Reduced Input Dimension Research Articles

Proposing an Uncertainty Management Framework to Implement the Evidence Theory for Vehicle Crash Applications

Abstract The purpose of this work is to enable the use of the Dempster–Shafer evidence theory (ET) for uncertainty propagation on computationally expensive automotive crash simulations. This is necessary as the results of these simulations are influenced by multiple possibly uncertain aspects. To avoid negative effects, it is important to detect these factors and their consequences. The challenge when pursuing this effort is the prohibitively high computational cost of the ET. To this end, we present a framework of existing methods that is specifically designed to reduce the necessary number of full model evaluations and parameters. An initial screening removes clearly irrelevant parameters to mitigate the curse of dimensionality. Next, we approximate the full-scale simulation using metamodels to accelerate output generation and thus enable the calculation of global sensitivity indices. These indicate effects of the parameters on the considered output and more profoundly sort out irrelevant parameters. After these steps, the ET can be performed rapidly and feasibly due to fast-responding metamodel and reduced input dimension. It yields bounds for the cumulative distribution function of the considered quantity of interest. We apply the proposed framework to a simplified crash test dummy model. The elementary effects method is used for screening, a kriging metamodel emulates the finite element simulation, and Sobol' sensitivity indices are determined before the ET is applied. The outcome of the framework provides engineers with information about the uncertainties they may face in hardware testing and that should be addressed in future vehicle design.

SurfCNN: A Descriptor Accelerated Convolutional Neural Network for Image-Based Indoor Localization

Convolutional neural network (CNN) is a powerful tool for many data applications. However, its high dimension nature, large network size and computational complexity, and the need of large amount of training data make it challenging to be used in edge computing applications, which are becoming increasingly popular, relevant and important. In this paper, we propose a descriptor based approach to accelerate convolutional neural network training, reduce input dimension and network size, which greatly facilitates the use of CNN for edge computating and even cloud computing. By using image descriptors to extract features from original images, we report a simpler convolutional neural network with fast training speed, low memory usage and outstanding accuracy without the need for a pre-trained network as opposed to the state of art. In indoor localization, our SURF descriptors accelerated CNN (SurfCNN) can reach an average position error of 0.28 m and orientation error of 9.2°. Compared to the conventional CNN that uses original images as input, our algorithm reduces the dimension of the input features by a factor of 48 without impairing the accuracy. Further, at an extreme feature reduction of 14,440 times, our model still retains an average position error retained at 0.41 m and orientation error at 14°.

The prediction of quality characteristics of cotton/elastane core yarn using artificial neural networks and support vector machines

Core yarn is a type of yarn that has a filament fiber in the center with a different fiber wrapped around it. This type of yarn is of growing importance in the textile industry. It is important to predict the quality characteristics of a core yarn before production to prevent the faulty production of fabrics. Therefore, the development of predictive models is a necessity in the textile industry. In this study, artificial neural network (ANN) and support vector machine (SVM) models are proposed to predict the quality characteristics of cotton/elastane core yarn, using fiber quality and spinning parameters. Principal component analysis and analysis of variance techniques are also used to reduce input dimensions, since high dimensional data may reduce a model’s potential for success in prediction. The prediction models are trained and tested using the data obtained from a textile production plant. The results of all the models are compared with each other on test data. Mean absolute percentage error (MAPE), mean absolute error (MAE) and correlation coefficient (R) are used to assess the prediction power of the models. Although on most of the tests SVM models fared slightly better than ANN models, both models provide accurate predictions for most of the yarn quality characteristics. The results show that the best models have over 90% success rate in MAPE and R. In particular, the Coefficient of Variance of mass (CVm) along the yarn, hairiness and Reisskilometer quality characteristics of the cotton/elastane core yarn are predicted with 91%, 93% and 95% accuracy, respectively.

Machine learning approach to predict center of pressure trajectories in a complete gait cycle: a feedforward neural network vs. LSTM network.

Center of pressure (COP) trajectories of human can maintain regulation of forward progression and stability of lateral sway during walking. The insole pressure system can only detect COP trajectories of each foot during single stance. In this study, we developed artificial neural network models that could present COP trajectories in an integrated coordinate system during a complete gait cycle using pressure information of the insole system. A feed forward artificial neural network (FFANN) and a long short-term memory (LSTM) model were developed. For FFANN, among 198 pressure sensors from Pedar-X insoles, proper input variables were selected using sequential forward selection to reduce input dimension. The LSTM model used all 198 signals as inputs because of its self-learning characteristic. As results of cross-validation, the FFANN model showed correlation coefficients of 0.98-0.99 and 0.93-0.95 in anterior/posterior and medial/lateral directions, respectively. For the LSTM model, correlation coefficients were similar to those of FFANN. However, the relative root mean square error (12.5%) of the FFANN model was higher than that (9.8%) of the LSTM model in medial/lateral direction (p = 0.03). This study can be used for quantitative evaluation of clinical diagnosis and rehabilitation status for patient with various diseases through further training using varied databases. Graphical abstract Architectures of neural networks developed in this study (a feed forward artificial neural network; b LSTM network).

A weighted auto regressive LSTM based approach for chemical processes modeling

Data-driven methods have been regarded as effective methods for modeling in chemical processes. However, with the increasing complexity of chemical processes in spatial domain and time domain, how to extract meaningful features and build corresponding models are keys for accurate modeling tasks. To retain temporal features of original inputs, a Recurrent Denosing Autoencoder (RDAE) is built to extract meaningful features and reduce input dimension in the spatial domain, and Cumulative Percent Variance (CPV) is introduced to decide the number of extracted features. Considering correction and prior knowledge effects of real history outputs, a Weighted Auto Regressive Long Short Term Memory (WAR-LSTM) structure is proposed as the basic cell, then multiple WAR-LSTMs are stacked as deep WAR-LSTM to extract high level representations from multi variables. Hence, spatial and temporal information are sufficiently used both in extracting features and building model. The benchmarked Tennessee Eastman process data and real process data from a Fluid Catalytic Cracking (FCC) unit verify the effectiveness of our work.

Slope displacement prediction using sequential intelligent computing algorithms

Slope displacement prediction plays a critical role in the early warning system for landslide and greatly helps prevent property damage and loss of human lives. Given the non-stationary and complex characteristic of the slope deformation, this paper proposes a slope displacement prediction model and an early warming framework based on a set of sequential intelligent computing algorithms that can take advantages of Rough Set theory (RS), Kernel principal component analysis (KPCA), quantum particle swarm optimization (QPSO), least square support vector machine (LSSVM), and Markov chain (MC). Firstly, based on the analysis results of field monitoring data, the environmental parameters affecting the landslide displacement are used as the initial input variables and discretized to remove the effects of dimension and magnitude. After this process, RS is utilized to identify the important influence factors in order to eliminate the multi-collinearity and redundancy of the attributes that were selected initially. As the extracted parameters are still high-dimensional, KPCA is employed to fuse them into a comprehensive indicator to further reduce input dimension and computational cost. The nonlinear relation model between the indicator and displacement is established by using LSSVM, with the parameters optimized through QPSO that has much faster and better global search ability. Once the QPSO-LSSVM model is established, MC is integrated to refine the prediction results. Five months continuous field measurements from the real time monitoring system of Tuyang landslide is applied to evaluate the effectiveness of the proposed model. The results demonstrate that the proposed approach achieves higher prediction accuracy, faster convergence, and better generalization compared with existing prevalent models.

Prediction of the Particle Size Distribution Parameters in a High Shear Granulation Process Using a Key Parameter Definition Combined Artificial Neural Network Model

Various methodologies have been proposed in the literatures to predict the particle size distribution (PSD) in agglomeration processes. However, there is no universal model that can completely satisfy industrial needs due to the complexity of the agglomeration process and the unclear mechanisms of particle growth. A systematic approach using an artificial neural network (ANN) model to predict PSD parameters of different agglomeration processes in high shear mixer was developed. In order to reduce input dimensions, several dimensional or dimensionless parameter groups are defined to represent the raw material properties, equipment geometries, and operation conditions as inputs. According to statistical analysis, an ANN model combined with the key parameter definitions can predict product mean particle size (MPS) within an average absolute relative error (AARE) of 7.18%, a standard deviation (σ) of 7.72%, and a correlation coefficient (R) of 0.9657 (while for distribution span prediction AARE is 5.76%, σ is...

Nonlinear identification of a gasoline HCCI engine using neural networks coupled with principal component analysis

Homogeneous charge compression ignition (HCCI) is a futuristic combustion technology that operates with high efficiency and reduced emissions. HCCI combustion is characterized by complex nonlinear dynamics which necessitates the use of a predictive model in controller design. Developing a physics based model for HCCI involves significant development times and associated costs arising from developing simulation models and calibration. In this paper, a neural networks (NN) based methodology is reported where black box type models are developed to predict HCCI combustion behavior during transient operation. The NN based approach can be considered a low cost and quick alternative to the traditional physics based modeling. A multi-input single-output model was developed each for indicated net mean effective pressure, combustion phasing, maximum in-cylinder pressure rise rate and equivalent air–fuel ratio. The two popular architectures namely multi-layer perceptron (MLP) and radial basis network (RBN) models were compared with respect to design, prediction performance and overall applicability to the transient HCCI modeling problem. A principal component analysis (PCA) is done as a pre-processing step to reduce input dimension thereby reducing memory requirements of the models. Also, PCA reduces the cross-validation time required to identify optimal model hyper-parameters. On comparing the model predictions with the experimental data, it was shown that neural networks can be a powerful approach for non-linear identification of a complex combustion system like the HCCI engine.

Element yield rate prediction in ladle furnace based on improved GA-ANFIS

The traditional prediction methods of element yield rate can be divided into experience method and data-driven method. But in practice, the experience formulae are found to work only under some specific conditions, and the sample data that are used to establish data-driven models are always insufficient. Aiming at this problem, a combined method of genetic algorithm (GA) and adaptive neuro-fuzzy inference system (ANFIS) is proposed and applied to element yield rate prediction in ladle furnace (LF). In order to get rid of the over reliance upon data in data-driven method and act as a supplement of inadequate samples, smelting experience is integrated into prediction model as fuzzy empirical rules by using the improved ANFIS method. For facilitating the combination of fuzzy rules, feature construction method based on GA is used to reduce input dimension, and the selection operation in GA is improved to speed up the convergence rate and to avoid trapping into local optima. The experimental and practical testing results show that the proposed method is more accurate than other prediction methods.

A novel information geometric approach to variable selection in MLP networks

In this paper, a novel information geometric-based variable selection criterion for multi-layer perceptron networks is described. It is based on projections of the Riemannian manifold defined by a multi-layer perceptron network on submanifolds defined by multi-layer perceptron networks with reduced input dimension. We show how the divergence between models can be used as a criterion for an efficient search in the space of networks with different inputs. Then, we show how the posterior probabilities of the models can be evaluated to rank the projected models. Finally, we test our algorithm on synthetic and real data, and compare its performances with other methods reported in literature.

Ensemble competitive learning neural networks with reduced input dimension.

Conventional neural networks utilize all the dimensions of the original input patterns for training and classification. However, a particular attribute of the input patterns does not necessarily contribute to classification and may even cause misclassification in certain cases. A new ensemble competitive learning method using the reduced input dimension is proposed. In contrast to the previous ensemble neural networks which adjust learning parameters, the proposed method takes advantage of the information in each dimension of the input patterns. Since the degree of contribution of each attribute to classification is not known beforehand, the different input data sets with one dimension reduced are presented to multiple neural networks. The classification information from each competitive learning neural network is then combined to make a final decision for classification. In order to improve classification accuracy, the ambiguous output neurons are eliminated which cannot be assigned to any class after training. We use three consensus schemes to judge the classification using ensemble neural networks. The experimental results with remote sensing and speech data indicate the improved performance of the proposed method.