A comparison of optimal MIMO linear and nonlinear models for brain-machine interfaces.

The field of brain-machine interfaces requires the estimation of a mapping from spike trains collected in motor cortex areas to the hand kinematics of the behaving animal. This paper presents a systematic investigation of several linear (Wiener filter, LMS adaptive filters, gamma filter, subspace Wiener filters) and nonlinear models (time-delay neural network and local linear switching models) applied to datasets from two experiments in monkeys performing motor tasks (reaching for food and target hitting). Ensembles of 100-200 cortical neurons were simultaneously recorded in these experiments, and even larger neuronal samples are anticipated in the future. Due to the large size of the models (thousands of parameters), the major issue studied was the generalization performance. Every parameter of the models (not only the weights) was selected optimally using signal processing and machine learning techniques. The models were also compared statistically with respect to the Wiener filter as the baseline. Each of the optimization procedures produced improvements over that baseline for either one of the two datasets or both.

Sleep spindle characteristics as a function of age.

The characteristics of sleep spindles as a function of age were studied with an automated analysis system. Five age groups (group 0, 3-5 years; I, 13 years; II, 25-34 years; III, 42-53 years; and IV, 67-79 years) were analyzed. Significant differences in spindle frequency were found between groups 0-I and II-III-IV. The frequency increased with increasing age. No age differences were found in spindle duration. Spindle amplitude reached a peak in the group I subjects, then decreased with increasing age.

3-D computer animation of electrophysiological responses.

Traditional methods for displaying electrophysiological data, that use time as the axis on a plot, are inadequate for displaying data from simultaneous multi-channel recordings. New methods proposed here plot the instantaneous value of the data on a third axis over a 2-dimensional spatial map of the tissue. The resulting 3-D computer-generated surfaces are animated over time to reveal simultaneous coherent waves of activity over the entire slice. This method was implemented for displaying multi-channel evoked potential data from rat hippocampal and human cortical slices. In rat hippocampal slice, stimulation of the Schaffer collateral-commissural pathway in stratum radiatum (SR) near CA2 elicited evoked extracellular responses along the length of CA1 from the alveus to stratum lacunosum moleculare (SLM). 3-D plotting and subsequent animation of these responses translated differential latencies of activation elicited across the slice into coherent moving patterns. These evoked waves of extracellular activity appeared to propagate along hippocampal laminae and were not readily visible in the individual plots. Human temporal cortical slices were stimulated in the white matter and evoked responses recorded in an array format. Upon plotting and animation, activity was seen to propagate vertically to the pial surface and thereafter move radially away from the stimulation site. Animation of 3-D plots of electrophysiological activity can provide instantaneous visual information on correlated changes in amplitude and latency over an entire brain slice. This means of displaying data can reduce a large number of complex wave forms to simple events and allow the simultaneous visualization of general patterns of activity in a large group of neurons.(ABSTRACT TRUNCATED AT 250 WORDS)

Sleep staging automaton based on the theory of evidence.

This paper addresses sleep staging as a medical decision problem. It develops a model for automated sleep staging by combining signal information, human heuristic knowledge in the form of rules, and a mathematical framework. The EEG/EOG/EMG events relevant for sleep staging are detected in real time by an existing front-end system and are summarized per minute. These token data are translated, normalized, and constitute the input alphabet to a finite state machine (automaton). The processed token events are used as partial belief in a set of anthropomimetic rules, which encode human knowledge about the occurrence of a particular sleep stage. The Dempster-Shafer theory of evidence weighs the partial beliefs and attributes the minutes sleep stage to the machine state transition that displays the highest final belief. Results are briefly presented.

TDAT--time domain analysis tool for EEG analysis.

An interactive design and analysis tool for displaying and quantifying multiple channels of data is presented. The system allows one to easily visualize multiple data channels and simultaneously observe the effects of filters on the data and to evaluate signal detection algorithms. The software is designed for a workstation environment; it will find application in a variety of applications where one needs to simultaneously visualize multiple data channels. TDAT is being used for the design and evaluation of filters and detection algorithms for electroencephalogram (EEG) waveforms, and it is serving as a prototype of a paperless system to be used by electroencephalographers. This paper describes the general software structure of the system and illustrates many of the system features with examples.

Chaotic activity during iron-induced "epileptiform" discharge in rat hippocampal slices.

Low-dimensional chaotic dynamics have been suggested in the rat hippocampal slice during iron-induced epileptiform activity. The dimensionality of this chaotic activity has been found to be similar in slices bathed in the same ionic extracellular medium. Some slices also displayed a drop in dimensionality prior to the onset of seizure-like activity. We suggest that techniques of nonlinear dynamical analysis are a useful reverse-engineering tool for studying the in vitro brain slice. We further conclude that neuronal circuits capable of displaying chaotic activity could exist at the level of the in vitro brain slice.

Multiresolution segmentation of respiratory electromyographic signals.

Analysis of respiratory electromyographic (EMG) signals in the study of respiratory control requires the detection of burst activity from background (signal segmentation), and focuses upon the determination of onset and cessation points of the burst activity (boundary estimation). This paper describes a new automated multiresolution technique for signal segmentation and boundary estimation. During signal segmentation, a new transitional segment is defined which contains the boundary between background and burst activity. Boundary estimation is then performed within this transitional segment. Boundary candidates are selected and a probability is attributed to each candidate, using an artificial neural network. The final boundary for a given transitional segment is the boundary estimate with the maximum a posteriori probability. This new method has proved accurate when compared to boundaries chosen by two investigators.

Target discrimination in synthetic aperture radar using artificial neural networks.

This paper addresses target discrimination in synthetic aperture radar (SAR) imagery using linear and nonlinear adaptive networks. Neural networks are extensively used for pattern classification but here the goal is discrimination. We show that the two applications require different cost functions. We start by analyzing with a pattern recognition perspective the two-parameter constant false alarm rate (CFAR) detector which is widely utilized as a target detector in SAR. Then we generalize its principle to construct the quadratic gamma discriminator (QGD), a nonparametrically trained classifier based on local image intensity. The linear processing element of the QCD is further extended with nonlinearities yielding a multilayer perceptron (MLP) which we call the NL-QGD (nonlinear QGD). MLPs are normally trained based on the L(2) norm. We experimentally show that the L(2) norm is not recommended to train MLPs for discriminating targets in SAR. Inspired by the Neyman-Pearson criterion, we create a cost function based on a mixed norm to weight the false alarms and the missed detections differently. Mixed norms can easily be incorporated into the backpropagation algorithm, and lead to better performance. Several other norms (L(8), cross-entropy) are applied to train the NL-QGD and all outperformed the L(2) norm when validated by receiver operating characteristics (ROC) curves. The data sets are constructed from TABILS 24 ISAR targets embedded in 7 km(2) of SAR imagery (MIT/LL mission 90).

Generalized information potential criterion for adaptive system training.

We have previously proposed the quadratic Renyi's error entropy as an alternative cost function for supervised adaptive system training. An entropy criterion instructs the minimization of the average information content of the error signal rather than merely trying to minimize its energy. In this paper, we propose a generalization of the error entropy criterion that enables the use of any order of Renyi's entropy and any suitable kernel function in density estimation. It is shown that the proposed entropy estimator preserves the global minimum of actual entropy. The equivalence between global optimization by convolution smoothing and the convolution by the kernel in Parzen windowing is also discussed. Simulation results are presented for time-series prediction and classification where experimental demonstration of all the theoretical concepts is presented.

Super-resolution of images based on local correlations.

An adaptive two-step paradigm for the superresolution of optical images is developed in this paper. The procedure locally projects image samples onto a family of kernels that are learned from image data. First, an unsupervised feature extraction is performed on local neighborhood information from a training image. These features are then used to cluster the neighborhoods into disjoint sets for which an optimal mapping relating homologous neighborhoods across scales can be learned in a supervised manner. A super-resolved image is obtained through the convolution of a low-resolution test image with the established family of kernels. Results demonstrate the effectiveness of the approach.

Training neural networks with additive noise in the desired signal.

A new global optimization strategy for training adaptive systems such as neural networks and adaptive filters [finite or infinite impulse response (FIR or IIR)] is proposed in this paper. Instead of adding random noise to the weights as proposed in the past, additive random noise is injected directly into the desired signal. Experimental results show that this procedure also speeds up greatly the backpropagation algorithm. The method is very easy to implement in practice, preserving the backpropagation algorithm and requiring a single random generator with a monotonically decreasing step size per output channel. Hence, this is an ideal strategy to speed up supervised learning, and avoid local minima entrapment when the noise variance is appropriately scheduled.

An analysis of the gamma memory in dynamic neural networks.

Presents a vector space framework to study short-term memory filters in dynamic neural networks. The authors define parameters to quantify the function of feedforward and recursive linear memory filters. They show, using vector spaces, what is the optimization problem solved by the PEs of the first hidden layer of the single input focused network architecture. Due to the special properties of the gamma bases, recursion brings an extra parameter lambda (the time constant of the leaky integrator) that displaces the memory manifold towards the desired signal when the mean square error is minimized. In contrast, for the feedforward memory filter the angle between the desired signal and the memory manifold is fixed for a given memory order. The adaptation of the feedback parameter can be done using gradient descent, but the optimization is nonconvex.

Incremental backpropagation learning networks.

How to learn new knowledge without forgetting old knowledge is a key issue in designing an incremental-learning neural network. In this paper, we present a new incremental learning method for pattern recognition, called the "incremental backpropagation learning network", which employs bounded weight modification and structural adaptation learning rules and applies initial knowledge to constrain the learning process. The viability of this approach is demonstrated for classification problems including the iris and the promoter domains.

Microcomputer-based system for the detection and quantification of petit mal epilepsy.

A petit mal seizure detector totally implemented in a 16 bit microcomputer and capable of analyzing on-line at least one channel of EEG data is described. The system uses the repetition period of the wave complexes as the primary parameter for the detection and performs well in clinically significant seizures. Besides characterizing the paroxysms in duration and time of occurrence, the system also evaluates on-line the mean values and variances of the detection parameters, yielding more quantitative information about the seizure data than previously described systems.

Extraction of principal components from biosignals by neural net.

This contribution gives the information on a useful application of principal component analysis (PCA) in the field of electroencephalogram (EEG) and laser-Doppler signal processing. The principal components are estimated by a neural network (NN) approach.

Detection of synchronized firings in multivariate neural spike trains during motor tasks.

This paper describes and compares two classical methods for the detection of neuron groups which exhibit synchronized firings in multivariate spike trains. These methods were compared on experimental and randomized data corresponding to the firing activity of 104 neurons located in motor, premotor, and parietal cortices in a monkey during movement tasks. Both methods exhibited high false positive rates in randomized data, but results showed that this rate can be advantageously reduced with a simple postprocessing. Otherwise, one method permitted to detect a significant number of synchronized groups of neurons related to the behavioral task.

Visualization of the short term maximum Lyapunov exponent topography in the epileptic brain.

In this paper, a new kind of brain topography is introduced and applied to data from four patients affected by intractable epilepsy. Experience has shown that the short term maximum Lyapunov exponent (STLmax) is a robust parameter when optimized for the dynamical analysis of the electroencephalography (EEG). The objective of this work is to map the spatial distribution of STLmax over time. STLmax is estimated from segments of each channel of long term continuous scalp EEG recordings and a movie of the STLmax segment estimates is created over the head. Movies allow for a simple visualization of which electrodes are related to the highest or lowest chaoticity for the longest time. We found out that the interictal epileptiform activity is related to the highest STLmax level, whereas the focal area is related to low STLmax levels during either the interictal and preictal stages.