, Conor J. Houghton

We propose that the critical function of sleep [3] is to prevent uncontrolled neuronal feedback while allowing rapid responses and prolonged retention of short-term memories. The goal of learning is...

Reverse-Mode AD in a Functional Framework: Lambda the Ultimate Backpropagator (2008)

Filtered Gaussian Processes for Learning with Large Data-Sets (2008)

We show that reverse-mode AD (Automatic Differentiation)—a generalized gradient-calculation operator—can be incorporated as a first-class function in an augmented lambda calculus, and therefore...

Jian Qing Shi, Roderick Murray-smith, D. Mike Titterington, Barak A. Pearlmutter

Abstract. Kernel-based non-parametric models have been applied widely over recent years. However, the associated computational complexity imposes limitations on the applicability of those methods to...

To appear in POPL 2007 First-Class Nonstandard Interpretations by Opening Closures (2008)

Algorithmic Differentiation, Functional Programming, and Iterate-to-Fixedpoint (2008)

Abstract We motivate and discuss a novel functional programming constructthat allows convenient modular run-time nonstandard interpretation via reflection on closure environments. This map-closure...

Discovering Convolutive Speech Phones using Sparseness and Non-Negativity Constraints (2008)

Algorithmic differentiation (AD) transforms straight line numeric code so that it calculates the derivative of the function originally calculated (Corliss et al., 2001). There are two varieties of...

Paul D. O’grady, Barak A. Pearlmutter

Abstract. Discovering a representation that allows auditory data to be parsimoniously represented is useful for many machine learning and signal processing tasks. Such a representation can be...

To appear in POPL 2007 Lazy Multivariate Higher-Order Forward-Mode AD (2008)

LETTER Communicated by Emilio Salinas Optimal Coding Predicts Attentional Modulation of Activity in Neural Systems (2008)

Abstract A method is presented for computing all higher-order partialderivatives of a multivariate function R n! R. This method works by evaluating the function under a nonstandard interpretation,...

Barak A. Pearlmutter, Lucas C. Parra

Neuronal activity in response to a fixed stimulus has been shown to change as a function of attentional state, implying that the neural code also changes with attention. We propose an...

Abstract (2008)

In the square linear blind source separation problem, one must nd a linear unmixing operator which can detangle the result xi(t) of mixing n unknown independent sources si(t) through an unknown n n...

Abstract (2008)

Sung C. Jun, Barak A. Pearlmutter

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Using Polyvariant Union-Free Flow Analysis to Compile a Higher-Order Functional-Programming Language with a First-Class Derivative Operator to Efficient Fortran-like Code (2008)

Siskind, Jeffrey M., Pearlmutter, Barak A.

We exhibit an aggressive optimizing compiler for a functionalprogramming language which includes a first-class forward automatic differentiation (AD) operator. The compiler's performance is...

Putting the Automatic Back into AD: Part I, What's Wrong (CVS: 1.1) (2008)

Siskind, Jeffrey M., Pearlmutter, Barak A.

Current implementations of automatic differentiation are far from automatic. We survey the difficulties encountered when applying four existing AD systems, ADIFOR, TAPENADE, ADIC, and FADBAD++, to...

Putting the Automatic Back into AD: Part II, Dynamic, Automatic, Nestable, and Fast (CVS: 1.1) (2008)

Pearlmutter, Barak A., Siskind, Jeffrey M.

This paper discusses a new AD system that correctly and automatically accepts nested and dynamic use of the AD operators, without any manual intervention. The system is based on a new formulation of...

Nesting Forward-Mode AD in a Functional Framework (2008)

Siskind, Jeffrey M., Pearlmutter, Barak A.

We discuss the augmentation of a functional-programming language with a derivative- taking operator implemented with forward-mode automatic differentiation (AD). The primary technical difficulty in...

Hemodynamics for brain-computer interfaces: optical correlates of control signals (2008)

Matthews, Fiachra, Pearlmutter, Barak A., Ward, Tomas E., Soraghan, Christopher, Markham, Charles

This article brings together the various elements that constitute the signal processing challenges presented by a hemodynamics-driven functional near-infrared spectroscopy (fNIRS) based...

The LOST Algorithm: Finding Lines and Separating Speech Mixtures (2008)

Paul D. O'Grady, Barak A. Pearlmutter

Robust clustering of data into linear subspaces is a frequently encountered problem. Here, we treat clustering of one-dimensional subspaces that cross the origin. This problem arises in blind source...

Isolating endogenous visuo-spatial attentional effects using the novel visual-evoked spread spectrum analysis (VESPA) technique (2007)

Lalor, Edmund C., Kelly, Simon P., Pearlmutter, Barak A., Reilly, Richard B., Foxe, John J.

In natural visual environments, we use attention to select between relevant and irrelevant stimuli that are presented simultaneously. Our attention to objects in our visual field is largely...

Barak A. Pearlmutter (2007)

Bap Cs Unm, Michael Zibulevsky, Barak A. Pearlmutter

In studies of repetitive responses, such as evoked responses in electro/magnetoencephalography, a signal is often of the same order of magnitude or even weaker than the noise. In order to recover the...

Blind Source Separation of Neuromagnetic Responses (2007)

Akaysha C. Tang, Barak A. Pearlmutter, Michael Zibulevsky

Magnetoencephalography (MEG) is a functional brain imaging technique with millisecond temporal resolution and millimeter spatial resolution. The high temporal resolution of MEG compared to fMRI and...

Brightness Illusions as Optimal Percepts Methods (2007)

Barak A. Pearlmutter, Santiago Jaramillo

We show that several brightness illusions can be largely accounted for by assuming that the perceptual system performs simple Bayesian inference using a Gaussian image prior with noisy retinal...

Subject-Independent Magnetoencephalographic Source Localization by a Multilayer Perceptron Sung C. Jun (2007)

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING (IN SUBMISSION) Blind Source Separation for the Superconducting (2007)

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Barak A. Pearlmutter, Robert H. Kraus, Petr L. Volegov

Abstract--Magnetic signals from brain responses are very weak compared to those from its surroundings. A system based on SQUIDs (Superconducting Quantum Interference Devices) was constructed to...

Equivalence Proofs for Multi-Layer Perceptron Classifiers and the Bayesian Discriminant Function (2007)

This paper presents a number of proofs that equate the outputs of a Multi-Layer Perceptron (MLP) classifier and the optimal Bayesian discriminant function for asymptotically large sets of...

y (2007)

A Normative Model of Attention: Modulation of Neuronal Response (2007)

Algorithmic differentiation (AD) transforms straight line numeric code so that it calculates the derivative of the function originally calculated (Corliss et al., 2001). There are two varieties of...

Transformations of Gaussian Process Priors (2007)

We introduce a normative model of top-down attention in which an attentional signal (whose origin is outside the scope of the model) breaks a symmetry implicit in the standard information-theoretic...

Roderick Murray-smith, Barak A. Pearlmutter

Gaussian processes-prior systems generally consist of noisy measurements of samples of the putatively Gaussian process of interest, where the samples serve to constrain the posterior estimate. Here...

Optimal Coding Predicts Attentional Modulation of Activity in Neural Systems (2007)

Jaramillo, Santiago, Pearlmutter, Barak A.

Neuronal activity in response to a fixed stimulus has been shown to change as a function of attentional state, implying that the neural code also changes with attention. We propose an...

Illusory Percepts from Auditory Adaptation (2007)

Parra, Lucas C., Pearlmutter, Barak A.

Phenomena resembling tinnitus and Zwicker phantom tone are seen to result from an auditory gain adaptation mechanism that attempts to make full use of a fixed-capacity channel. In the case of...

Lazy multivariate higher-order forward-mode AD (2007)

Discovering Convolutive Speech Phones using Sparseness and Non-Negativity Constraints (2007)

A method is presented for computing all higher-order partial derivatives of a multivariate function R n → R. This method works by evaluating the function under a nonstandard interpretation, lifting...

O'Grady, Paul D., Pearlmutter, Barak A.

Discovering a representation that allows auditory data to be parsimoniously represented is useful for many machine learning and signal processing tasks. Such a representation can be constructed by...

Use your powers wisely: resource allocation in parallel channels (2006)

Pearlmutter, Barak A., Jaramillo, Santiago

This study evaluates various resource allocation strategies for simultaneous estimation of two independent signals from noisy observations. We focus on strategies that make use of the underlying...

Convolutive non-negative matrix factorisation with a sparseness constraint (2006)

Pearlmutter, Barak A., O'Grady, Paul D.

Discovering a representation which allows auditory data to be parsimoniously represented is useful for many machine learning and signal processing tasks. Such a representation can be constructed by...

Linear program differentiation for single-channel speech separation (2006)

Pearlmutter, Barak A., Olsson, Rasmus K.

Many apparently difficult problems can be solved by reduction to linear programming. Such problems are often subproblems within larger systems. When gradient optimisation of the entire larger system...

A Dual-Channel Optical Brain-Computer Interface In A Gaming Environment (2006)

Soraghan , Christopher J, Matthews, Fiachra, Kelly, Dan, Markham, Charles, Pearlmutter, Barak A., O'Neill, Ray

This paper explores the viability of using a novel optical Brain-Computer Interface within a gaming environment. We describe a system that incorporates a 3D gaming engine and an optical BCI. This...

Brightness Illusions as Optimal Percepts (2006)

Sparse representations for the cocktail party problem (2006)

We show that Mach bands and a number of other low-level brightness illusions can be accounted for by assuming that the perceptual system performs simple Bayesian inference using a Gaussian image...

Hiroki Asari, Barak A. Pearlmutter, Anthony M. Zador

Bounds on Query Convergence (2005)

Pearlmutter, Barak A.

The problem of finding an optimum using noisy evaluations of a smooth cost function arises in many contexts, including economics, business, medicine, experiment design, and foraging theory. We derive...

Multimodal Integration: fMRI, MRI, EEG, MEG (2005)

Halchenko, Yaroslav O., Hanson, Stephen Jose, Pearlmutter, Barak A.

This chapter provides a comprehensive survey of the motivations, assumptions and pitfalls associated with combining signals such as fMRI with EEG or MEG. Our initial focus in the chapter concerns...

Fast Robust Subject-Independent Magnetoencephalographic Source Localization Using an Artificial Neural Network (2005)

Jun, Sung Chan, Pearlmutter, Barak A.

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Transformations of Gaussian Process Priors (2005)

Murray-Smith, Roderick, Pearlmutter, Barak A.

Gaussian process prior systems generally consist of noisy measurements of samples of the putatively Gaussian process of interest, where the samples serve to constrain the posterior estimate. Here we...

Linear Program Differentiation For Single-CHannel Speech Separation (2005)

Pearlmutter, Barak A., Olsson, Rasmus K.

Many apparently difficult problems can be solved by reduction to linear programming. Such problems are often subproblems within larger systems. When gradient optimisation of the entire larger system...

Transformations of Gaussian Process priors (2005)

Pearlmutter, Barak A., Murray-Smith, Roderick

Abstract. Gaussian process prior systems generally consist of noisy measurements of samples of the putatively Gaussian process of interest, where the samples serve to constrain the posterior...

Multimodal Integration: fMRI, MRI, EEG, MEG (2005)

Halchenko, Yaroslav O., Hanson, Stephen Jose, Pearlmutter, Barak A.

This chapter provides a comprehensive survey of the motivations, assumptions and pitfalls associated with combining signals such as fMRI with EEG or MEG. Our initial focus in the chapter concerns...

Fast Robust Subject-Independent Magnetoencephalographic Source Localization Using an Artificial Neural Network (2005)

Jun, Sung Chan, Pearlmutter, Barak A.

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Transformations of Gaussian Process Priors (2005)

Murray-Smith, Roderick, Pearlmutter, Barak A.

Gaussian process prior systems generally consist of noisy measurements of samples of the putatively Gaussian process of interest, where the samples serve to constrain the posterior estimate. Here we...

Linear Program Differentiation For Single-CHannel Speech Separation (2005)

Pearlmutter, Barak A., Olsson, Rasmus K.

Many apparently difficult problems can be solved by reduction to linear programming. Such problems are often subproblems within larger systems. When gradient optimisation of the entire larger system...

Perturbation Confusion and Referential Transparency: Correct Functional Implementation of Forward-Mode AD (2005)

Siskind, Jerey Mark, Pearlmutter, Barak A.

It is tempting to incorporate dierentiation operators into functional-programming languages. Making them rst-class citizens, however, is an enterprise fraught with danger. We discuss a potential...

Fast robust subject-independent magnetoencephalographic source localization using an artificial neural network (2005)

Sung Chan Jun, Barak A. Pearlmutter

Abstract: We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP)...

Fast robust subject-independent magnetoencephalographic source localization using an artificial neural network (2005)

Fast robust subject-independent magnetoencephalographic source localization using an artificial neural network (2005)

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Barak A. Pearlmutter, Guido Nolte

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Survey of sparse and non-sparse methods in source separation (2005)

Paul D. O’grady, Barak A. Pearlmutter, Scott T. Rickard

ABSTRACT: Source separation arises in a variety of signal processing applications, ranging from speech processing to medical image analysis. The separation of a superposition of multiple signals is...

Soft-LOST: EM on a mixture of oriented lines (2004)

Paul D. O'grady, Barak A Pearlmutter

Abstract. Robust clustering of data into overlapping linear subspaces is a common problem. Here we consider one-dimensional subspaces that cross the origin. This problem arises in blind source...

Soft-LOST: EM on a mixture of oriented lines (2004)

Paul D. O’grady, Barak A. Pearlmutter

Abstract. Robust clustering of data into overlapping linear subspaces is a common problem. Here we consider one-dimensional subspaces that cross the origin. This problem arises in blind source...

Monaural source separation using spectral cues (2004)

Barak A. Pearlmutter, Anthony M. Zador

Abstract. The acoustic environment poses at least two important challenges. First, animals must localise sound sources using a variety of binaural and monaural cues; and second they must separate...

Soft-LOST: EM on a mixture of oriented lines (2004)

Paul D. O’grady, Barak A. Pearlmutter

Abstract. Robust clustering of data into overlapping linear subspaces is a common problem. Here we consider one-dimensional subspaces that cross the origin. This problem arises in blind source...

MEG source localization using an MLP with a distributed output representation (2003)

Jun, Sung Chan, Pearlmutter, Barak A., Nolte, Guido

We present a system that takes realistic magnetoencephalographic (MEG) signals and localizes a single dipole to reasonable accuracy in real time. At its heart is a multilayer perceptron (MLP) which...

A Normative Model of Attention: Modulation of Neural Response (2003)

Santiago Jaramillo Barak, Barak A. Pearlmutter

When a sensory stimulus is encoded in a lossy fashion for efficient transmission, there are necessarily tradeoffs between the represented fidelity of various aspects of the input pattern. In the...

MEG Source Localization Using an MLP with a Distributed Output Representation (2003)

Transformations of Gaussian process priors (2003)

We present a system that takes realistic magnetoencephalographic (MEG) signals and localizes a single dipole to reasonable accuracy in real time. At its heart is a multilayer perceptron (MLP) which...

Roderick Murray-smith, Barak A. Pearlmutter

Abstract. Gaussian process prior systems generally consist of noisy measurements of samples of the putatively Gaussian process of interest, where the samples serve to constrain the posterior...

MEG source localization using a MLP with a distributed output representation (2003)

Subject-Independent Magnetoencephalographic Source Localization by a Multilayer Perceptron (2003)

Abstract—We present a system that takes realistic magnetoencephalographic (MEG) signals and localizes a single dipole to reasonable accuracy in real time. At its heart is a multilayer perceptron...

Sung C. Jun, Barak A. Pearlmutter

We describe a system that localizes a single dipole to reasonable accuracy from noisy magnetoencephalographic (MEG) measurements in real time. At its core is a multilayer perceptron (MLP) trained to...

Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise (2002)

Fast robust MEG source localization using MLPs (2002)

Abstract. Iterative gradient methods like Levenberg-Marquardt (LM) are in widespread use for source localization from electroencephalographic (EEG) and magnetoencephalographic (MEG) signals....

Independent components of magnetoencephalography: localization (2002)

Source localization from MEG data in real time requires algorithms which are robust, fully automatic, and very fast. We present two neural network systems which are able to localize a single dipole...

Akaysha C. Tang, Barak A. Pearlmutter, Dan B. Phung, Scott A. Carter

Blind source separation (BSS) decomposes multidimensional magnetoencephalography (MEG) data into a set of components. While many components reflect the activity from known and unknown noise sources,...

Independent Components of Magnetoencephalography: Localization (2002)

Akaysha C. Tang, Barak A. Pearlmutter, Natalie A. Malaszenko, Dan B. Phung, Bethany C. Reeb

We applied second-order blind identification (SOBI), an independent component...

Independent components of magnetoencephalography: localization (2002)

Akaysha C. Tang, Barak A. Pearlmutter

Independent component analysis (ICA) is a class of decomposition methods that separate sources from mixtures of signals. In this chapter, we used second order blind identification (SOBI), one of the...

Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise (2002)

Independent components of magnetoencephalography: localization (2002)

Abstract. Iterative gradient methods like Levenberg-Marquardt (LM) are in widespread use for source localization from electroencephalographic (EEG) and magnetoencephalographic (MEG) signals....

Akaysha C. Tang, Barak A. Pearlmutter

Independent component analysis (ICA) is a class of decomposition methods that separate sources from mixtures of signals. In this chapter, we used second order blind identification (SOBI), one of the...

Independent components of magnetoencephalography: localization (2002)

Akaysha C. Tang, Barak A. Pearlmutter

Independent component analysis (ICA) is a class of decomposition methods that separate sources from mixtures of signals. In this chapter, we used second order blind identification (SOBI), one of the...

Independent components of magnetoencephalography: localization (2002)

Akaysha C. Tang, Barak A. Pearlmutter, Natalie A. Malaszenko, Dan B. Phung, Bethany C. Reeb

We applied second-order blind identification (SOBI), an independent component analysis (ICA) method, to MEG data collected during cognitive tasks. We explored SOBI’s ability to help isolate...

Fast robust MEG source localization using MLPs (2002)

Fast robust MEG source localization using MLPs (2002)

Source localization from MEG data in real time requires algorithms which are robust, fully automatic, and very fast. We present two neural network systems which are able to localize a single dipole...

Jun, Sung Chan, Pearlmutter, Barak A., Nolte, Guido

Source localization from MEG data in real time requires algorithms which are robust, fully automatic, and very fast. We present two neural network systems which are able to localize a single dipole...

Blind Source Separation by Sparse Decomposition in a Signal Dictionary (2001)

Blind Separation Of Sources With Sparse Representations In A Signal Dictionary (2001)

The blind source separation problem is to extract the underlying source signals from a set of linear mixtures, where the mixing matrix is unknown. This situation is common, in acoustics, radio,...

Blind source separation by sparse decomposition in a signal dictionary (2001)

The blind source separation problem is to extract the underlying source signals from a set of linear mixtures, where the mixing matrix is unknown. We consider a two-stage separation process. First, a...

Second Order Blind Source Separation By Recursive Splitting Of Signal Subspaces (2000)

1 Introduction In blind source separation an N-channel sensor signal x(t) arises from M unknown scalar source signals si(t), linearly mixed together by an unknown N \Theta M matrix A, and possibly...

Michael Zibulevsky Barak, Michael Zibulevsky, Barak A. Pearlmutter

We present an approach to blind source separation based on delayed correlations. This method recursively splits separation space into subspaces spanned by groups of sources. The inner loop consists...

Independent Components of Magnetoencephalography, Part I: Localization (2000)

Akaysha C. Tang, Barak A. Pearlmutter, Dan B. Phung, Scott A. Carter

Blind source separation (BSS) decomposes a multidimensional time series into a set of components, each with a one-dimensional time course and a xed spatial distribution. For EEG and MEG, the former...

An MEG Study of Response Latency and Variability in the Human Visual System During a Visual-Motor Integration Task (2000)

Akaysha C. Tang, Barak A. Pearlmutter, Tim A. Hely, Michael Zibulevsky, Michael P. Weisend

Human reaction times during sensory-motor tasks vary considerably. To begin to understand how this variability arises, we examined neuronal populational response time variability at early versus late...

Localization Of Independent Components From Magnetoencephalography (2000)

Akaysha C. Tang, Dan Phung, Barak A. Pearlmutter, Robert Christner

Blind source separation (BSS) decomposes a multidimensional time series into a set of sources, each with a one-dimensional time course and a xed spatial distribution. For EEG and MEG, the former...

Blind Source Separation by Sparse Decomposition in a Signal Dictionary (2000)

Second Order Blind Source Separation (2000)

The blind source separation problem is to extract the underlying source signals from a set of linear mixtures, where the mixing matrix is unknown. This situation is common, in acoustics, radio,...

Recursive Splitting Of, Michael Zibulevsky, Barak A. Pearlmutter

based on delayed correlations. This method recursively splits separation space into subspaces spanned by groups of sources. The inner loop consists of repeated application of a standard eigenvalue...

Differentiating Functions of the Jacobian with Respect to the Weights (1999)

Gary William Flake, Barak A. Pearlmutter

For many problems, the correct behavior of a model depends not only on its input-output mapping but also on properties of its Jacobian matrix, the matrix of partial derivatives of the model's...

Differentiating Functions of the Jacobian with Respect to the Weights (1999)

Gary William Flake, Barak A. Pearlmutter

For many problems, the correct behavior of a model depends not only on its input-output mapping but also on properties of its Jacobian matrix, the matrix of partial derivatives of the model's...

Blind Source Separation by Sparse Decomposition (1999)

Learning State Space Trajectories in Recurrent Neural Networks: A Preliminary Report. (1998)

The blind source separation problem is to extract the underlying source signals from a set of their linear mixtures, where the mixing matrix is unknown. This situation is common, eg in acoustics,...

Pearlmutter, Barak A.

We describe a procedure for finding learning state space trajectories in recurrent neural networks. Keywords: Connectionism; Learning algorithm; Trajectories following; Minimizing functionals. (JES)

Results of the Abbadingo One DFA Learning Competition and a New Evidence Driven State Merging Algorithm (1998)

Kevin J. Lang, Barak A. Pearlmutter, Rodney A. Price

This paper first describes the structure and results of the Abbadingo One DFA Learning Competition. The competition was designed to encourage work on algorithms that scale well---both to larger DFAs...

Results of the Abbadingo One DFA Learning Competition and a New Evidence Driven State Merging Algorithm (1998)

Kevin J. Lang, Barak A. Pearlmutter, Rodney Price

This paper first describes the structure and results of the Abbadingo One DFA Learning Competition. The competition was designed to encourage work on algorithms that scale well -- both to larger DFAs...

Results of the abbadingo one DFA learning competition and a new evidence-driven state merging algorithm (1998)

Kevin J. Lang, Barak A. Pearlmutter, Rodney A. Price

Abstract. This paper first describes the structure and results of the Abbadingo One DFA Learning Competition. The competition was designed to encourage work on algorithms that scale well—both to...

Maximum likelihood blind source separation: A context-sensitive generalization of ICA (1997)

Barak A. Pearlmutter, Lucas C. Parra

In the square linear blind source separation problem, one must find a linear unmixing operator which can detangle the result x i (t) of mixing n unknown independent sources s i (t) through an unknown...

Doing the twist: diagonal meshes are isomorphic to twisted toroidal Meshes (1996)

Pearlmutter, Barak A.

We show that a k x n diagonal mesh is isomorphic to a n+k/2 x n+k/2 - nk/2 twisted toroidal mesh, i.e., a network similar to a standard n+k/2 x n-k/2 toroidal mesh, but with opposite handed twists of...

VC Dimension of an Integrate-and-Fire Neuron Model (1996)

Anthony Zador, Barak A. Pearlmutter

We compute the VC dimension of a leaky integrate-and-fire neuron model. The VC dimension quantifies the ability of a function class to partition an input pattern space, and can be considered a...

Doing the Twist: Diagonal Meshes are Isomorphic to Twisted Toroidal Meshes (1996)

VC Dimension of an Integrate-and-Fire Neuron Model (1996)

We show that a k \Theta n diagonal mesh is isomorphic to a n+k 2 \Theta n+k 2 \Gamma n\Gammak 2 \Theta n\Gammak 2 twisted toroidal mesh, i.e. a network similar to a standard n+k 2 \Theta n+k 2...

Anthony Zador, Barak A. Pearlmutter

We find the VC dimension of a leaky integrate-andfire neuron model. The VC dimension quantifies the ability of a function class to partition an input pattern space, and can be considered a measure of...

A context-sensitive generalization of ICA (1996)

Barak A. Pearlmutter, Lucas Parra

— Source separation arises in a surprising number of signal processing applications, from speech recognition to EEG analysis. In the square linear blind source separation problem without time...

Time-Skew Hebb Rule in a Nonisopotential Neuron (1995)

Gradient Calculations for Dynamic Recurrent Neural Networks: A Survey (1995)

In an isopotential neuron with rapid response, it has been shown that the receptive fields formed by Hebbian synaptic modulation depend on the principal eigenspace of Q(0), the input autocorrelation...

Relating Egomotion and Image Evolution (1995)

We survey learning algorithms for recurrent neural networks with hidden units, and put the various techniques into a common framework. We discuss fixedpoint learning algorithms, namely recurrent...

Barak A. Pearlmutter, Leonid Gurvits

By considering the dynamics of the apparent motion of a stationary object relative to a moving observer, we construct a partial differential equation that relates the changes in an image to the...

Fast exact multiplication by the Hessian (1994)

Fast Exact Multiplication by the Hessian (1994)

Just storing the Hessian H (the matrix of second derivatives @

Playing the Matching-Shoulders Lob-Pass Game with Logarithmic Regret (1994)

Just storing the Hessian H (the matrix of second derivatives ¶ 2 E=¶w i ¶w j of the error E with respect to each pair of weights) of a large neural network is difficult. Since a common use of a...

Joe Kilian, Kevin J. Lang, Barak A. Pearlmutter

The best previous algorithm for the matching shoulders lob-pass game, arthur (Abe and Takeuchi 1993), suffered O(t 1=2 ) regret. We prove that this is the best possible performance for any algorithm...

Fast exact multiplication by the Hessian (1994)

Efficient Computation of Sparse Elements of the Inverse of a Sparse Near-Tridiagonal Matrix with Application to the Nerve Equation (1993)

To appear in Neural Computation Just Hessian storing the (the matrix of second derivatives ∂2E¡∂wi∂wj of the error E with respect to each pair of weights) of a large neural network is...

Anthony M. Zador, Barak A. Pearlmutter

Standard algorithms for computing the inverse of a tridiagonal matrix (or more generally, any Hines matrix) compute the entire inverse, which is not sparse. For some problems, only the elements of...

Dynamic Recurrent Neural Networks (1990)