Search

This paper addresses road anomaly detection by formulating it as a classification problem and applying deep learning approaches to solve it. Besides conventional road anomalies, additional ones are introduced from the perspective of a vehicle. In order to facilitate the learning process, the paper pays a close attention to pattern representation, and proposes three sets of numeric features for representing road conditions. Also, three deep learning approaches, i.e. Deep Feedforward Network (DFN), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN), are considered to tackle the classification problem. The detectors, with respect to the three deep learning approaches, are trained and evaluated through data collected from a test vehicle driven on various road anomaly conditions. The comparison study on the detection performances is conducted by setting key hyper-parameters to certain sets of fixed values. Also, the comparison study on performances of each detector with respect to different pattern representations is conducted. The results have shown the effectiveness of the proposed approaches and the efficiency of the proposed feature representations in road anomaly detection.

Deep feedforward network (DFN) is the general structure of many well-known deep neural networks (DNN) for image classification. The recent research emphasizes on going deeper and wider network architecture to achieve higher accuracy and lower misclassification rate. This paper provides a study and investigation on stacking three basic operation of neural layers, i:e: convolutional layer, pooling layer and fully connected layer. As a result, a new framework of convolutional deep feedforward network (C-DFN) is proposed in this paper. C-DFN performed significantly better than deep feedforward network (DFN), deep belief network (DBN), and convolutional deep belief network (C-DBN) in MNIST dataset, INRIA pedestrian dataset and Daimler pedestrian dataset. The convolutional layer acts as a trainable feature extractor improving the network performance significantly. Moreover, it reduced 14% of the trainable parameters in DFN. With the use of trainable activation function such as PReLU in the C-DFN, it achieves an average misclassification rate of 9.22% of the three benchmark datasets. [ABSTRACT FROM AUTHOR]

Photovoltaics has gained popularity as a renewable energy source in recent decades. The main challenge for this energy source is the instability in the amount of generated energy owing to its strong dependency on the weather. Therefore, prediction of solar power generation is important for reliable and efficient operation. Popular data sources for predictors are largely divided into recent weather records and numerical weather predictions. This study proposes adequate deep neural networks that can utilise each data source or both. Focusing on a 24‐hour‐ahead prediction problem, the authors first design two deep neural networks for prediction: a deep feedforward network that uses the weather forecast data and a recurrent neural network that uses recent weather observations. Finally, a hybrid network, named PVHybNet, combines the both networks to enhance their prediction performance. In predicting the solar power generation by Yeongam power plant in South Korea, the final model yields an R‐squared value of 92.7%. The results support the effectiveness of the combined network that utilises both weather forecasts and recent weather observations. The authors also demonstrate that the hybrid model outperforms several machine learning models. [ABSTRACT FROM AUTHOR]

Three-phase PWM rectifiers are adopted extensively in industry because of their excellent properties and potential advantages. However, while the IGBT has an open-circuit fault, the system does not crash suddenly, the performance will be reduced for instance voltages fluctuation and current harmonics. A fault diagnosis method based on deep feedforward network with transient synthetic features is proposed to reduce the dependence on the fault mathematical models in this paper, which mainly uses the transient phase current to train the deep feedforward network classifier. Firstly, the features of fault phase current are analyzed in this paper. Secondly, the historical fault data after feature synthesis is employed to train the deep feedforward network classifier, and the average fault diagnosis accuracy can reach 97.85% for transient synthetic fault data, the classifier trained by the transient synthetic features obtained more than 1% gain in performance compared with original transient features. Finally, the online fault diagnosis experiments show that the method can accurately locate the fault IGBTs, and the final diagnosis result is determined by multiple groups results, which has the ability to increase the accuracy and reliability of the diagnosis results. [ABSTRACT FROM AUTHOR]

A spectrum auction is a promising approach with respect to efficiently allocating spectrum among unlicensed users. In this paper, we study the spectrum auction based on the waveform and air-interface of wireless users, the interests of the channel for the auction, and the interference they suffered during communication as well as their economic capability. How to make the analysis and the integration of such multiple factors is a key problem for multi-slot spectrum auction. To address the problem, we adopt the deep feedforward network algorithm to perform waveform and air-interface data analysis and integration for multi-slot spectrum auction. Simulation results are presented to verify the effectiveness of the proposed algorithm in the small cell network. Our approach could be used to 5G where heterogeneous wireless networks will be applied extensively and spectrum auction decision is made based on deep learning and different user patterns.

Abstract The neural architecture is very substantial in order to construct a neural network model that produce a minimum error. Several factors among others include the input choice, the number of hidden layers, the series length, and the activation function. In this paper we present a design of experiment in order to optimize the neural network model. We conduct a simulation study by modeling the data generated from a nonlinear time series model, called subset 3 exponential smoothing transition auto-regressive (ESTAR ([3]). We explore a deep learning model, called deep feedforward network and we compare it to the single hidden layer feedforward neural network. Our experiment resulted in that the input choice is the most important factor in order to improve the forecast performance as well as the deep learning model is the promising approach for forecasting task. [ABSTRACT FROM AUTHOR]

Spectrum auction is considered a suitable approach to efficiently allocate spectrum among unlicensed users. In this paper, for the first time ever, we study this problem from Pattern Recognit. point of view, where all wireless users could be classified based on their economic capability, the interests of the channel for the auction, and the interference they need to suffer during communication. These factors from wireless users are multimodal data ranging from linguistic to numerical data. How to make analysis and integration of such multimodal data is a key for multi-slot spectrum auction. We adopt the Deep Feedforward Network algorithm to perform multimodal data analysis and integration for multi-slot spectrum auction. Simulation results are presented to verify the effectiveness of the proposed algorithm in the small cell network. Our approach is essentially a Pattern Recognit. approach where spectrum auction decision is made based on deep learning and different user patterns. [ABSTRACT FROM AUTHOR]