Adaptive Network Model Research Articles

Sparse Neural Network Models of Antimicrobial Peptide-Activity Relationships.

We present an adaptive neural network model for chemical data classification. The method uses an evolutionary algorithm for optimizing the network structure by seeking sparsely connected architectures. The number of hidden layers, the number of neurons in each layer and their connectivity are free variables of the system. We used the method for predicting antimicrobial peptide activity from the amino acid sequence. Visualization of the evolved sparse network structures suggested a high charge density and a low aggregation potential in solution as beneficial for antimicrobial activity. However, different training data sets and peptide representations resulted in greatly varying network structures. Overall, the sparse network models turned out to be less accurate than fully-connected networks. In a prospective application, we synthesized and tested 10 de novo generated peptides that were predicted to either possess antimicrobial activity, or to be inactive. Two of the predicted antibacterial peptides showed cosiderable bacteriostatic effects against both Staphylococcus aureus and Escherichia coli. None of the predicted inactive peptides possessed antibacterial properties. Molecular dynamics simulations of selected peptide structures in water and TFE suggest a pronounced peptide helicity in a hydrophobic environment. The results of this study underscore the applicability of neural networks for guiding the computer-assisted design of new peptides with desired properties.

A Feedback-oriented Data Delay Modeling in a Dynamic Neural Network for Time Series Forecasting

In this study, we develop a neural network with a time shifting approach to forecast time series patterns. We investigate the impact of different layer-weight configurations to capture the trends in the forms of seasonal, chaotic, etc. We also hypothesize the combined effect of the delayed inputs and the forward connections to introduce a dynamical structure. The effect of overfitting issue is procedurally monitored to gain the resistance property from the early stoppage of training process and to reduce the predictions' error. Finally, the performance of the proposed network is challenged by six well-known deterministic and non-deterministic time series and compared by the autoregression (AR), artificial neural network (ANN), adaptive k-nearest neighbors (AKN), and adaptive neural network (ADNN) models. The results show that the proposed network outperforms the conventional models, particularly in forecasting the chaotic and seasonal time series.

Controller Integrity Monitoring in Adaptive Learning Systems Towards Trusted Autonomy

This paper presents a controller integrity monitoring method for a class of second-order nonlinear uncertain systems incorporating neural network-based adaptive control algorithms. The adaptive neural network model is employed to ensure robust tracking performance in the presence of certain modeling uncertainty under consideration during the offline controller design process. Based on Lyapunov stability analysis, an online controller integrity monitoring scheme is developed to detect the occurrence of controller software/algorithm faults and unanticipated physical component faults, which may lead to unstable learning behaviors and malfunctions of the adaptive controller. Adaptive thresholds for detecting controller malfunctions are derived, ensuring the robustness with respect to modeling uncertainty and neural network approximation error. Additionally, the detectability conditions are investigated, characterizing the class of detectable software faults and unanticipated physical faults. An upper bound on the fault detection time is also established. Simulation results are shown to illustrate the effectiveness of the proposed method.

The impact of socio-technical communication styles on the diversity and innovation potential of global science collaboratories

Emerging cyber-infrastructure tools are enabling scientists to transparently co-develop, share, and communicate about real-time diverse forms of knowledge artifacts. In these environments, communication preferences of scientists are posited as an important factor affecting innovation capacity and robustness of social and knowledge network structures. Scientific knowledge creation in such communities is called global participatory science (GPS). Recently, using agent-based modeling and collective action theory as a basis, a complex adaptive social communication network model (CollectiveInnoSim) is implemented. This work leverages CollectiveInnoSim implementing communication preferences of scientists. Social network metrics and knowledge production patterns are used as proxy metrics to infer innovation potential of emergent knowledge and collaboration networks. The objective is to present the underlying communication dynamics of GPS in a form of computational model and delineate the impacts of various communication preferences of scientists on innovation potential of the collaboration network. Gained insight can ultimately help policy-makers to design GPS environments and promote innovation.

A H-Infinity Control for Path Tracking with Fuzzy Hyperbolic Tangent Model

To achieve the goal of driver-less underground mining truck, a fuzzy hyperbolic tangent model is established for path tracking on an underground articulated mining truck. Firstly, the sample data of parameters are collected by the driver controlling articulated vehicle at a speed of 3 m/s, including both the lateral position deviation and the variation of heading angle deviation. Then, according to the improved adaptive BP neural network model and deriving formula of mediation rate of error estimator by the method of Cauchy robust, the weights are identified. Finally, H-infinity control controller is designed to control steering angle. The results of hardware-in-the-loop simulation show that lateral position deviation, heading angle deviation, and steering angle of the vehicle can be controlled, respectively, at 0.024 m, 0.08 rad, and 0.21 rad. All the deviations are asymptotically stable, and error control is in less than 2%. The method is demonstrated to be effective and reliable in path tracking for the underground vehicles.

Demography-based adaptive network model reproduces the spatial organization of human linguistic groups.

The distribution of human linguistic groups presents a number of interesting and nontrivial patterns. The distributions of the number of speakers per language and the area each group covers follow log-normal distributions, while population and area fulfill an allometric relationship. The topology of networks of spatial contacts between different linguistic groups has been recently characterized, showing atypical properties of the degree distribution and clustering, among others. Human demography, spatial conflicts, and the construction of networks of contacts between linguistic groups are mutually dependent processes. Here we introduce an adaptive network model that takes all of them into account and successfully reproduces, using only four model parameters, not only those features of linguistic groups already described in the literature, but also correlations between demographic and topological properties uncovered in this work. Besides their relevance when modeling and understanding processes related to human biogeography, our adaptive network model admits a number of generalizations that broaden its scope and make it suitable to represent interactions between agents based on population dynamics and competition for space.

From epidemics to information propagation: Striking differences in structurally similar adaptive network models

The continuous-time adaptive susceptible-infected-susceptible (ASIS) epidemic model and the adaptive information diffusion (AID) model are two adaptive spreading processes on networks, in which a link in the network changes depending on the infectious state of its end nodes, but in opposite ways: (i) In the ASIS model a link is removed between two nodes if exactly one of the nodes is infected to suppress the epidemic, while a link is created in the AID model to speed up the information diffusion; (ii) a link is created between two susceptible nodes in the ASIS model to strengthen the healthy part of the network, while a link is broken in the AID model due to the lack of interest in informationless nodes. The ASIS and AID models may be considered as first-order models for cascades in real-world networks. While the ASIS model has been exploited in the literature, we show that the AID model is realistic by obtaining a good fit with Facebook data. Contrary to the common belief and intuition for such similar models, we show that the ASIS and AID models exhibit different but not opposite properties. Most remarkably, a unique metastable state always exists in the ASIS model, while there an hourglass-shaped region of instability in the AID model. Moreover, the epidemic threshold is a linear function in the effective link-breaking rate in the AID model, while it is almost constant but noisy in the AID model.

Adaptive elastic networks as models of supercooled liquids.

The thermodynamics and dynamics of supercooled liquids correlate with their elasticity. In particular for covalent networks, the jump of specific heat is small and the liquid is strong near the threshold valence where the network acquires rigidity. By contrast, the jump of specific heat and the fragility are large away from this threshold valence. In a previous work [Proc. Natl. Acad. Sci. USA 110, 6307 (2013)], we could explain these behaviors by introducing a model of supercooled liquids in which local rearrangements interact via elasticity. However, in that model the disorder characterizing elasticity was frozen, whereas it is itself a dynamic variable in supercooled liquids. Here we study numerically and theoretically adaptive elastic network models where polydisperse springs can move on a lattice, thus allowing for the geometry of the elastic network to fluctuate and evolve with temperature. We show numerically that our previous results on the relationship between structure and thermodynamics hold in these models. We introduce an approximation where redundant constraints (highly coordinated regions where the frustration is large) are treated as an ideal gas, leading to analytical predictions that are accurate in the range of parameters relevant for real materials. Overall, these results lead to a description of supercooled liquids, in which the distance to the rigidity transition controls the number of directions in phase space that cost energy and the specific heat.

Allosteric transitions of ATP-binding cassette transporter MsbA studied by the adaptive anisotropic network model.

The transporter MsbA is a kind of multidrug resistance ATP-binding cassette transporter that can transport lipid A, lipopolysaccharides, and some amphipathic drugs from the cytoplasmic to the periplasmic side of the inner membrane. In this work, we explored the allosteric pathway of MsbA from the inward- to outward-facing states during the substrate transport process with the adaptive anisotropic network model. The results suggest that the allosteric transitions proceed in a coupled way. The large-scale closing motions of the nucleotide-binding domains occur first, accompanied with a twisting motion at the same time, which becomes more obvious in middle and later stages, especially for the later. This twisting motion plays an important role for the rearrangement of transmembrane helices and the opening of transmembrane domains on the periplasmic side that mainly take place in middle and later stages respectively. The topological structure plays an important role in the motion correlations above. The conformational changes of nucleotide-binding domains are propagated to the transmembrane domains via the intracellular helices IH1 and IH2. Additionally, the movement of the transmembrane domains proceeds in a nonrigid body, and the two monomers move in a symmetrical way, which is consistent with the symmetrical structure of MsbA. These results are helpful for understanding the transport mechanism of the ATP-binding cassette exporters.

Oscillating epidemics in a dynamic network model: stochastic and mean-field analysis.

An adaptive network model using SIS epidemic propagation with link-type-dependent link activation and deletion is considered. Bifurcation analysis of the pairwise ODE approximation and the network-based stochastic simulation is carried out, showing that three typical behaviours may occur; namely, oscillations can be observed besides disease-free or endemic steady states. The oscillatory behaviour in the stochastic simulations is studied using Fourier analysis, as well as through analysing the exact master equations of the stochastic model. By going beyond simply comparing simulation results to mean-field models, our approach yields deeper insights into the observed phenomena and help better understand and map out the limitations of mean-field models.

Wind power forecasting based on principle component phase space reconstruction

Forecasting of wind power is very important for both power grid and electricity market. Wind power forecasting based only on historical wind power data is carried out in this work. In a first treatment to the wind power data, Phase Space Reconstruction (PSR) is used to reconstruct the phase space of the wind dynamical system. Secondly, Principle Component Analysis (PCA) is used to minimize the influence from improper selection of the delay time and phase dimension. Finally, a prediction model, using Resource Allocating Network (RAN), is built for nonlinear mapping between the historical wind power data and the forecasting. Performance of the proposed method is compared with Persistence (PER), New-Reference (NR), and Adaptive Wavelet Neural Network (AWNN) models by using data from the US National Renewable Energy Laboratory (NREL). Analysis results indicate that the forecasting error of the proposed method is about 3% for 48 look-ahead hours, which is remarkably below the errors obtained with other forecast methods and has a probability close to 80% for 48 look-ahead hours forecasting within 12.5% error. The proposed method can also forecast wind power for turbines of different capacity and at different elevations below 10% error.

A Combined Adaptive Neural Network and Nonlinear Model Predictive Control for Multirate Networked Industrial Process Control

This paper investigates the multirate networked industrial process control problem in double-layer architecture. First, the output tracking problem for sampled-data nonlinear plant at device layer with sampling period T(d) is investigated using adaptive neural network (NN) control, and it is shown that the outputs of subsystems at device layer can track the decomposed setpoints. Then, the outputs and inputs of the device layer subsystems are sampled with sampling period T(u) at operation layer to form the index prediction, which is used to predict the overall performance index at lower frequency. Radial basis function NN is utilized as the prediction function due to its approximation ability. Then, considering the dynamics of the overall closed-loop system, nonlinear model predictive control method is proposed to guarantee the system stability and compensate the network-induced delays and packet dropouts. Finally, a continuous stirred tank reactor system is given in the simulation part to demonstrate the effectiveness of the proposed method.

Allosteric Transitions of Glutamine-Binding Protein Studied by the Elastic Network Model

Glutamine-Binding Protein (GlnBP) from Escherichia coli is an extremely important periplasmic binding protein. Binding of glutamine at the cleft between two domains causes a conformational change corresponding to a closure of two domains around the ligand. Because this process has a close relationship with protein function, studying this process has an important biological significance. It is very difficult for the existing experimental methods to obtain the conformational transition of GlnBP. In this paper, we studied the allosteric mechanism of GlnBP by using the adaptive Anisotropic Network Model (aANM) proposed by Bahar and obtained a relatively reasonable allosteric pathway. This work is helpful for the understanding the role of the topology structure in GlnBP conformational transition.

Adaptive structure radial basis function network model for processes with operating region migration

An adaptive structure radial basis function (RBF) network model is proposed in this paper to model nonlinear processes with operating point migration. The recursive orthogonal least squares algorithm (ROLS) is adopted to select new centers on-line, as well as to train the network weights. Based on the R matrix in the orthogonal decomposition, an initial center bank is formed and updated in each sample period. A new learning strategy is proposed to gain information from the new data for network structure adaptation. A center grouping algorithm is also developed to divide the centers into active and non-active groups, so that a structure with a smaller size is maintained in the final network model used for output prediction. The proposed RBF model is evaluated and compared with the three existing adaptive structure RBF networks by modeling a nonlinear time-varying numerical example. Simulation results demonstrate that the proposed algorithm has several advantages in term of the adaptive tracking ability and a better recovery speed over the existing methods during the migration of system׳s operating point.

A hybrid model for integrated day‐ahead electricity price and load forecasting in smart grid

Load and price forecasting are two key issues for market participants and system operators in electricity markets. Most existing works predict load and price separately. However, a dynamic pricing type scenario is envisioned in smart grid, where consumers may have the ability to react to constantly changing price signals by shifting electricity usages from expensive hours to others, which would consequently impact electricity prices. Thus, price and load signals are strongly coupled, and previous separated forecast models may not be as effective. In this study, a two-stage integrated price and load forecasting framework is developed. The first stage provides initial price and load forecasts separately, and the second stage considers load and price interaction with initial forecasts as inputs. At each stage, a hybrid time-series and adaptive wavelet neural network (AWNN) model is used, in which multivariate autoregressive integrated moving average catches the linear relationship of price and load log return series, generalised autoregressive conditional heteroscedastic unveils heteroscedastic character of residuals and AWNN presents non-linear impacts. Several criteria such as average mean absolute percentage error and error variance are used to measure the forecasting accuracy. Illustrative forecasting examples of the New York Independent System Operator market are presented to show the efficiency of the proposed method.

Multivariate adaptive regression splines and neural network models for prediction of pile drivability

Piles are long, slender structural elements used to transfer the loads from the superstructure through weak strata onto stiffer soils or rocks. For driven piles, the impact of the piling hammer induces compression and tension stresses in the piles. Hence, an important design consideration is to check that the strength of the pile is sufficient to resist the stresses caused by the impact of the pile hammer. Due to its complexity, pile drivability lacks a precise analytical solution with regard to the phenomena involved. In situations where measured data or numerical hypothetical results are available, neural networks stand out in mapping the nonlinear interactions and relationships between the system's predictors and dependent responses. In addition, unlike most computational tools, no mathematical relationship assumption between the dependent and independent variables has to be made. Nevertheless, neural networks have been criticized for their long trial-and-error training process since the optimal configuration is not known a priori. This paper investigates the use of a fairly simple nonparametric regression algorithm known as multivariate adaptive regression splines (MARS), as an alternative to neural networks, to approximate the relationship between the inputs and dependent response, and to mathematically interpret the relationship between the various parameters. In this paper, the Back propagation neural network (BPNN) and MARS models are developed for assessing pile drivability in relation to the prediction of the Maximum compressive stresses (MCS), Maximum tensile stresses (MTS), and Blow per foot (BPF). A database of more than four thousand piles is utilized for model development and comparative performance between BPNN and MARS predictions.

Reduction in impulse noise in digital images through a new adaptive artificial neural network model

In this paper, an adaptive artificial neural network model is developed in order to restore severely corrupted images. The proposed new and effective impulse noise reduction filter is named as adaptive neural network models with an algorithm based on artificial neural networks. Networks trained at different noise intensities get activated according to the intensity of the noise and estimate the most suitable neighboring pixel that can replace the corrupted pixel. The proposed algorithm reduces impulse noise effectively while also protecting the details. Experimental results show that the proposed algorithm performs better compared with other traditional filters.

Time Series Forecasting Using a Hybrid Adaptive Particle Swarm Optimization and Neural Network Model

AbstractFor time series forecasting, the problem that we often encounter is how to increase the prediction accuracy as much as possible with the irregular and noise data. This study proposes a novel multilayer feedforward neural network based on the improved particle swarm optimization with adaptive genetic operator (IPSO- MLFN). In the proposed IPSO, inertia weight is dynamically adjusted according to the feedback from particles’ best memories, and acceleration coefficients are controlled by a declining arccosine and an increasing arccosine function. Further, a crossover rate which only depends on generation and does not associate with the individual fitness is designed. Finally, the parameters of MLFN are optimized by IPSO. The empirical results on the container throughput forecast of Shenzhen Port show that forecasts with IPSO-MLFN model are more conservative and credible.

An Adaptive Fuzzy Wavelet Network with Gradient Learning for Nonlinear Function Approximation

AbstractIn this article, a new adaptive fuzzy wavelet neural network (AFWNN) model is proposed for nonlinear function approximation problems. The AFWNN model is based on the traditional Takagi-Sugeno-Kang fuzzy system. Specifically, this model replaces the membership functions of fuzzy rules with wavelet basis functions, which are known to have time and frequency localization properties, i.e., they can approximate patterns both in the time and frequency domains. The structure of the AFWNN model is derived from that of the adaptive neuro-fuzzy inference system (ANFIS). However, the AFWNN improves over the ANFIS by replacing Gaussian functions in the hidden layer with wavelet basis functions. The AFWNN model is trained using a gradient-based optimization algorithm. The AFWNN is then tested on three function approximation problems of time series prediction. For certain types of nonlinear time series, for instance fractal processes, the AFWNN is found to be substantially more accurate than alternative methods.

Modeling of technology adoption in Botswana

Purpose– The conceptualisation of technology adoption has largely been based on the Bass or some Bass-derived model – notably, the logistic model. Logistic-type models offer limited insights regarding the adoption process dynamics or the utility value of their results. The purpose of this paper is to outline an alternative technology adoption framework based on complex adaptive networks.Design/methodology/approach– An agent-based methodological approach is proposed. In it the actors, factors, goals, and adaptive learning influences driving solar energy technology adoption (SETA) process are first substantiated by empirical evidence gathered using field questionnaires and then incorporated in the simulation of a dynamic complex adaptive network of SETA. The complex adaptive network model is based on simple heuristic rules applied using a modified preferential attachment scheme within a NetLogo simulation environment.Findings– The interim results suggest an emergent network where prominent hub “driver” agents underlining the robustness of the model are statistically discernible.Research limitations/implications– The research is limited to solar photovoltaic and solar water heating technology adoption in Botswana households; however, its results are far-reaching.Practical implications– These results can be related to sustainable energy policy design. There, targeted incentive mechanisms can be formulated against the backdrop of the identified environmental factors and actors; the aim being to accelerate and cascade SETA.Social implications– The results could also be cascaded to other sectors and other non-solar technologies, thus providing a general alternative framework for enabling the widespread adoption of technologies.Originality/value– This research therefore represents a novel way of utilizing the new science of networks to accelerate SETA.