VLSI neuroprocessors for video motion detection.

The system design of a locally connected competitive neural network for video motion detection is presented. The motion information from a sequence of image data can be determined through a two-dimensional multiprocessor array in which each processing element consists of an analog neuroprocessor. Massively parallel neurocomputing is done by compact and efficient neuroprocessors.

Application of a maximum likelihood estimator in an experimental study in ultrasonic diffraction tomography.

A recently proposed approach to the inverse problem of detecting the presence and estimating the location of a known object from data collected in a set of diffraction tomographic experiments is evaluated. Experimental data are used to validate of the filtered backpropagation algorithms used, and their robustness to modeling errors and to severe limitations in the angular coverage of the tomographic data is demonstrated. A potential application to medical imaging of soft tissue is illustrated.

A VLSI neural processor for image data compression using self-organization networks.

An adaptive electronic neural network processor has been developed for high-speed image compression based on a frequency-sensitive self-organization algorithm. The performance of this self-organization network and that of a conventional algorithm for vector quantization are compared. The proposed method is quite efficient and can achieve near-optimal results.

A neural detector for seismic reflectivity sequences.

A commonly used routine in seismic signal processing is deconvolution, which comprises two operations: reflectivity detection and magnitude estimation. Existing statistical detectors are computationally expensive. In the paper, a Hopfield neural network is constructed to perform the reflectivity detection operation.

Adaptive fuzzy systems for backing up a truck-and-trailer.

Fuzzy control systems and neural-network control systems for backing up a simulated truck, and truck-and-trailer, to a loading dock in a parking lot are presented. The supervised backpropagation learning algorithm trained the neural network systems. The robustness of the neural systems was tested by removing random subsets of training data in learning sequences.

Learning and convergence analysis of neural-type structured networks.

A class of feedforward neural networks, structured networks, has recently been introduced as a method for solving matrix algebra problems in an inherently parallel formulation. A convergence analysis for the training of structured networks is presented. Since the learning techniques used in structured networks are also employed in the training of neural networks, the issue of convergence is discussed not only from a numerical algebra perspective but also as a means of deriving insight into connectionist learning.

Neuromagnetic localization using magnetic resonance images.

Ionic flow associated with neural activation of the brain produces a magnetic field, called the neuromagnetic field, that can be measured outside the head using a highly sensitive superconducting quantum interference device (SQUID)-based neuromagnetometer. Under certain conditions, the sources producing the neuromagnetic field can be localized from a sampling of the neuromagnetic field. Neuromagnetic measurements alone, however, do not contain sufficient information to visualize brain structure.

Correction of T2 distortion in multi-excitation RARE sequence.

Correction schemes have been implemented to correct for T2 distortions in a multiexcitation RARE (rapid acquisition with relaxation enhancement) sequence where data from multiple echoes and multiple excitations are combined. Computer simulation studies and human imaging studies have been conducted to develop and test the correction procedures. A direct method and an iterative technique have been investigated.

Stochastic competitive learning.

Competitive learning systems are examined as stochastic dynamical systems. This includes continuous and discrete formulations of unsupervised, supervised, and differential competitive learning systems. These systems estimate an unknown probability density function from random pattern samples and behave as adaptive vector quantizers.

Differential competitive learning for centroid estimation and phoneme recognition.

A comparison is made of a differential-competitive-learning (DCL) system with two supervised competitive-learning (SCL) systems for centroid estimation and for phoneme recognition. DCL provides a form of unsupervised adaptive vector quantization. Standard stochastic competitive-learning systems learn only if neurons win a competition for activation induced by randomly sampled patterns.

Cumulant-based parameter estimation using structured networks.

A two-level three-layer structured network is developed to estimate the moving-average model parameters based on second-order and third-order cumulant matching. The structured network is a multilayer feedforward network composed of linear summers in which the weights of these summers have a clear physical meaning. The first level is composed of random access memory units, which are used to control the connectivities of the second-level summers.

Derivation and analysis of a filtered backprojection algorithm for cone beam projection data.

The authors address the problem of three-dimensional image reconstruction from cone beam projections. Modifying a result due to A.A.

Computation of 3-D velocity fields from 3-D cine CT images of a human heart.

A method of computing the three-dimensional (3-D) velocity field from 3-D cine computer tomographs (CTs) of a beating heart is proposed. Using continuum theory, the authors develop two constraints on the 3-D velocity field generated by a beating heart. With these constraints, the computation of the 3-D velocity field is formulated as an optimization problem and a solution to the optimization problem is developed using the Euler-Lagrange method.

Unsupervised learning in noise.

A new hybrid learning law, the differential competitive law, which uses the neuronal signal velocity as a local unsupervised reinforcement mechanism, is introduced, and its coding and stability behavior in feedforward and feedback networks is examined. This analysis is facilitated by the recent Gluck-Parker pulse-coding interpretation of signal functions in differential Hebbian learning systems. The second-order behavior of RABAM (random adaptive bidirectional associative memory) Brownian-diffusion systems is summarized by the RABAM noise suppression theorem: the mean-squared activation and synaptic velocities decrease exponentially quickly to their lower bounds, the instantaneous noise variances driving the system.

A generalized EM algorithm for 3-D Bayesian reconstruction from Poisson data using Gibbs priors.

A generalized expectation-maximization (GEM) algorithm is developed for Bayesian reconstruction, based on locally correlated Markov random-field priors in the form of Gibbs functions and on the Poisson data model. For the M-step of the algorithm, a form of coordinate gradient ascent is derived. The algorithm reduces to the EM maximum-likelihood algorithm as the Markov random-field prior tends towards a uniform distribution.

The role of the periciliary fluid in mucociliary flows: flow velocity profiles in frog palate mucus.

Observations of flow velocity profiles over frog mucociliated palate are used to estimate viscosity, shear rate and shear stress in the periciliary flow field. The ability of cilia to generate significant shear stress at long distances and their utility as rhoeometers are examined. It is proposed that the depth of significant ciliary shear penetration into the periciliary fluid is sufficient to move mucus masses well beyond the ciliary tips, obviating the need for tip penetration where anchoring phenomena are sufficiently reduced.