Narrow Results

This paper proposes the idea that Hindmarsh–Rose (HR) neuronal model can be used to develop a new type of central pattern generator (CPG). Some key properties of HR model are studied and proved to meet the requirements of CPG. Pros and cons of HR model are provided. A CPG network based on HR model is developed and the related properties are investigated. We explore the bipedal primary gaits generated by the CPG network. The preliminary applications of HR model are tested on humanoid locomotion model and functional electrical stimulation (FES) walking system. The positive results of stimulation and experiment show the feasibility of HR model as a valid CPG.

Target waves in excitable media such as neuronal network can regulate the spatial distribution and orderliness as a continuous pacemaker. Three different schemes are used to develop stable target wave in the network, and the potential mechanism for emergence of target waves in the excitable media is investigated. For example, a local pacing driven by external periodical forcing can generate stable target wave in the excitable media, furthermore, heterogeneity and local feedback under self-feedback coupling are also effective to generate continuous target wave as well. To discern the difference of these target waves, a statistical synchronization factor is defined by using mean field theory and artificial defects are introduced into the network to block the target wave, thus the robustness of these target waves could be detected. However, these target waves developed from the above mentioned schemes show different robustness to the blocking from artificial defects. A regular network of Hindmarsh–Rose neurons is designed in a two-dimensional square array, target waves are induced by using three different ways, and then some artificial defects, which are associated with anatomical defects, are set in the network to detect the effect of defects blocking on the travelling waves. It confirms that the robustness of target waves to defects blocking depends on the intrinsic properties (ways to generate target wave) of target waves.

In this paper, burst synchronization and rhythm dynamics of a small-world neuronal network consisting of mixed bursting types of neurons coupled via inhibitory–excitatory chemical synapses are explored. Two quantities, the synchronization parameter and average width factor, are used to characterize the synchronization degree and rhythm dynamics of the neuronal network. Numerical results show that the percentage of the inhibitory synapses in the network is the major factor for we get a similarly bell-shaped dependence of synchronization on it, and the decrease of the average width factor of the network. We also find that not only the value of the coupling strength can promote the synchronization degree, but the probability of random edges adding to the small-world network also can. The ratio of the long bursting neurons has little effect on the burst synchronization and rhythm dynamics of the network.

Spiral waves were observed in the biological experiment on rat brain cortex with the application of carbachol and bicuculline which can block inhibitory coupling from interneurons to pyramidal neurons. To simulate the experimental spiral waves, a two-dimensional neuronal network composed of pyramidal neurons and inhibitory interneurons was built. By decreasing the percentage of active inhibitory interneurons, the random-like spatial patterns change to spiral waves and to random-like spatial patterns or nearly synchronous behaviors. The spiral waves appear at a low percentage of inhibitory interneurons, which matches the experimental condition that inhibitory couplings of the interneurons were blocked. The spiral waves exhibit a higher order or signal-to-noise ratio (SNR) characterized by spatial structure function than both random-like spatial patterns and nearly synchronous behaviors, which shows that changes of the percentage of active inhibitory interneurons can induce spatial coherence resonance-like behaviors. In addition, the relationship between the coherence degree and the spatial structures of the spiral waves is identified. The results not only present a possible and reasonable interpretation to the spiral waves observed in the biological experiment on the brain cortex with disinhibition, but also reveal that the spiral waves exhibit more ordered degree in spatial patterns.

Pattern estimation and selection in media can give important clues to understand the collective response to external stimulus by detecting the observable variables. Both reaction–diffusion systems (RDs) and neuronal networks can be treated as multi-agent systems from molecular level, intrinsic cooperation, competition. An external stimulus or attack can cause collapse of spatial order and distribution, while appropriate noise can enhance the consensus in the spatiotemporal systems. Pattern formation and synchronization stability can bridge isolated oscillators and the network by coupling these nodes with appropriate connection types. As a result, the dynamical behaviors can be detected and discussed by developing different spatial patterns and realizing network synchronization. Indeed, the collective response of network and multi-agent system depends on the local kinetics of nodes and cells. It is better to know the standard bifurcation analysis and stability control schemes before dealing with network problems. In this review, dynamics discussion and synchronization control on low-dimensional systems, pattern formation and synchronization stability on network, wave stability in RDs and neuronal network are summarized. Finally, possible guidance is presented when some physical effects such as polarization field and electromagnetic induction are considered.

Through introducing the ingredients of electromagnetic induction and coupled time delay into the original Fitzhugh–Nagumo (FHN) neuronal network, the dynamics of stochastic resonance in a model of modified FHN neuronal network in the environment of phase noise is explored by numerical simulations in this study. On one hand, we demonstrate that the phenomenon of stochastic resonance can appear when the intensity of phase noise is appropriately adjusted, which is further verified to be robust to the edge-added probability of small-world network. Moreover, under the influence of electromagnetic induction, the phase noise-induced resonance response is suppressed, meanwhile, a large noise intensity is required to induce stochastic resonance as the feedback gain of induced current increases. On the other hand, when the coupled time delay is incorporated into this model, the results indicate that the properly tuned time delay can induce multiple stochastic resonances in this neuronal network. However, the phenomenon of multiple stochastic resonances is found to be restrained upon increasing feedback gain of induced current. Surprisingly, by changing the period of phase noise, multiple stochastic resonances can still emerge when the coupled time delay is appropriately tuned to be integer multiples of the period of phase noise.

Using the Hindmarsh–Rose (HR) model, a two-layer neuronal network is constructed to study the spiral wave dynamics. The first layer generates spiral wave induced by random values of boundary under appropriate coupling intensity and external force, and the second layer is in the different states. Coupling channels between the two layers are set in multiple areas and spiral wave of first layer affect second layer via the coupling channels. The spatiotemporal pattern of neuronal network is investigated in the second layer. It is shown that spiral wave can be found under appropriate conditions, multi-area channels are more likely to generate spiral waves and target waves than only one local coupling area. A statistical variable on the second layer is calculated by increasing intensity of channels between layers. The larger the coupling areas between layers, the more obvious the synchronism phenomenon is.

In this paper, a modified Hindmarsh–Rose neuron model is presented, which has a fractional-order threshold magnetic flux. The dynamics of the model is investigated by bifurcation diagrams and Lyapunov exponents in two cases of presence and absence of the external electromagnetic induction. Then the emergence of the spiral waves in the network of the proposed model is studied. To find the effects of different factors on the formation and destruction of spiral waves, the external current, the coupling strength and the external stimuli amplitude are varied. It is observed that all of these parameters have significant impacts on the spiral waves. Furthermore, the external electromagnetic induction influences the existence of spiral waves in specific external current values.

Noise-induced single spatial coherence resonance (CR) and multiple spatial CRs simulated in a network have been reported independently in previous studies. In this paper, the relationship between the single and multiple spatial CRs is established by adjusting the initial values of the network composed of Morris–Lecar (ML) model neurons. The ML model manifests a saddle-node bifurcation on an invariant cycle through which a resting state is changed to a stable limit cycle corresponding to period-1 firing. Under resting state, a stable node, a saddle, and an unstable focus coexist. The membrane potential of the unstable focus is much higher than that of the stable node. When the initial value is closer to the unstable focus, the residence time of membrane potential on a high level is longer; correspondingly, the spatial CRs appear more frequently with respect to noise intensity and the coherence degree becomes stronger. The single spatial CR is induced by noise with high intensity. Multiple spatial CRs are induced by noise with high, middle, and even low noise intensities, respectively. When the initial values are closer to an unstable focus, the residence time of membrane potentials on a higher level is longer, which is important to the generation of multiple CRs, and builds a relationship between single and multiple spatial CRs.

We investigate the effects of disorder on the synchronized state of a network of Hindmarsh–Rose neuronal models. Disorder, introduced as a perturbation of the neuronal parameters, destroys the network activity by wrecking the synchronized state. The dynamics of the synchronized state is analyzed through the Kuramoto order parameter, adapted to the neuronal Hindmarsh–Rose model. We find that the coupling deeply alters the dynamics of the single units, thus demonstrating that coupling not only affects the relative motion of the units, but also the dynamical behavior of each neuron; Thus, synchronization results in a structural change of the dynamics. The Kuramoto order parameter allows to clarify the nature of the transition from perfect phase synchronization to the disordered states, supporting the notion of an abrupt, second order-like, dynamical phase transition. We find that the system is resilient up to a certain disorder threshold, after that the network abruptly collapses to a desynchronized state. The loss of perfect synchronization seems to occur even for vanishingly small values of the disorder, but the degree of synchronization (as measured by the Kuramoto order parameter) gently decreases, and the completely disordered state is never reached.

In this paper, we focus on investigating the effects of time delay on burst synchronization transitions of a neuronal network which is locally modeled by Hindmarsh–Rose neurons. Here, neurons inside the neuronal network are connected through electrical synapses or chemical synapses. With the numerical results, it is revealed that burst synchronization transitions of both electrically and chemically coupled neuronal networks could be induced by time delay just when the coupling strength is large enough. Meanwhile, it is found that, in electrically and excitatory chemically coupled neuronal networks, burst synchronization transitions are observed through change of spiking number per burst when coupling strength is large enough; while in inhibitory chemically coupled neuronal network, burst synchronization transitions are observed for large enough coupling strength through changing fold-Hopf bursting activity to fold-homoclinic bursting activity and vice versa. Namely, two types of burst synchronization transitions are observed. One type of burst synchronization transitions occurs through change of spiking numbers per burst and the other type of burst synchronization transition occurs through change of bursting types.

The present work focuses on the existence of chimera states in two-layer networks of locally, nonlocally, and globally coupled Hindmarsh–Rose neurons. We show that chimera states occur in all three coupled neuronal networks by changing synaptic coupling strength and the number of coupled neighbors, and traveling chimera state exists in networks by using local and nonlocal couplings. Particularly, we find that in the nonlocally coupled network, there exist regions of coherent state, which can produce a wave such that the period becomes longer with the increase of the number of coupled neighbors. Interestingly, we observe a new chimera state with the coexistence of interlayer synchronous and asynchronous chimera states, which we named as the interlayer semi-synchronous chimera state. In addition, the results indicate that the interaction between layers in neuronal networks can induce different types of chimera states and firing patterns, which could be useful to control neural firing patterns and deepen the understanding of neuronal evolution where coherent and incoherent dynamics coexist.

Fractional-order models describing neuronal dynamics often exhibit better compatibility with diverse neuronal firing patterns that can be observed experimentally. Due to the overarching significance of synchronization in neuronal dynamics, we here study synchronization in multiplex neuronal networks that are composed of fractional-order Hindmarsh–Rose neurons. We compute the average synchronization error numerically for different derivative orders in dependence on the strength of the links within and between network layers. We find that, in general, fractional-order models synchronize better than integer-order models. In particular, we show that the required interlayer and intralayer coupling strengths for interlayer or intralayer synchronization can be weaker if we reduce the derivative order of the model describing the neuronal dynamics. Furthermore, the dependence of the interlayer or intralayer synchronization on the intralayer or interlayer coupling strength vanishes with decreasing derivative order. To support these results analytically, we use the master stability function approach for the considered multiplex fractional-order neuronal networks, by means of which we obtain sufficient conditions for the interlayer and intralayer synchronizations that are in agreement with numerical results.

In this paper, we propose a novel scheme to learn high-level representative features and conduct classification for hyperspectral image (HSI) data in an automatic fashion. The proposed method is a collaboration of a wavelet-based extended morphological profile (WTEMP) and a deep autoencoder (DAE) (“WTEMP-DAE”), with the aim of exploiting the discriminative capability of DAE when using WTEMP features as the input. Each part of WTEMP-DAE is ingenious and contributes to the final classification performance. Specifically, in WTEMP-DAE, the spatial information is extracted from the WTEMP, which is then joined with the wavelet denoised spectral information to form the spectral-spatial description of HSI data. The obtained features are fed into DAE as the original input, where the good weights and bias of the network are initialized through unsupervised pre-training. Once the pre-training is completed, the reconstruction layers are discarded and a logistic regression (LR) layer is added to the top of the network to perform supervised fine-tuning and classification. Experimental results on two real HSI data sets demonstrate that the proposed strategy improves classification performance in comparison with other state-of-the-art hand-crafted feature extractors and their combinations.

Ion-channel current recordings based on an incubation type planar patch clamp were first reported in 2008, using HEK293 cells expressed with TRPV1 channels and capsaicin as ligand molecules. At first the success probability (number of devices which normally worked/total number of devices fabricated) was extremely low (several %). Several years later, we have succeeded in significantly decreasing the base line noise by using a salt-bridge-type Ag/AgCl electrode and successfully demonstrated the application of an incubation type planar patch clamp to ligand gated ion-channel biosensors and light gated ion-channel biosensors using HEK293 cells expressed with ChRWR. Furthermore, a spontaneous ion-channel current from a neuronal network was successfully observed by using a planar patch clamp chip, on which the neuronal network was occasionally formed with a soma of a neuron on a micro through-hole. Although the neuronal network was not controlled, this success shows the high potential of realizing a high throughput screening device on the basis of channel current measurements, which contain the most important information on network conditions.