José Manuel Ferrández

Population coding in spike trains of simultaneously recorded retinal ganglion cells.

To achieve a better understanding of the parallel information processing that takes place in the nervous system, many researchers have recently begun to use multielectrode techniques to obtain high spatial- and temporal-resolution recordings of the firing patterns of neural ensembles. Apart from the complexities of acquiring and storing single unit responses from large numbers of neurons, the multielectrode technique has provided new challenges in the analysis of the responses from many simultaneously recorded neurons. This paper provides insights into the problem of coding/decoding of retinal images by ensembles of retinal ganglion cells. We have simultaneously recorded the responses of 15 ganglion cells to visual stimuli of various intensities and wavelengths and analyzed the data using discriminant analysis. Models of stimulus encoding were generated and discriminant analysis used to estimate the wavelength and intensity of the stimuli. We find that the ganglion cells we have recorded from are non-redundant encoders of these stimulus features. While single ganglion cells are poor classifiers of the stimulus parameters, examination of the responses of only a few ganglion cells greatly enhances our ability to specify the stimulus wavelength and intensity. Of the parameters studied, we find that the rate of firing of the ganglion cells provides the most information about these stimulus parameters, while the timing of the first action potential provides almost as much information. While we are not suggesting that the brain is using these variables, our results show how a population of sensory neurons can encode stimulus features and suggest that the brain could potentially deduce reliable information about stimulus features from response patterns of retinal ganglion cell populations.

DATA-MEAns: An open source tool for the classification and management of neural ensemble recordings

The number of laboratories using techniques that allow to acquire simultaneous recordings of as many units as possible is considerably increasing. However, the development of tools used to analyse this multi-neuronal activity is generally lagging behind the development of the tools used to acquire these data. Moreover, the data exchange between research groups using different multielectrode acquisition systems is hindered by commercial constraints such as exclusive file structures, high priced licenses and hard policies on intellectual rights. This paper presents a free open-source software for the classification and management of neural ensemble data. The main goal is to provide a graphical user interface that links the experimental data to a basic set of routines for analysis, visualization and classification in a consistent framework. To facilitate the adaptation and extension as well as the addition of new routines, tools and algorithms for data analysis, the source code and documentation are freely available.

A retinomorphic architecture based on discrete-time cellular neural networks using reconfigurable computing

This paper describes a novel architecture for the hardware implementation of non-linear multi-layer cellular neural networks (CNNs). This makes it feasible to design CNNs with millions of neurons accommodated in low price FPGA devices, being able to process standard video in real time. This architecture has been used to build a CNN-based model of the synapsis I of the fovea region, with the aim of implementing the basic spatial processing of the retina in reconfigurable hardware. The model is based on the receptive fields of the bipolar cells and mimics the retinal architecture achieving its processing capabilities.

Stress detection using wearable physiological and sociometric sensors

Stress remains a significant social problem for individuals in modern societies. This paper presents a machine learning approach for the automatic detection of stress of people in a social situation by combining two sensor systems that capture physiological and social responses. We compare the performance using different classifiers including support vector machine, AdaBoost, and [Formula: see text]-nearest neighbor. Our experimental results show that by combining the measurements from both sensor systems, we could accurately discriminate between stressful and neutral situations during a controlled Trier social stress test (TSST). Moreover, this paper assesses the discriminative ability of each sensor modality individually and considers their suitability for real-time stress detection. Finally, we present an study of the most discriminative features for stress detection.

RetinaStudio: A bioinspired framework to encode visual information

Abstract The retina is a very complex neural structure, which performs spatial, temporal, and chromatic processing on visual information and converts it into a compact ‘digital’ format composed of neural impulses. This paper presents a new compiler-based framework able to describe, simulate and validate custom retina models. The framework is compatible with the most usual neural recording and analysis tools, taking advantage of the interoperability with these kinds of applications. Furthermore it is possible to compile the code to generate accelerated versions of the visual processing models compatible with COTS microprocessors, FPGAs or GPUs. The whole system represents an ongoing work to design and develop a functional visual neuroprosthesis. Several case studies are described to assess the effectiveness and usefulness of the framework.

New emulated discrete model of CNN architecture for FPGA and DSP applications

A new approach to Cellular Neural Networks discrete model is proposed. This approach is focused on CNN implementation on reconfigurable hardware architectures and DSP microprocessors. CNN are analysed from the perspective of Systems Theory, giving rise to an alternative model to those found in the literature available. Dynamic equations and their solutions, stability analysis and real-time implementation architecture are described in this paper as the most relevant points in the development of our model. The main results, obtained from different simulations, evidence the usefulness and functionality of the model.

V-Proportion: A method based on the Voronoi diagram to study spatial relations in neuronal mosaics of the retina

The visual system plays a predominant role in the human perception. Although all components of the eye are important to perceive visual information, the retina is a fundamental part of the visual system. In this work we study the spatial relations between neuronal mosaics in the retina. These relations have shown its importance to investigate possible constraints or connectivities between different spatially colocalized populations of neurons, and to explain how visual information spreads along the layers before being sent to the brain. We introduce the V-Proportion, a method based on the Voronoi diagram to study possible spatial interactions between two neuronal mosaics. Results in simulations as well as in real data demonstrate the effectiveness of this method to detect spatial relations between neurons in different layers.

Study of the contrast processing in the early visual system using a neuromorphic retinal architecture

In this paper we propose a retinal architecture that incorporates the neural circuits found in the different retinal regions. It is implemented in a reconfigurable system for observing in real time the contrast processing capabilities of each retinal region over the provided stimuli. The retina model is based on a discrete-time cellular neural network (DTCNN) that will be implemented on reconfigurable architecture (FPGA) with a time multiplexing approach. This architecture is able to incorporate 50 million neurons in its structure for processing video in real time. It has been observed that the contrast detection and the detail resolution are influenced by the convergence factor of neurons and by the lateral inhibition, specific characteristics of each neural circuit.

A neural network approach for the analysis of multineural recordings in retinal ganglion cells

In this paper the coding capabilities of individual retinal ganglion cells are compared with respect to the coding capabilities of small population of cells using different neural networks. This approach allows not only the identification of the most discriminating cells, but also detection of the parameters that are more important for the discrimination task. Our results show that the spike rate together with the exact timing of the first spike at light-ON were the most important parameters for encoding stimulus features. Furthermore we found that whereas single ganglion cells are poor classifiers of visual stimuli, a population of only 15 cells can distinguish stimulus color and intensity reasonable well. This demonstrates that visual information is coded as the overall set of activity levels across neurons rather than by single cells.

A biological neuroprocessor for robotic guidance using a center of area method

This paper introduces a neuroinformatic system using human neuroblastoma cultures and centre of area learning for basic robotic guidance. Multielectrode Arrays Setups have been designed for direct culturing neural cells over silicon or glass substrates, providing the capability to stimulate and record simultaneously populations of neural cells. The main objective of this work will be to control a robot using this biological neuroprocessor and a simple centre of area learning scheme. The final system could be applied for testing how chemicals affect the behaviour of the robot or to establish the basis for new hybrid optogenetic learning.