Motion detection by biological asymmetrical neural network

Nonlinearity is an important factor in the biological visual neural networks. Among prominent features of the visual networks, movement detections are carried out in the visual cortex. The visual cortex for the movement detection, consist of two layered networks, called the primary visual cortex (V1), followed by the middle temporal area (MT), in which nonlinear functions will play important roles in the visual systems. These networks will be decomposed to asymmetric sub-networks with nonlinearities. In this paper, the fundamental characteristics in asymmetric and symmetric neural networks with nonlinearities are developed for the detection of the changing stimulus or the movement detection in these neural networks. By the optimization of the asymmetric networks, movement detection Equations are derived. Then, it was clarified that the even – odd nonlinearity combined asymmetric networks, has the ability of generating directional vector in the stimulus change detection or movement detection, while symmetric networks need the time memory to have the same ability. Further, the vector operations in the neural network are developed. These facts are applied to two layered networks, V1 and MT.

Visual motion information is useful for many complex tasks in the biological and robotic systems. Models for motion processing in the biological systems have been studied to use conventional symmetric quadrature functions with Gabor filters. This paper proposes a model of the another bio-inspired asymmetric neural networks. The prominent features are the nonlinear characteristics as the squaring and rectification functions, which are observed in the retinal and visual cortex networks. In this paper, the asymmetric network with Gabor filters is compared with that of the conventional symmetric networks. It is shown that the biological asymmetric network with nonlinearities is effective for detecting the inputted phase information and directional movements from the network computations. The responses to the frequency characteristics and to the complex motion stimulus are computed in the asymmetric networks, which are not derived for the conventional energy model.

To make clear the mechanism of the visual movement is important in the visual system. The prominent feature is the nonlinear characteristics as the squaring and rectification functions, which are observed in the retinal and visual cortex networks. Conventional model for motion processing in cortex, is the use of symmetric quadrature functions with Gabor filters. This paper proposes a new motion sensing processing model in the asymmetric networks. To make clear the behavior of the asymmetric nonlinear network, white noise analysis and Wiener kernels are applied. It is shown that the biological asymmetric network with nonlinearities is effective and general for generating the directional movement from the network computations. The qualitative analysis is performed between the asymmetrical network and the conventional quadrature model. The results are applicable to the V1 and MT model of the neural networks in the cortex.

1. Simultaneous intracellular recordings of responses to light flashes were obtained from rod-horizontal cell and rod-hyperpolarizing bipolar cell pairs in isolated retinae of the toad. The gain and temporal filtering of synaptic transfer were characterized throughout the rods' range of light responses. 2. Paired rod-horizontal cell and rod-bipolar cell responses to dim flashes (less than 0.4 Rh*, where Rh* denotes effective photoisomerizations per rod per flash) exhibited nearly the same time course. Analysis of the onset of the horizontal cell responses revealed a temporal lag equivalent to a single stage of low-pass filtering (tau f = 75-200 ms). No filtering was discerned in the transfer of dim-flash responses from rods to bipolars. On average, horizontal cells were five times as sensitive (mV/Rh*) and hyperpolarizing bipolar cells 10.7 times as sensitive as their paired rods. 3. For brighter flashes, up to 1600 Rh*, the rising and return phases of bipolar responses appeared to be simple scaled versions of the rod responses. The scaling factor was equal to the ratio of flash sensitivities for dim flashes. Rod responses greater than about 2 mV produced a saturation of the bipolar cell response. 4. The return phases of the horizontal cell responses were kinetically similar, scaled versions of the rod responses for rod potentials less than about 5 mV. However, the rising phases lagged significantly behind those of the rod. The effective time constant of the lag increased proportionally with flash intensity. For the brighter flashes, the horizontal cell response peaked as much as a second after the rod response. 5. The linear scaling, minimal temporal filtering and saturation of the bipolar cell responses were satisfactorily reproduced by a model of synaptic transfer that assumed that the rate of transmitter release followed the rod voltage exponentially and that the postsynaptic conductance followed Michaelis-Menten saturation (Falk & Fatt, 1972). 6. The progressively longer lag in the horizontal cell responses to brighter flashes was satisfactorily simulated by a kinetically limited Falk and Fatt model which postulated that the effective electrical time constant of the horizontal cell membrane strongly depended on synaptic or voltage-modulated conductances. 7. Satisfactory model simulations of all postsynaptic responses required that an e-fold change in the release rate of transmitter from the rod be obtained with a 2 mV change in the rod potential.

There are two types of horizontal cells, the luminosity and the chromaticity cells, in the retina of the kissing gourami, Helostoma rudolfi. Luminosity cells occupy the outermost layer proximal to the receptor terminals, whereas chromaticity cells form a layer proximal to the layer of luminosity cells. Neither type of cell has axons. Responses were evoked by light from red and green light-emitting diodes. The two stimuli were modulated either by a pulsatile or a white-noise signal. The luminosity cell always produced a hyperpolarizing response. The chromaticity cell produced a hyperpolarizing response when stimulated by only one color. However, in the presence of a steady or modulated green input, a red stimulus produced a depolarizing response. Such chromaticity cells were similar to the (spectral) biphasic chromaticity horizontal cells observed in other retinae. The depolarizing phase of the red response was produced by the balance of intensity of the two inputs, red and green. We used white-noise methodology to identify the dynamics of the horizontal cell's modulation response by taking advantage of the fact that a Wiener kernel is a measure of a cell's incremental sensitivity, which includes its response dynamics. Under all conditions, a steady state modulation response by both luminosity and chromaticity cells always was related linearly to the input modulation. The average mean square error (MSE) of the model predicted by the first-order kernel was approximately 8% for both luminosity (n = 116) and chromaticity (n = 23) cells. In some cases, the MSE was a few percent even when the peak-to-peak response amplitude was nearly 30 mV. The ratio of inputs from red and green cones to both types of horizontal cells was variable; the major input for luminosity cells came from red cones, whereas the major input for chromaticity cells came from green cones. First-order kernels generated by the major input were robust in terms of waveform in the sense that the waveform remained unchanged whether or not there was a steady or modulated illumination by the opposing color. The results reported here do not address the question of the neural circuitry that generates horizontal cell responses, in particular, the depolarizing response. However, whatever that circuitry might be, the high degree of linearity of the modulation response by both types of cell under various stimulus conditions imposes restrictions on the performance of any proposed model as well as on mechanisms that underlie the generation of the horizontal cell response.

To make clear the mechanism of the visual movement is important in he visual system. The prominent feature is the nonlinear characteristics as the squaring and rectification functions, which are observed in the retinal and visual cortex networks. Many popular models for motion processing in cortex, is the use of symmetric quadrature functions with Gabor filters. This paper proposes a new motion sensing processing model in the asymmetric networks. To make clear the behavior of the asymmetric nonlinear network, white noise analysis and Wiener kernels are applied. It is shown that the biological asymmetric network with nonlinearities is effective and superior for generating the directional movement from the network computations. The directive equations for the stimulus were derived by applying the white noise analysis and kernels to characterize stimulus changes. The results are applicable to the V1 and MT model of the neural networks in the cortex.

The influence of HEPES on light responses of rabbit horizontal cells

The prominent feature is the nonlinear characteristics as the squaring and rectification functions, which are observed in the retinal and visual cortex networks. Conventional model for motion processing in cortex, uses a symmetric quadratic functions with Gabor filters. This paper proposes a new motion processing model in the asymmetric networks. First, the asymmetric network is analyzed using Wiener kernels. It is shown that the asymmetric network with nonlinearities is effective and general for generating the directional movement compared with the conventional quadratic model. Second, independence maximization of data is an important issue in computational neural networks. To make clear the characteristics of the asymmetric network with Gabor functions, orthogonality is computed, which shows independent characteristics of the asymmetric network without maximizing optimization of independence in the quadratic model. The orthogonal analyses for the independence of the asymmetric networks are applied to the V1 and MT neural networks to generate independent subspaces by using selective Gabor functions.

Reconfigurable logic gates in biological crossbar neural networks using STDP learning.

This paper deals not only with pointwise and uniform convergence but also Y-valued fractional approximation results by univariate symmetrized neural network (SNN) operators on Banach space Y,.. Moreover, our main motivation in this work is to compare the convergence results obtained by classical neural network (NN) operators and symmetric neural network (SNN) operators and try to convert them into numerical examples and graphs through computer programming language Python codes. As a result of this experimental study conducted under the regime of certain parameters, the convergence speed and results of SNN operators are superior to those of classical NN operators.

The classical interchange (permutation) of atoms of similar identity does not have an effect on the overall potential energy. In this study, we present feed-forward neural network structures that provide permutation symmetry to the potential energy surfaces of molecules. The new feed-forward neural network structures are employed to fit the potential energy surfaces for two illustrative molecules, which are H(2)O and ClOOCl. Modifications are made to describe the symmetric interchange (permutation) of atoms of similar identity (or mathematically, the permutation of symmetric input parameters). The combined-function-derivative approximation algorithm (J. Chem. Phys. 2009, 130, 134101) is also implemented to fit the neural-network potential energy surfaces accurately. The combination of our symmetric neural networks and the function-derivative fitting effectively produces PES fits using fewer numbers of training data points. For H(2)O, only 282 configurations are employed as the training set; the testing root-mean-squared and mean-absolute energy errors are respectively reported as 0.0103 eV (0.236 kcal/mol) and 0.0078 eV (0.179 kcal/mol). In the ClOOCl case, 1693 configurations are required to construct the training set; the root-mean-squared and mean-absolute energy errors for the ClOOCl testing set are 0.0409 eV (0.943 kcal/mol) and 0.0269 eV (0.620 kcal/mol), respectively. Overall, we find good agreements between ab initio and NN prediction in term of energy and gradient errors, and conclude that the new feed-forward neural-network models advantageously describe the molecules with excellent accuracy.

Digital image halftoning is treated as an optimization problem to which neural networks provide an efficient parallel solution. An image distortion measure is introduced in which the gray-tone image is approximated by a filtered version of the halftoned image. This distortion measure is minimized by using a near-neighborhood-connected symmetric neural network. The filter used in the distortion measure is estimated on the basis of a training set of gray-tone and bilevel images. This filter is then used for the reconstruction of a gray-tone image from its halftoned version. The above procedure, combined with a postprocessing of the reconstructed image by a nonlinear edge-preserving noise-smoothing filter, provides images of good quality. >

1. The electrical properties of horizontal cells in the mudpuppy in light and dark were measured with a pair of micropipettes separated by about 1 mum with low coupling resistance so that no bridge circuitry was required. 2. All horizontal cells studied showed significant anomalous rectification: the current-voltage characteristic for about 60 per cent of the cells studied had a slope resistance of about 20-30 M omega at the dark potential level; the slope resistance increased by about 15% for each 10 mV depolarization and decreased by about 15% for each 10 mV hyperpolarization. The remaining 40% of the horizontal cells showed a higher input resistance at corresponding potential levels but had similar rectifying properties. 3. The increase in resistance with depolrization developed with a time course of about 1/2 sec when steady steps of outward current were passed across the membrane, but the time course for resistance decrease with hyperpolarization was much shorter for steady inward current steps. In about half the horizontal cells there was a transient decrease in resistance lasting about 100 msec immediately following the outward current steps superimposed upon the slower sustained resistance increase. 4. The normal 20-30 mV hyperpolarizing light response was associated with little or no change in input resistance. However, if the membrane potential was held at the dark potential level with extrinsic current, thereby eliminating the potential-dependent resistance change, a light-elicited resistance increase of about 10 M omega was measured. 5. The time-dependent change in membrane resistance elicited by polarizing steps of current obscured the reversal potential for the response. However, when the reversal potential was measured at short times following polarization of the membrane, before the time-dependent resistance change developed, it was estimated at between +15 and +50 m V. 6. The results suggest that the horizontal cell response is mediated by a light-elicited resistance increase at the synaptic membrane which is obscured by a potential- and time-dependent resistance decrease at another part of the membrane.

One of the central tasks in retinal neuroscience is to understand the circuitry of retinal neurons and how those connections are responsible for shaping the signals transmitted to the brain. Photons are detected in the retina by rod and cone photoreceptors, which convert that energy into an electrical signal, transmitting it to other retinal neurons, where it is processed and communicated to central targets in the brain via the optic nerve. Important early insights into retinal circuitry and visual processing came from the histological studies of Cajal1,2 and, later, from electrophysiological recordings of the spiking activity of retinal ganglion cells - the output cells of the retina3,4. A detailed understanding of visual processing in the retina requires an understanding of the signaling at each step in the pathway from photoreceptor to retinal ganglion cell. However, many retinal cell types are buried deep in the tissue and therefore relatively inaccessible for electrophysiological recording. This limitation can be overcome by working with vertical slices, in which cells residing within each of the retinal layers are clearly visible and accessible for electrophysiological recording. Here, we describe a method for making vertical sections of retinas from larval tiger salamanders (Ambystoma tigrinum). While this preparation was originally developed for recordings with sharp microelectrodes5,6, we describe a method for dual whole-cell voltage clamp recordings from photoreceptors and second-order horizontal and bipolar cells in which we manipulate the photoreceptor's membrane potential while simultaneously recording post-synaptic responses in horizontal or bipolar cells. The photoreceptors of the tiger salamander are considerably larger than those of mammalian species, making this an ideal preparation in which to undertake this technically challenging experimental approach. These experiments are described with an eye toward probing the signaling properties of the synaptic ribbon - a specialized synaptic structure found in a only a handful of neurons, including rod and cone photoreceptors, that is well suited for maintaining a high rate of tonic neurotransmitter release7,8 - and how it contributes to the unique signaling properties of this first retinal synapse.

The short and long-term effects of temperature on the dynamic range of signalling in horizontal cells of carp retina