Search

The profile of urban microclimates is important in many engineering fields, such as occupant's thermal comfort and health, and other building engineering. To predict the profile of urban microclimate, this study applies the artificial neural network and long short-term memory network predictive models, and an urban microclimate dataset was obtained with a long-term monitoring from year 2017 to 2019 with 5-min resolution including temperature, relative humidity, and solar radiation. Two predictive models were applied, and the first (Model 1) is to apply the predictive techniques to predict the urban microclimate in the real-time sequence, and then extract the characteristics of urban microclimate, while the second (Model 2) is to directly extract the characteristics of the microclimate, and then predict the characteristics of the microclimate. Backpropagation artificial neural network (BP-ANN) and long-short term memory (LSTM) techniques were applied in both models. The results show Model 1 with as the time-series prediction can reach the best (99.92%) of correlation coefficient and 98% of the mean average percentage error (MAPE), for temperature, while 99.66% and 98.18% for relative humidity, respectively, while accuracies in Model 2 decreased to 79% and 88.6% of MAPE for temperature and relative humidity, respectively. The prediction of solar radiation using ANN and LSTM are 51.1% and 57.8% of the correlation coefficient, respectively. [ABSTRACT FROM AUTHOR]

This paper focused on the problem of diagnosis of Alzheimer's disease via the combination of deep learning and radiomics methods. We proposed a classification model for Alzheimer's disease diagnosis based on improved convolution neural network models and image fusion method and compared it with existing network models. We collected 182 patients in the ADNI and PPMI database to classify Alzheimer's disease, and reached 0.906 AUC in training with single modality images, and 0.941 AUC in training with fusion images. This proved the proposed method has better performance in the fusion images. The research may promote the application of multimodal images in the diagnosis of Alzheimer's disease. Fusion images dataset based on multi-modality images has higher diagnosis accuracy than single modality images dataset. Deep learning methods and radiomics significantly improve the diagnosing accuracy of Alzheimer's disease diagnosis. [ABSTRACT FROM AUTHOR]

Financial bankruptcy prediction is crucial for financial institutions in assessing the financial health of companies and individuals. Such work is necessary for financial institutions to establish effective prediction models to make appropriate lending decisions. In recent decades, various bankruptcy prediction models have been developed for academics and practitioners to predict the likelihood that a loan customer will go bankrupt. Among them, Artificial Neural Networks (ANNs) have been widely and effectively applied in bankruptcy prediction. Inspired by the mechanism of biological neurons, we propose an evolutionary pruning neural network (EPNN) model to conduct financial bankruptcy analysis. The EPNN possesses a dynamic dendritic structure that is trained by a global optimization learning algorithm: the Adaptive Differential Evolution algorithm with Optional External Archive (JADE). The EPNN can reduce the computational complexity by removing the superfluous and ineffective synapses and dendrites in the structure and is simultaneously able to achieve a competitive classification accuracy. After simplifying the structure, the EPNN can be entirely replaced by a logic circuit containing the comparators and the logic NOT, AND, and OR gates. This mechanism makes it feasible to apply the EPNN to bankruptcy analysis in hardware implementations. To verify the effectiveness of the EPNN, we adopt two benchmark datasets in our experiments. The experimental results reveal that the EPNN outperforms the Multilayer Perceptron (MLP) model and our previously developed preliminary pruning neural network (PNN) model in terms of accuracy, convergence speed, and Area Under the Receiver Operating Characteristics (ROC) curve (AUC). In addition, the EPNN also provides competitive and satisfactory classification performances in contrast with other commonly used classification methods. [ABSTRACT FROM AUTHOR]

With the number of deaths due to liver diseases increasing steadily in recent years, early detection and treatment of these diseases has been one of the most active research fields using computational intelligence techniques. In this paper, we propose a more realistic single-neuron model with synaptic nonlinearities in a dendritic tree for liver disorder diagnosis. The computation on the neuron is performed as a combination of dimensional reduction and nonlinearity, which has a neuron-pruning function that can remove useless synapses and dendrites during learning, forming a distinct synaptic and dendritic morphology. The nonlinear interactions in a dendrite tree are expressed using the Boolean logic AND (conjunction), OR (disjunction), and NOT (negation), which can be easily implemented in hardware. Furthermore, an error backpropagation (BP) algorithm is used to train the neuron model, and the performance is compared with a traditional BP neural network in terms of accuracy, sensitivity, and specificity. We use the BUPA liver disorder datasets obtained from the UCI Machine Learning Repository to verify the proposed method. Simulation results show promise for the use of this single-neuron model as an effective pattern classification method in liver disorder diagnostics. © 2016 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc. [ABSTRACT FROM AUTHOR]

As a novel optimization technique, chaos has gained much attention and some applications during the past decade. For a given energy or cost function, by following chaotic ergodic orbits, a chaotic dynamic system may eventually reach the global optimum or its good approximation with high probability. To enhance the performance of the local search method (LS), which is based on the generalized reduced gradient algorithm, hybrid local search method is proposed by incorporating chaos. Thus, LS and chaos are hybridized to form a chaotic dynamic local search method (CDLS), which reasonably combines the searching ability of LS and chaotic searching behavior. In this paper, a CDLS method based on the logistic equation is presented to learn Multiple-Valued Logic (MVL) Networks. Simulation results and comparisons with the traditional back propagation algorithm (BP) and the standard LS method show that the CDLS can effectively enhance the searching efficiency and greatly improve the searching quality within reasonable number of iterations. [ABSTRACT FROM AUTHOR]

Recently, biological superior information processing ability has been researched to be able to be useful for a computer system. In particular, neural networks for the brain and nervous system have been researched. Immunity network models based on the biological immune response network also have been studied, but most of the researchers consider only a basic part of the interaction between B cells and T cells. Recent research has reported that the function of T cell receptors (TCR) is important in the practical immune response network. Therefore, we propose a new immunity network model which, unlike the traditional immunity networks, has the same function as that of a TCR. We applied the proposed immunity network model and the traditional immunity network model to the pattern recognition system. And we compared the performance of the proposed immunity network model with the traditional immunity network model, and showed the usability of the proposed immunity network model. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(10): 41–48, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10040 [ABSTRACT FROM AUTHOR]

The complex-valued backpropagation algorithm has been widely used in fields of dealing with telecommunications, speech recognition and image processing with Fourier transformation. However, the local minima problem usually occurs in the process of learning. To solve this problem and to speed up the learning process, we propose a modified error function by adding a term to the conventional error function, which is corresponding to the hidden layer error. The simulation results show that the proposed algorithm is capable of preventing the learning from sticking into the local minima and of speeding up the learning. [ABSTRACT FROM AUTHOR]

Abstract: In this article, we propose a method for improving the transiently chaotic neural network (TCNN) by introducing several time-dependent parameters. This method allows the network to have rich chaotic dynamics in its initial stage and to reach a state in which all neurons are stable soon after the last bifurcation. This enables the network to have rich search ability initially and to use less CPU time to reach a stable state. The simulation results on the N-queen problem confirm that this method effectively improves both the solution quality and convergence speed of TCNN. [Copyright &y& Elsevier]

Abstract: In this paper, we propose a discrete competitive Hopfield neural network (DCHNN) for the cellular channel assignment problem (CAP). The DCHNN can always satisfy the problem constraint and therefore guarantee the feasibility of the solutions for the CAP. Furthermore, the DCHNN permits temporary energy increases to escape from local minima by introducing stochastic dynamics. Simulation results show that the DCHNN has superior ability for the CAP within reasonable number of iterations. [Copyright &y& Elsevier]

In this paper, by adding a nonlinear self-feedback to the maximum neural network, we propose a parallel algorithm for the maximum clique problem that introduces richer and more flexible dynamics and can prevent the network from getting stuck at local minima. A large number of instances have been simulated to verify the proposed algorithm. [Copyright &y& Elsevier]

The minimum vertex cover problem is a classic graph optimization problem. It is well known that it is an NP-complete problem. In this paper, a new motion equation with a special term of the Hopfield neural network is presented for the minimum vertex cover problem. Extensive simulations are performed, and the simulation results show that the Hopfield neural network with the new motion equation works much better than the original Hopfield network and a sequential ratio-2 algorithm on random graphs, and has higher convergence rate to optimal solutions for benchmark graphs. [Copyright &y& Elsevier]

A method of supervised learning for multilayer artificial neural networks to escape local minima is proposed. The learning model has two phases: a backpropagation phase and a gradient ascent phase. The backpropagation phase performs steepest descent on a surface in weight space whose height at any point in weight space is equal to an error measure, and it finds a set of weights minimizing this error measure. When the backpropagation gets stuck in local minima, the gradient ascent phase attempts to fill up the valley by modifying gain parameters in a gradient ascent direction of the error measure. The two phases are repeated until the network gets out of local minima. The algorithm has been tested on benchmark problems, such as exclusive-or (XOR), parity, alphabetic characters learning, Arabic numerals with a noise recognition problem, and a realistic real-world problem: classification of radar returns from the ionosphere. For all of these problems, the systems are shown to be capable of escaping from the backpropagation local minima and converge faster when using the new proposed method than using the simulated annealing techniques. [ABSTRACT FROM AUTHOR]

In this letter, we utilize the Hopfield network learning method to adjust the balance between constraint term and cost term of the energy function so that the local minimum that the network once falls into vanishes and the network can continue updating in a gradient descent direction of energy. We applied the proposed learning method to the traveling salesman problem to show that the method is capable of finding an optimal solution or a near-optimal solution in a shorter time. [Copyright &y& Elsevier]

In this paper, we propose a hill-climbing learning method for Hopfield networks in which the energy of the network is intentionally raised in the weight space so that the network can escape from local minima. This learning method involves repeated updating of the Hopfield network in state space and modification of the weights in weight space if it settles into a local minimum, so that the energy is raised. Shifting of the global minimum during learning can be avoided by updating the state on the unlearned Hopfield network in state space, using equilibrium states obtained from learning as the initial state. Simulations using this learning method show that the Hopfield network can escape from a local minimum which depends on the initial state. A simulation experiment is first conducted with a two-variable Hopfield network, and the convergence of learning and the change of the domain from a local minimum to a global minimum are demonstrated. Then, using an example of the traveling-salesman problem, a simulation is performed using concrete numerical values, and the validity of the method is demonstrated. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 84(7): 28–40, 2001 [ABSTRACT FROM AUTHOR]

• A super-resolution (SR) imaging model of electrical capacitance tomography is proposed. • It consists of a low-resolution (LR) imaging network and a high-resolution (HR) imaging network. • The imaging performance, construction efficiency, and generalization ability of the trained SR imaging model are evaluated. • The particle concentration distribution on a fluidized bed is measured with the trained SR imaging model. Electrical capacitance tomography (ECT) has been widely used to investigate the flow dynamics of the gas–solid fluidized bed. Deep learning-based tomographic imaging of ECT is of great potential to reconstruct the particle concentration distribution with compatible image quality and reconstruction speed. However, it suffers from low model construction efficiency as numerous data samples need to be collected for training the deep neural network. Here a super-resolution (SR) imaging model based on a two-stage imaging network is proposed to improve the model construction efficiency. A low-resolution (LR) imaging network is used to reconstruct the LR image with the measured capacitance matrix while a subsequent high-resolution (HR) network calculates the HR image with the reconstructed LR image. The imaging performance, construction efficiency, and generalization ability of the trained SR imaging model are evaluated by reconstructing the samples in the test set and the unseen typical flow patterns. Considering the training set includes 100,000 data samples, the time consumption of numerically collecting the dataset for training the SR imaging model is 2.70 h, 2.2 % of that for a conventional deep learning-based imaging model of ECT, and the overall model construction time is only 4.44 h when the grid numbers of the LR and HR images are 246 and 8053, respectively. Experimental measurements are further carried out on a fluidized bed and the preliminary results show good prediction and generality of the trained model for reconstructing the particle concentration distribution. [ABSTRACT FROM AUTHOR]

Abstract: In this paper, we propose a new rewiring rule that generates small-world networks with larger clustering coefficient and smaller average path length. Unlike the random rewiring rule in the WS model described by Watts and Strogatz, we use the “rich-gets-richer” rule that links a vertex that already has a large number of connections and has a higher probability. Simulation results also verify that the novel “rich-gets-richer” rule based small-world network is an improvement over the WS model. [Copyright &y& Elsevier]

Abstract: This paper presents an improved adaptive neuro-fuzzy inference system (ANFIS) for the application of time-series prediction. Because ANFIS is based on a feedforward network structure, it is limited to static problem and cannot effectively cope with dynamic properties such as the time-series data. To overcome this problem, an improved version of ANFIS is proposed by introducing self-feedback connections that model the temporal dependence. A batch type local search is suggested to train the proposed system. The effectiveness of the presented system is tested by using three benchmark time-series examples and comparison with the various models in time-series prediction is also shown. The results obtained from the simulation show an improved performance. [Copyright &y& Elsevier]

Abstract: In this paper, coefficient sensitiveness of transiently chaotic neural network (TCNN) for maximum clique problem (MCP) is analyzed. By introducing a new definition of adaptivity into TCNN and utilizing a neuron filter, the domain of coefficient selection is enlarged to search globally optimal solution and near-optimal solution on MCP. Based on our analysis on the characteristic of TCNN, we propose a TCNN filter algorithm to shrink the network scale and improve solution quality. The algorithm effectively relaxes the coefficient sensitiveness and improves network ability to solve MCP. Simulations have been performed to verify the validity of our algorithm. [Copyright &y& Elsevier]

Abstract: In this article, we propose a new elastic net method to solve the traveling salesman problem by introducing some time-dependent parameters. This method can help network neurons move quickly near to the cities at early stage, and gradually increase the strength that pulls neurons towards their neighbours on the path to minimize the total path length. This enables the network to have superior ability of searching for cities, and converge sooner to a saturated state. Simulation results illustrate that the proposed network performs better than the classical elastic net for optimization both in solution quality and convergence speed. [Copyright &y& Elsevier]

Abstract: We propose a Lagrangian object relaxation technique that can obtain a more near-optimal solution for the traveling salesman problem (TSP). It consists of two stages. First, a feasible solution is calculated and second, a more near-optimal solution is calculated by a Hopfield neural network (HNN). The Lagrangian object relaxation technique can help the HNN escape from the local minimum by correcting Lagrangian multipliers. The Lagrangian object relaxation neural network is analyzed theoretically and evaluated experimentally through simulating the TSP. The simulation results based on some TSPLIB benchmark problems show that the proposed method can find 100% valid solutions which are near-optimal solutions. [Copyright &y& Elsevier]