Adaptive Neural Network Structure Optimization Algorithm Based on Dynamic Nodes

Large-scale artificial neural networks have many redundant structures, making the network fall into the issue of local optimization and extended training time. Moreover, existing neural network topology optimization algorithms have the disadvantage of many calculations and complex network structure modeling. We propose a Dynamic Node-based neural network Structure optimization algorithm (DNS) to handle these issues. DNS consists of two steps: the generation step and the pruning step. In the generation step, the network generates hidden layers layer by layer until accuracy reaches the threshold. Then, the network uses a pruning algorithm based on Hebb’s rule or Pearson’s correlation for adaptation in the pruning step. In addition, we combine genetic algorithm to optimize DNS (GA-DNS). Experimental results show that compared with traditional neural network topology optimization algorithms, GA-DNS can generate neural networks with higher construction efficiency, lower structure complexity, and higher classification accuracy.     

Successful Reconstruction of a Physiological Circuit with Known Connectivity from Spiking Activity Alone

Identifying the structure and dynamics of synaptic interactions between neurons is the first step to understanding neural network dynamics. The presence of synaptic connections is traditionally inferred through the use of targeted stimulation and paired recordings or by post-hoc histology. More recently, causal network inference algorithms have been proposed to deduce connectivity directly from electrophysiological signals, such as extracellularly recorded spiking activity. Usually, these algorithms have not been validated on a neurophysiological data set for which the actual circuitry is known. Recent work has shown that traditional network inference algorithms based on linear models typically fail to identify the correct coupling of a small central pattern generating circuit in the stomatogastric ganglion of the crab Cancer borealis. In this work, we show that point process models of observed spike trains can guide inference of relative connectivity estimates that match the known physiological connectivity of the central pattern generator up to a choice of threshold. We elucidate the necessary steps to derive faithful connectivity estimates from a model that incorporates the spike train nature of the data. We then apply the model to measure changes in the effective connectivity pattern in response to two pharmacological interventions, which affect both intrinsic neural dynamics and synaptic transmission. Our results provide the first successful application of a network inference algorithm to a circuit for which the actual physiological synapses between neurons are known. The point process methodology presented here generalizes well to larger networks and can describe the statistics of neural populations. In general we show that advanced statistical models allow for the characterization of effective network structure, deciphering underlying network dynamics and estimating information-processing capabilities.     

Loss of self-inhibition is a cellular mechanism for episodic rhythmic behavior

BACKGROUND: Rhythmic motor behaviors can be generated continuously (e.g., breathing) or episodically (e.g., locomotion, swallowing), when short or long bouts of rhythmic activity are interspersed with periods of quiescence. Although the mechanisms of rhythm generation are known in detail in many systems, there is very little understanding of how the episodic nature of rhythmic behavior is produced at the neuronal level. RESULTS: Using a well-established episodic rhythm-generating neural circuit controlling molluscan feeding, we demonstrate that quiescence between bouts of activity arises from active, maintained inhibition of an otherwise rhythmically active network. We show that the source of the suppressive drive is within the circuit itself; a single central pattern generator (CPG) interneuron type that fires tonically to inhibit feeding during quiescence. Suppression of the tonic activity of this neuron by food is sufficient to change the network from an inactive to a rhythmically active state, with the cell switching function to fire phasically as part of the food-evoked rhythmogenesis. Furthermore, the absolute level of intrinsic suppressive control is modulated extrinsically by the animal's behavioral state (e.g., hunger/satiety), increasing the probability of episodes of feeding when the animal is hungry. CONCLUSIONS: By utilizing the same intrinsic member of a CPG network in both rhythm-generation and suppression, this system has developed a simple and efficient mechanism for generating a variable level of response to suit the animal's changing behavioral demands.     

Physical models of neural networks

The state of art in computer modelling of neural networks with associative memory is reviewed. The available experimental data are considered on learning and memory of small neural systems, on isolated synapses and on molecular level. Computer simulations demonstrate that realistic models of neural ensembles exhibit properties which can be interpreted as image recognition, categorization, learning, prototype forming, etc. A bilayer model of associative neural network is proposed. One layer corresponds to the short-term memory, the other one to the long-term memory. Patterns are stored in terms of the synaptic strength matrix. We have studied the relaxational dynamics of neurons firing and suppression within the short-term memory layer under the influence of the long-term memory layer. The interaction among the layers has found to create a number of novel stable states which are not the learning patterns. These synthetic patterns may consist of elements belonging to different non-intersecting learning patterns. Within the framework of a hypothesis of selective and definite coding of images in brain one can interpret the observed effect as the "idea? generating" process.     

Emergence of clusters in the hidden layer of a dynamic recurrent neural network

The neural integrator of the oculomotor system is a privileged field for artificial neural network simulation. In this paper, we were interested in an improvement of the biologically plausible features of the Arnold-Robinson network. This improvement was done by fixing the sign of the connection weights in the network (in order to respect the biological Dale's Law). We also introduced a notion of distance in the network in the form of transmission delays between its units. These modifications necessitated the introduction of a general supervisor in order to train the network to act as a leaky integrator. When examining the lateral connection weights of the hidden layer, the distribution of the weights values was found to exhibit a conspicuous structure: the high-value weights were grouped in what we call clusters. Other zones are quite flat and characterized by low-value weights. Clusters are defined as particular groups of adjoining neurons which have strong and privileged connections with another neighborhood of neurons. The clusters of the trained network are reminiscent of the small clusters or patches that have been found experimentally in the nucleus prepositus hypoglossi, where the neural integrator is located. A study was conducted to determine the conditions of emergence of these clusters in our network: they include the fixation of the weight sign, the introduction of a distance, and a convergence of the information from the hidden layer to the motoneurons. We conclude that this spontaneous emergence of clusters in artificial neural networks, performing a temporal integration, is due to computational constraints, with a restricted space of solutions. Thus, information processing could induce the emergence of iterated patterns in biological neural networks. Received: 18 September 1996 / Accepted in revised form: 7 January 1997     

Analyzing Self-Similar and Fractal Properties of the C. elegans Neural Network

The brain is one of the most studied and highly complex systems in the biological world. While much research has concentrated on studying the brain directly, our focus is the structure of the brain itself: at its core an interconnected network of nodes (neurons). A better understanding of the structural connectivity of the brain should elucidate some of its functional properties. In this paper we analyze the connectome of the nematode Caenorhabditis elegans. Consisting of only 302 neurons, it is one of the better-understood neural networks. Using a Laplacian Matrix of the 279-neuron “giant component” of the network, we use an eigenvalue counting function to look for fractal-like self similarity. This matrix representation is also used to plot visualizations of the neural network in eigenfunction coordinates. Small-world properties of the system are examined, including average path length and clustering coefficient. We test for localization of eigenfunctions, using graph energy and spacial variance on these functions. To better understand results, all calculations are also performed on random networks, branching trees, and known fractals, as well as fractals which have been “rewired” to have small-world properties. We propose algorithms for generating Laplacian matrices of each of these graphs.     

A neural network model of ventriloquism effect and aftereffect

Presenting simultaneous but spatially discrepant visual and auditory stimuli induces a perceptual translocation of the sound towards the visual input, the ventriloquism effect. General explanation is that vision tends to dominate over audition because of its higher spatial reliability. The underlying neural mechanisms remain unclear. We address this question via a biologically inspired neural network. The model contains two layers of unimodal visual and auditory neurons, with visual neurons having higher spatial resolution than auditory ones. Neurons within each layer communicate via lateral intra-layer synapses; neurons across layers are connected via inter-layer connections. The network accounts for the ventriloquism effect, ascribing it to a positive feedback between the visual and auditory neurons, triggered by residual auditory activity at the position of the visual stimulus. Main results are: i) the less localized stimulus is strongly biased toward the most localized stimulus and not vice versa; ii) amount of the ventriloquism effect changes with visual-auditory spatial disparity; iii) ventriloquism is a robust behavior of the network with respect to parameter value changes. Moreover, the model implements Hebbian rules for potentiation and depression of lateral synapses, to explain ventriloquism aftereffect (that is, the enduring sound shift after exposure to spatially disparate audio-visual stimuli). By adaptively changing the weights of lateral synapses during cross-modal stimulation, the model produces post-adaptive shifts of auditory localization that agree with in-vivo observations. The model demonstrates that two unimodal layers reciprocally interconnected may explain ventriloquism effect and aftereffect, even without the presence of any convergent multimodal area. The proposed study may provide advancement in understanding neural architecture and mechanisms at the basis of visual-auditory integration in the spatial realm.     

Full-state tracking control of a mobile robot using neural networks

In this paper a nonholonomic mobile robot with completely unknown dynamics is discussed. A mathematical model has been considered and an efficient neural network is developed, which ensures guaranteed tracking performance leading to stability of the system. The neural network assumes a single layer structure, by taking advantage of the robot regressor dynamics that expresses the highly nonlinear robot dynamics in a linear form in terms of the known and unknown robot dynamic parameters. No assumptions relating to the boundedness is placed on the unmodeled disturbances. It is capable of generating real-time smooth and continuous velocity control signals that drive the mobile robot to follow the desired trajectories. The proposed approach resolves speed jump problem existing in some previous tracking controllers. Further, this neural network does not require offline training procedures. Lyapunov theory has been used to prove system stability. The practicality and effectiveness of the proposed tracking controller are demonstrated by simulation and comparison results.     

A hierarchical neural network model for associative memory

A hierarchical neural network model with feedback interconnections, which has the function of associative memory and the ability to recognize patterns, is proposed. The model consists of a hierarchical multi-layered network to which efferent connections are added, so as to make positive feedback loops in pairs with afferent connections. The cell-layer at the initial stage of the network is the input layer which receives the stimulus input and at the same time works as an output layer for associative recall. The deepest layer is the output layer for pattern-recognition. Pattern-recognition is performed hierarchically by integrating information by converging afferent paths in the network. For the purpose of associative recall, the integrated information is again distributed to lower-order cells by diverging efferent paths. These two operations progress simultaneously in the network. If a fragment of a training pattern is presented to the network which has completed its self-organization, the entire pattern will gradually be recalled in the initial layer. If a stimulus consisting of a number of training patterns superposed is presented, one pattern gradually becomes predominant in the recalled output after competition between the patterns, and the others disappear. At about the same time when the recalled pattern reaches a steady state in he initial layer, in the deepest layer of the network, a response is elicited from the cell corresponding to the category of the finally-recalled pattern. Once a steady state has been reached, the response of the network is automatically extinguished by inhibitory signals from a steadiness-detecting cell. If the same stimulus is still presented after inhibition, a response for another pattern, formerly suppressed, will now appear, because the cells of the network have adaptation characteristics which makes the same response unlikely to recur. Since inhibition occurs repeatedly, the superposed input patterns are recalled one by one in turn.     

A template matching model for pattern recognition: Self-organization of templates and template matching by a disinhibitory neural network

A template matching model for pattern recognition is proposed. By following a previouslyproposed algorithm for synaptic modification (Hirai, 1980), the template of a stimulus pattern is selforganized as a spatial distribution pattern of matured synapses on the cells receiving modifiable synapses. Template matching is performed by the disinhibitory neural network cascaded beyond the neural layer composed of the cells receiving the modifiable synapses. The performance of the model has been simulated on a digital computer. After repetitive presentations of a stimulus pattern, a cell receiving the modifiable synapses comes to have the template of that pattern. And the cell in the latter layer of the disinhibitory bitory neural network that receives the disinhibitory input from that cell becomes electively sensitive to that pattern. Learning patterns are not restricted by previously learned ones. They can be subset or superset patterns of the ones previously learned. If an unknown pattern is presented to the model, no cell beyond the disinhibitory neural network will respond. However, if previously learned patterns are embedded in that pattern, the cells which have the templates of those patterns respond and are assumed to transmit the information to higher center. The computer simulation also shows that the model can organize a clean template under a noisy environment.     

A neural network model for the intersensory coordination involved in goal-directed movements

A neural network model for a sensorimotor system, which was developed to simulate oriented movements in man, is presented. It is composed of a formal neural network comprising two layers: a sensory layer receiving and processing sensory inputs, and a motor layer driving a simulated arm. The sensory layer is an extension of the topological network previously proposed by Kohonen (1984). Two kinds of sensory modality, proprioceptive and exteroceptive, are used to define the arm position. Each sensory cell receives proprioceptive inputs provided by each arm-joint together with the exteroceptive inputs. This sensory layer is therefore a kind of associative layer which integrates two separate sensory signals relating to movement coding. It is connected to the motor layer by means of adaptive synapses which provide a physical link between a motor activity and its sensory consequences. After a learning period, the spatial map which emerges in the sensory layer clearly depends on the sensory inputs and an associative map of both the arm and the extra-personal space is built up if proprioceptive and exteroceptive signals are processed together. The sensorimotor transformations occuring in the junctions linking the sensory and motor layers are organized in such a manner that the simulated arm becomes able to reach towards and track a target in extra-personal space. Proprioception serves to determine the final arm posture adopted and to correct the ongoing movement in cases where changes in the target location occur. With a view to developing a sensorimotor control system with more realistic salient features, a robotic model was coupled with the formal neural network. This robotic implementation of our model shows the capacity of formal neural networks to control the displacement of mechanical devices.     

A Canonical Circuit for Generating Phase-Amplitude Coupling

‘Phase amplitude coupling’ (PAC) in oscillatory neural activity describes a phenomenon whereby the amplitude of higher frequency activity is modulated by the phase of lower frequency activity. Such coupled oscillatory activity – also referred to as ‘cross-frequency coupling’ or ‘nested rhythms’ – has been shown to occur in a number of brain regions and at behaviorally relevant time points during cognitive tasks; this suggests functional relevance, but the circuit mechanisms of PAC generation remain unclear. In this paper we present a model of a canonical circuit for generating PAC activity, showing how interconnected excitatory and inhibitory neural populations can be periodically shifted in to and out of oscillatory firing patterns by afferent drive, hence generating higher frequency oscillations phase-locked to a lower frequency, oscillating input signal. Since many brain regions contain mutually connected excitatory-inhibitory populations receiving oscillatory input, the simplicity of the mechanism generating PAC in such networks may explain the ubiquity of PAC across diverse neural systems and behaviors. Analytic treatment of this circuit as a nonlinear dynamical system demonstrates how connection strengths and inputs to the populations can be varied in order to change the extent and nature of PAC activity, importantly which phase of the lower frequency rhythm the higher frequency activity is locked to. Consequently, this model can inform attempts to associate distinct types of PAC with different network topologies and physiologies in real data.     

基于BP神经网络与ETM+遥感数据的盐城滨海自然湿地覆被分类

高效而精确的湿地遥感分类是大范围湿地资源动态监测与管理的必要保障。本研究使用ETM 遥感数据,借助Matlab神经网络工具箱,构建了基于BP神经网络的滨海湿地覆被分类模型,并将其应用于江苏盐城沿海湿地珍禽国家级自然保护区的核心区的自然湿地覆被分类研究中。本研究选择3、4、7、8波段作为输入层变量,单隐藏层设为10个节点,输出层变量对应待划分的8种覆被类型,构建三层式BP神经网络滨海湿地覆被分类模型。结果显示,BP分类总精度为85.91%,Kappa系数为0.8328,与最小距离法和极大似然法的分类总精度相比,分别提高了7.99%和6.08%,Kappa系数也相比提高。研究结果表明,BP神经网络分类法是一种较为有效的湿地遥感影像分类技术,能够提高分类精度。     

Synchronization without oscillatory neurons

Experimental results suggest that neurons in the cortex synchronize their action potentials on the millisecond time scale. More importantly this binding expresses functional relationships between the neurons. A model of neuronal interactions is proposed in which simultaneous discharges of neurons develop through specialized synaptic circuits. As an important prerequisite for this synchronization it is demonstrated that SynFire chains, generating different levels of excitation, propagate their activity waves at distinct velocities. Two chains were coupled by excitatory synapses and their activity was initiated at different times. Due to synaptic interactions, activity in the earlier-initiated chain accelerates propagation in the other chain until the two activity waves are synchronized. Compared with several neural network models with oscillatory units, physiologically more plausible neurons are simulated. It is still under debate whether neurons in the cortex show oscillatory dischargesper se. In particular, a high rate of noise relative to very weak synaptic gains cannot impair our results in the neural network simulations.     

Discovering molecular features of intrinsically disordered regions by using evolution for contrastive learning

A major challenge to the characterization of intrinsically disordered regions (IDRs), which are widespread in the proteome, but relatively poorly understood, is the identification of molecular features that mediate functions of these regions, such as short motifs, amino acid repeats and physicochemical properties. Here, we introduce a proteome-scale feature discovery approach for IDRs. Our approach, which we call “reverse homology”, exploits the principle that important functional features are conserved over evolution. We use this as a contrastive learning signal for deep learning: given a set of homologous IDRs, the neural network has to correctly choose a held-out homolog from another set of IDRs sampled randomly from the proteome. We pair reverse homology with a simple architecture and standard interpretation techniques, and show that the network learns conserved features of IDRs that can be interpreted as motifs, repeats, or bulk features like charge or amino acid propensities. We also show that our model can be used to produce visualizations of what residues and regions are most important to IDR function, generating hypotheses for uncharacterized IDRs. Our results suggest that feature discovery using unsupervised neural networks is a promising avenue to gain systematic insight into poorly understood protein sequences.     

Motor pattern selection by combinatorial code of interneuronal pathways

We use a modeling approach to examine ideas derived from physiological network analyses, pertaining to the switch of a motor control network between two opposite control modes. We studied the femur–tibia joint control system of the insect leg, and its switch between resistance reflex in posture control and “active reaction” in walking, both elicited by the same sensory input. The femur–tibia network was modeled by fitting the responses of model neurons to those obtained in animals. The strengths of 16 interneuronal pathways that integrate sensory input were then assigned three different values and varied independently, generating a database of more than 43 million network variants. We demonstrate that the same neural network can produce the two different behaviors, depending on the combinatorial code of interneuronal pathways. That is, a switch between behaviors, such as standing to walking, can be brought about by altering the strengths of selected sensory integration pathways. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

LaDIVA: A neurocomputational model providing laryngeal motor control for speech acquisition and production

Many voice disorders are the result of intricate neural and/or biomechanical impairments that are poorly understood. The limited knowledge of their etiological and pathophysiological mechanisms hampers effective clinical management. Behavioral studies have been used concurrently with computational models to better understand typical and pathological laryngeal motor control. Thus far, however, a unified computational framework that quantitatively integrates physiologically relevant models of phonation with the neural control of speech has not been developed. Here, we introduce LaDIVA, a novel neurocomputational model with physiologically based laryngeal motor control. We combined the DIVA model (an established neural network model of speech motor control) with the extended body-cover model (a physics-based vocal fold model). The resulting integrated model, LaDIVA, was validated by comparing its model simulations with behavioral responses to perturbations of auditory vocal fundamental frequency (f_o) feedback in adults with typical speech. LaDIVA demonstrated capability to simulate different modes of laryngeal motor control, ranging from short-term (i.e., reflexive) and long-term (i.e., adaptive) auditory feedback paradigms, to generating prosodic contours in speech. Simulations showed that LaDIVA’s laryngeal motor control displays properties of motor equivalence, i.e., LaDIVA could robustly generate compensatory responses to reflexive vocal f_o perturbations with varying initial laryngeal muscle activation levels leading to the same output. The model can also generate prosodic contours for studying laryngeal motor control in running speech. LaDIVA can expand the understanding of the physiology of human phonation to enable, for the first time, the investigation of causal effects of neural motor control in the fine structure of the vocal signal.     

K-way neural network graph partitioning with separator vertices

A neural network for partitioning graphs into any number of subgraphs using a k-way procedure is presented. The resulting neural network optimises all subgraphs so that they contain approximately the same number of vertices with the exception of a `separator' subgraph. The latter separates all the other subgraphs and is optimised to contain a minimum number of vertices. Expressions for the neuron link weights for such a network are derived analytically, and the recall mechanism of the mean field theorem neural network is used to obtain the graph partitioning. Applications focus on partitioning graphs associated with finite element meshes which, together with parallel domain decomposition solution methods, provide the motivation for this work. Received: 14 January 1998 / Accepted in revised form: 24 November 1998     

A neural network model for limb trajectory formation

This paper deals with the problem of representing and generating unconstrained aiming movements of a limb by means of a neural network architecture. The network produced time trajectories of a limb from a starting posture toward targets specified by sensory stimuli. Thus the network performed a sensory-motor transformation. The experimenters trained the network using a bell-shaped velocity profile on the trajectories. This type of profile is characteristic of most movements performed by biological systems. We investigated the generalization capabilities of the network as well as its internal organization. Experiments performed during learning and on the trained network showed that: (i) the task could be learned by a three-layer sequential network; (ii) the network successfully generalized in trajectory space and adjusted the velocity profiles properly; (iii) the same task could not be learned by a linear network; (iv) after learning, the internal connections became organized into inhibitory and excitatory zones and encoded the main features of the training set; (v) the model was robust to noise on the input signals; (vi) the network exhibited attractor-dynamics properties; (vii) the network was able to solve the motorequivalence problem. A key feature of this work is the fact that the neural network was coupled to a mechanical model of a limb in which muscles are represented as springs. With this representation the model solved the problem of motor redundancy.