RFN-Nest: An end-to-end residual fusion network for infrared and visible images

In the image fusion field, the design of deep learning-based fusion methods is far from routine. It is invariably fusion-task specific and requires a careful consideration. The most difficult part of the design is to choose an appropriate strategy to generate the fused image for a specific task in hand. Thus, devising learnable fusion strategy is a very challenging problem in the community of image fusion. To address this problem, a novel end-to-end fusion network architecture (RFN-Nest) is developed for infrared and visible image fusion. We propose a residual fusion network (RFN) which is based on a residual architecture to replace the traditional fusion approach. A novel detail-preserving loss function, and a feature enhancing loss function are proposed to train RFN. The fusion model learning is accomplished by a novel two-stage training strategy. In the first stage, we train an auto-encoder based on an innovative nest connection (Nest) concept. Next, the RFN is trained using the proposed loss functions. The experimental results on public domain data sets show that, compared with the existing methods, our end-to-end fusion network delivers a better performance than the state-of-the-art methods in both subjective and objective evaluation. The code of our fusion method is available at https://github.com/hli1221/imagefusion-rfn-nest.     

Connectionist approach to time series prediction: an empirical test

Among the various potential applications of neural networks, forecasting is considered to be a major application. Several researchers have reported their experiences with the use of neural networks in forecasting, and the evidence is inconclusive. This paper presents the results of a forecasting competition between a neural network model and a Box-Jenkins automatic forecasting expert system. Seventy-five series, a subset of data series which have been used for comparison of various forecasting techniques, were analysed using the Box-Jenkins approach and a neural network implementation. The results show that the simple neural net model tested on this set of time series could forecast about as well as the Box-Jenkins forecasting system.     

Human lower extremity joint moment prediction: A wavelet neural network approach

Joint moment is one of the most important factors in human gait analysis. It can be calculated using multi body dynamics but might not be straight forward. This study had two main purposes; firstly, to develop a generic multi-dimensional wavelet neural network (WNN) as a real-time surrogate model to calculate lower extremity joint moments and compare with those determined by multi body dynamics approach, secondly, to compare the calculation accuracy of WNN with feed forward artificial neural network (FFANN) as a traditional intelligent predictive structure in biomechanics.To aim these purposes, data of four patients walked with three different conditions were obtained from the literature. A total of 10 inputs including eight electromyography (EMG) signals and two ground reaction force (GRF) components were determined as the most informative inputs for the WNN based on the mutual information technique. Prediction ability of the network was tested at two different levels of inter-subject generalization. The WNN predictions were validated against outputs from multi body dynamics method in terms of normalized root mean square error (NRMSE (%)) and cross correlation coefficient (ρ).Results showed that WNN can predict joint moments to a high level of accuracy (NRMSE < 10%, ρ > 0.94) compared to FFANN (NRMSE < 16%, ρ > 0.89). A generic WNN could also calculate joint moments much faster and easier than multi body dynamics approach based on GRFs and EMG signals which released the necessity of motion capture. It is therefore indicated that the WNN can be a surrogate model for real-time gait biomechanics evaluation.     

用RBF神经网络预测材料力学性能的研究

由于径向基函数(RBF)神经网络有易学,动态仿真性强,较强的输入输出映射功能和全局最优逼近的结构特点,因此将之用于预测麦杆增强复合板材力学性能。高斯函数表示形式简单,径向对称,光滑性好和解析性好,所以模型采用高斯函数作为隐含层基函数,k均值聚类法确定径向基函数的参数,运用最小二乘法确定权值。结合影响复合板材力学性能因素的特点和变化规律,以成型温度、成型压力、纤维含量、保温时间、拉伸强度、冲击韧性等为对象建立预测复合板材力学性能的模型,用它来优化模压成型的工艺参数,找出最佳工艺参数的范围。结果表明,径向基函数神经网络具有较好的学习和泛化能力,在预测力学性能中效果较好。     

Bearing condition prediction considering uncertainty: An interval type-2 fuzzy neural network approach

Rolling-element bearings are critical components of rotating machinery. It is important to accurately predict in real-time the health condition of bearings so that maintenance practices can be scheduled to avoid malfunctions or even catastrophic failures. In this paper, an Interval Type-2 Fuzzy Neural Network (IT2FNN) is proposed to perform multi-step-ahead condition prediction of faulty bearings. Since the IT2FNN defines an interval type-2 fuzzy logic system in the form of a multi-layer neural network, it can integrate the merits of each, such as fuzzy reasoning to handle uncertainties and neural networks to learn from data. The interval type-2 fuzzy linguistic process in the IT2FNN enables the system to handle prediction uncertainties, since the type-2 fuzzy sets are such sets whose membership grades are type-1 fuzzy sets that can be used in failure prediction due to the difficult determination of an exact membership function for a fuzzy set. Noisy data of faulty bearings are used to validate the proposed predictor, whose performance is compared with that of a prevalent type-1 condition predictor called Adaptive Neuro-Fuzzy Inference System (ANFIS). The results show that better prediction accuracy can be achieved via the IT2FNN.     

Tool condition monitoring in metal cutting: A neural network approach

This paper discusses the application of neural network-based pattern recognition techniques for monitoring the metal-cutting process. The specific application considered is in-process monitoring of the condition of the cutting tool. Tool condition monitoring is an important prerequisite for successful automation of the metal cutting process. In this paper, we demonstrate the application of supervised and unsupervised neural network paradigms to pattern recognition of sensor signal features. The supervised technique used is backpropagation and the unsupervised technique used is adaptive resonance theory (ART). The results support the premise that, despite excellent classification accuracy by both networks, the unsupervised system holds greater promise in a real world setting. The advantages are discussed and a framework for exploiting them in tool condition monitoring systems is presented.This work was completed as part of graduate research at University of California, Berkeley, Department of Mechanical Engineering.     

Individuals’ personal network characteristics and patterns of Facebook use: A social network approach

Using the theoretical framework of ego-centric networks, this study examines the associations between the characteristics of both Facebook-specific and pre-existing personal networks and patterns of Facebook use. With data from an ego-network survey of college students, the study discovered that various dimensions of Facebook-specific network characteristics, such as multiplexity, proximity, density, and heterogeneity in race, were positively associated with usage patterns, including time spent on Facebook, posting messages, posting photos, and lurking. In contrast, network characteristics of pre-existing relationships, such as density and heterogeneity in race, were negatively associated with Facebook usage patterns. Theoretical implications and limitations were discussed.     

PPNNP: A privacy-preserving neural network prediction with separated data providers using multi-client inner-product encryption

In a neural network of deep learning, it needs a series of algorithms that endeavor to recognize underlying relationships in a set of data. In order to protect the privacy of user’s datasets, traditional schemes can perform the prediction task by setting only a single data provider in the system. However, the data may come from multiple separated data providers rather than single data source in real world since each data provider might hold partial features of a complete prediction sample. It requires that multiple data providers cooperate to perform the prediction for the neural networks by sending their own local data to a well-trained prediction model deployed on a remote cloud server to obtain a predictive label. However, the data owned by multiple data providers usually contain a large amount of private information, which can lead to serious security problems once leaked. To resolve the security and privacy issues of the data owned by multiple data providers, in this paper, we propose a Privacy-Preserving Neural Network Prediction model (PPNNP) that deploys multi-client inner-product functional encryption to the first layer of prediction model. Multiple data providers encrypt their data and upload it to a well-trained model deployed on cloud server, and the server makes predictions by calculating inner-products related to them. It can provide sufficient privacy and security for the data while deploying different neural network architectures with activation functions that are even non-linear on the remote server. We evaluate our scheme based on the real datasets and provide a comparison with the related schemes. Experimental results demonstrate that our scheme can reduce the computational cost of the whole process while significantly reducing the encryption time. It can obtain an accuracy of over 90% in different network architectures with even non-linear activation functions. Meanwhile, our solution can reduce communication overhead in the whole protocol.     

A new approach to supply chain network equilibrium models

We establish nonlinear complementarity formulations for the supply chain network equilibrium models. The formulations have simple structures and facilitate us to study qualitative properties of the models. In this setting, we obtain weaker conditions to guarantee the existence and uniqueness of the equilibrium pattern for a supply chain. A smoothing Newton algorithm that exploits the network structure is proposed for solving these models. Not only is the smoothing Newton algorithm proved to be globally convergent without requiring the assumptions of monotonicity and Lipschitz continuity, but also it can overcome the flaw that the performance of the modified projection method heavily depends on the choice of the predetermined step size. Numerical results indicate the advantages of the nonlinear complementarity formulation and the smoothing Newton algorithm.     

Energy efficient routing algorithm for fiber-wireless access networks: A network formation game approach

New multimedia services and ubiquitous networking pose great challenges on existing access network infrastructures. To cope with such requirements new access technologies, such as the fiber-wireless (FiWi), are being developed. Together with the emergence of new access networks, efforts are being made to reduce the amount of energy required to provide services. Indeed, this issue plays an increasingly important role. Here we propose an energy efficient routing algorithm for FiWi access networks. The main idea is to exploit the multipath capabilities of the wireless mesh front end of FiWi access networks to create energy efficient routes that optimize the sleeping and active periods of all ONUs and wireless nodes. To achieve this goal, an energy efficient network model based on network formation game theory is used. This model allows several network formation processes to be compared in regard to the energy efficiency of the routes they generate. Our results reveal that the farsighted network formation process establishes the most energy efficient routes, meaning that the choices done by this formation process were the best ones. However, this farsighted process is computationally expensive. For this reason a heuristic algorithm is developed, which explores the most energy efficient choices taken by the network formation processes, and farsighted process in particular. Results show that the proposed heuristic is able to obtain results close to the farsighted process.