Research on the anti-interference characteristics of neural networks with different scales

Being difficult to attain the precise mathematical models, traditional control methods such as proportional integral (PI) and proportional integral differentiation (PID) cannot meet the demands for real time and robustness when applied in some nonlinear systems. The neural network controller is a good replacement to overcome these shortcomings. However, the performance of neural network controller is directly determined by neural network model. In this paper, a new neural network model is constructed with a structure topology between the regular and random connection modes based on complex network, which simulates the brain neural network as far as possible, to design a better neural network controller. Then, a new controller is designed under small-world neural network model and is investigated in both linear and nonlinear systems control. The simulation results show that the new controller basing on small-world network model can improve the control precision by 30% in the case of system with random disturbance. Besides the good performance of the new controller in tracking square wave signals, which is demonstrated by the experiment results of direct drive electro-hydraulic actuation position control system, it works well on anti-interference performance.

We study the evolutionary snowdrift game in a heterogeneous Newman–Watts small-world network. The heterogeneity of the network is controlled by the number of hubs. It is found that the moderate heterogeneity of the network can promote the cooperation best. Besides, we study how the hubs affect the evolution of cooperative behaviours of the heterogeneous Newman–Watts small-world network. Simulation results show that both the initial states of hubs and the connections between hubs can play an important role. Our work gives a further insight into the effect of hubs on the heterogeneous networks.

Quantification of degeneracy in Hodgkin–Huxley neurons on Newman–Watts small world network

Small-world networks have the higher clustering coefficient and shorter average path length. According to the design requirements of topology and routing algorithm of the WSN, we apply small world theory into the WSN, and propose the routing algorithm based on Newman Watts small-world network model. This algorithm judge the cluster number whether same to decide the communication type. Data is transmitted to super node firstly, and then the packets are sent by the shortest transmission paths which get from the super node ring. Experiments show that the routing algorithm improves the network throughput and network transmission efficiency, the common node energy consumption become small, so the service life of wireless sensor network is prolonged.

Stochastic resonance on Newman–Watts networks of Hodgkin–Huxley neurons with local periodic driving

The synapses and network topology enhance neural synchronization and anti-interference, enabling the bio-inspired brain model to mimic biological noise resilience effectively. This study numerically simulates the effects of synapses and network topology on the synchronous discharge and anti-interference of neuronal networks. The Hodgkin-Huxley neuron model, the electrical synapses (ES), the Hansel chemical synapse (HS), and the Rabinovich chemical synapse (RS) were used to construct the neural networks with the ring structure and the Newman-Watts (NW) small-world topology. The sine wave and the sine wave with superimposed Gaussian white noise interference were selected as the stimulation signals. The MATLAB and Simulink platforms were employed to implement the numerical simulation. For the ring network with the sine wave stimulation, the correlation coefficients of one set of neuron pairs (neuron 1 and neuron 25) were 0.292 (ES), 0.236 (HS), and 0.168 (RS), respectively. However, after superimposed interference, the correlation coefficients become 0.099, 0.086, and 0.379, respectively. For the NW small-world topology with sinusoidal stimulation, the correlation coefficients of the same neuron pair were 0.569 (ES), 0.563 (HS), and 0.969 (RS), respectively. The correlation coefficients after superposition interference become 0.569, 0.163, and 0.88, respectively. The HS-coupled network exhibits severe signal latency (Ring network: Latency >200 ms, NW small-world network: Latency >150 ms). While RS-coupled network demonstrates dramatically reduced delays (<50 ms) across both topologies. The results suggest that the synchronization of the RS coupling network is much better than that of both ES and HS coupling networks. Ring networks coupled via HS demonstrate performance metrics comparable to those of ES-coupled ring networks, albeit with significant action potential propagation delays observed in both configurations. The NW small-world network can reduce the delay of signal transmission in the network by increasing the number of pathways. As network topological complexity increases, distal neurons demonstrate reduced spike timing variability and enhanced firing synchrony, collectively improving interference suppression efficacy.

The scientific researches are focused on network topologies and training algorithms fields because they reduce overfitting problem in artificial neural networks. In this context, we showed in our previous work that Newman–Watts small-world feed-forward artificial neural networks present better classification and prediction performance than conventional feed-forward artificial neural networks. In this study, we investigate the effects of the Resilient back-propagation algorithm on SW network topology and propose a Resilient Newman–Watts small-world feed-forward artificial neural network model by assuming fixed initial topological conditions. We find that Resilient small-world network further reduces overfitting and further increases the network performance when compared to the conventional feed-forward artificial neural networks. Furthermore, it is shown that the proposed network model does not increase the algorithmic complexity as per other models. The obtained results imply that the proposed model can contribute to the solving of overfitting problem encountered in both deep neural networks and conventional artificial neural networks.

The nature of associations between variables is important for constructing theory about psychological phenomena. In the last decade, this topic has received renewed interest with the introduction of psychometric network models. In psychology, network models are often contrasted with latent variable (e.g., factor) models. Recent research has shown that differences between the two tend to be more substantive than statistical. One recently developed algorithm called the Loadings Comparison Test (LCT) was developed to predict whether data were generated from a factor or small-world network model. A significant limitation of the current LCT implementation is that it's based on heuristics that were derived from descriptive statistics. In the present study, we used artificial neural networks to replace these heuristics and develop a more robust and generalizable algorithm. We performed a Monte Carlo simulation study that compared neural networks to the original LCT algorithm as well as logistic regression models that were trained on the same data. We found that the neural networks performed as well as or better than both methods for predicting whether data were generated from a factor, small-world network, or random network model. Although the neural networks were trained on small-world networks, we show that they can reliably predict the data-generating model of random networks, demonstrating generalizability beyond the trained data. We echo the call for more formal theories about the relations between variables and discuss the role of the LCT in this process.

Neural network techniques have been widely adopted for various tasks, including sequence time-series prediction. Researchers tend to develop large-scale neural network models with multi-level frameworks and extensive numbers of parameters. However, this strategy might not be appropriate for the dynamic response prediction (DRP) problem because large-scale neural network models may tend to overfit the data rather than learn the system’s inherent dynamic property. This paper advocates that neural network models with concise architectures and the ability to make step-by-step moving-forward predictions are better suited for DRP problems. Building on this opinion, the Elman recurrent, long short-term memory, discrete-time state-space, and autoregressive neural networks are adopted in this paper to construct surrogate models for DRP tasks. Numerical and experimental evaluations of these models and other large-scale neural network models are conducted, demonstrating that the four selected types of neural network models, built with limited numbers of parameters, could achieve good performance and minimal overfitting, with overall R-squared values exceeding 90%. It is recommended to utilize these small-scale neural networks or their modifications for DRP problems rather than to construct complicated and large-scale neural networks.

Numerous real-world systems can be described by complex network models, and now the phenomenon of small-world is very common and widespread in many real networks. Over the last several years, instant messaging chat has been developing in a strikingly fast pace, which has the features different from those of the two classical small-world networks, WS and NW network models. To better describe the instant messaging chat network, we here present an algorithm for building a new small-world network model by combining the two classic small-world networks. Then the properties of the new small-world network are illustrated. By showing some numerical simulation results and comparing the new model with WS and NW small-world networks, it is demonstrated that our new small-world network model may get closer to the real instant messaging chat network.

基于网络效能分析的生境网络构建与优化——以苏锡常地区白鹭为例

Synchronizability of multi-layer small-world dynamical networks

Effects of autapse and ion channel block on the collective firing activity of Newman–Watts small-world neuronal networks

We investigate the evolutionary prisoner's dilemma game (PDG) on weighted Newman–Watts (NW) networks. In weighted NW networks, the link weight wij is assigned to the link between the nodes i and j as: wij = (κi · κj)β, where κi(κj) is the degree of node i(j) and β represents the strength of the correlations. Obviously, the link weight can be tuned by only one parameter β. We focus on the cooperative behavior and wealth distribution in the system. Simulation results show that the cooperator frequency is promoted by a large range of β and there is a minimal cooperation frequency around β = – 1. Moreover, we also employ the Gini coefficient to study the wealth distribution in the population. Numerical results show that the Gini coefficient reaches its minimum when β ≈ – 1. Our work may be helpful in understanding the emergence of cooperation and unequal wealth distribution in society.

We investigate the prisoner's dilemma game based on a new rule: players will change their current strategies to opposite strategies with some probability if their neighbours' average payoffs are higher than theirs. Compared with the cases on regular lattices (RL) and Newman–Watts small world network (NW), cooperation can be best enhanced on the scale-free Barabási–Albert network (BA). It is found that cooperators are dispersive on RL network, which is different from previously reported results that cooperators will form large clusters to resist the invasion of defectors. Cooperative behaviours on the BA network are discussed in detail. It is found that large-degree individuals have lower cooperation level and gain higher average payoffs than that of small-degree individuals. In addition, we find that small-degree individuals more frequently change strategies than do large-degree individuals.