Dimensionality Of Input Data Research Articles

Kernel principal component analysis-based least squares support vector machine optimized by improved grey wolf optimization algorithm and application in dynamic liquid level forecasting of beam pump

Considering the blind parameters selection and the high dimension of input data in least squares support vector machine (LSSVM) modeling process, a kernel principal component analysis (KPCA)-based LSSVM forecasting method optimized by improved grey wolf optimization (GWO) algorithm is proposed. As an excellent forecasting model, the regression forecasting performance of LSSVM is greatly affected by parameters selection of the model. An improved GWO algorithm with better performance is proposed to determine the optimal parameters of LSSVM. This improved GWO algorithm improves the optimization precision and global optimization ability of the standard GWO algorithm. The parameters of LSSVM model are taken as the optimization object that is optimized by improved GWO algorithm. At the same time, the input variables of LSSVM are correlated and redundant. KPCA algorithm can eliminate the correlation and redundancy between input variables. The reduction of input variables reduces the complexity and training time of modeling process, and the coupling between input variables, to improve the prediction accuracy of LSSVM. The dynamic liquid level of beam pump is chosen as the research object. The proposed forecasting method is applied to the prediction of dynamic liquid level. The simulation comparison on actual collected dynamic liquid level data is performed. The simulation results show that the proposed forecasting method has better predictive performance for the dynamic liquid level.

Crowdsourced Wireless Spectrum Anomaly Detection

Automated wireless spectrum monitoring across frequency, time and space will be essential for many future applications. Manual and fine-grained spectrum analysis is becoming impossible because of the large number of measurement locations and complexity of the spectrum use landscape. Detecting unexpected behaviors in the wireless spectrum from the collected data is a crucial part of this automated monitoring, and the control of detected anomalies is a key functionality to enable interaction between the automated system and the end user. In this paper we look into the wireless spectrum anomaly detection problem for crowdsourced sensors. We first analyze in detail the nature of these anomalies and design effective algorithms to bring the higher dimensional input data to a common feature space across sensors. Anomalies can then be detected as outliers in this feature space. In addition, we investigate the importance of user feedback in the anomaly detection process to improve the performance of unsupervised anomaly detection. Furthermore, schemes for generalizing user feedback across sensors are also developed to close the anomaly detection loop.

Deep Learning in Music Recommendation Systems

Like in many other research areas, deep learning (DL) is increasingly adopted in music recommender systems (MRS). Deep neural networks are used in this area particularly for extracting latent factors of music items from audio signals or metadata and for learning sequential patterns of music items (tracks or artists) from music playlists or listening sessions. Latent item factors are commonly integrated into content-based filtering and hybrid MRS, whereas sequence models of music items are used for sequential music recommendation, e.g., automatic playlist continuation. This review article explains particularities of the music domain in RS research. It gives an overview of the state of the art that employs deep learning for music recommendation. The discussion is structured according to the dimensions of neural network type, input data, recommendation approach (content-based filtering, collaborative filtering, or both), and task (standard or sequential music recommendation). In addition, we discuss major challenges faced in MRS, in particular in the context of the current research on deep learning.

Unveiling phase transitions with machine learning

The classification of phase transitions is a central and challenging task in condensed matter physics. Typically, it relies on the identification of order parameters and the analysis of singularities in the free energy and its derivatives. Here, we propose an alternative framework to identify quantum phase transitions, employing both unsupervised and supervised machine learning techniques. Using the axial next-nearest neighbor Ising (ANNNI) model as a benchmark, we show how unsupervised learning can detect three phases (ferromagnetic, paramagnetic, and a cluster of the antiphase with the floating phase) as well as two distinct regions within the paramagnetic phase. Employing supervised learning we show that transfer learning becomes possible: a machine trained only with nearest-neighbour interactions can learn to identify a new type of phase occurring when next-nearest-neighbour interactions are introduced. All our results rely on few and low dimensional input data (up to twelve lattice sites), thus providing a computational friendly and general framework for the study of phase transitions in many-body systems.

Compressing Recurrent Neural Networks with Tensor Ring for Action Recognition

Recurrent Neural Networks (RNNs) and their variants, such as Long-Short Term Memory (LSTM) networks, and Gated Recurrent Unit (GRU) networks, have achieved promising performance in sequential data modeling. The hidden layers in RNNs can be regarded as the memory units, which are helpful in storing information in sequential contexts. However, when dealing with high dimensional input data, such as video and text, the input-to-hidden linear transformation in RNNs brings high memory usage and huge computational cost. This makes the training of RNNs very difficult. To address this challenge, we propose a novel compact LSTM model, named as TR-LSTM, by utilizing the low-rank tensor ring decomposition (TRD) to reformulate the input-to-hidden transformation. Compared with other tensor decomposition methods, TR-LSTM is more stable. In addition, TR-LSTM can complete an end-to-end training and also provide a fundamental building block for RNNs in handling large input data. Experiments on real-world action recognition datasets have demonstrated the promising performance of the proposed TR-LSTM compared with the tensor-train LSTM and other state-of-the-art competitors.

Reconstruction of three-dimensional anisotropic media based on analysis of morphological completeness

The physical properties of porous media can be more effectively observed and understood when analyzing three-dimensional (3D) structures. The use of computer-based mathematical modeling methods has been proven effective for reconstructing 3D structures of porous media from two-dimensional (2D) images. However, reconstructing a heterogeneous 3D structure from a single 2D image is currently challenging for computer-based mathematical modeling methods. Focusing on the reconstruction of heterogeneous 3D structures from a single anisotropic training image (TI), this paper first introduces a morphological completeness analysis method, which presents a completely new approach to reconstruction in which only two sections are controlled. Then, to enhance the constraint on the simulation points in this new approach to reconstruction, a double-layer and multi-grid system (DLMGS) is proposed. The updated pattern density function simulation, based on the DLMGS, is then taken as the specific reconstruction algorithm in this work. Finally, our reconstruction method is tested using a core image, and the performance of the reconstruction result is assessed using multiple functions. Detailed experiments show that our method can successfully reconstruct a heterogeneous 3D structure from a single anisotropic image, which reduces the required dimensionality of input data for 3D reconstruction and extends the applicable scope.

Deep Learning Model for Personality Traits Classification from Text Emphasis on Data Slicing

Text analytics is the progression of find out insight knowledge over written text such as consumer attitude, product criticisms, emotion analysis, customer response, etc. Previous works fundamentally based on organized data, then again, writings, (for example, tweets) are inexactly organized, have poor spelling, regularly contain linguistic use errors, and they are multilingual. This made the development of much more perplexing and fascinating. The two common methodologies are used, Text-based analytics and Bag of Words. Sentiment parsing emphasis on the structure and grammar of words where Bag of words model neglects the grammar and word type and, in its place, concentrating on signifying text (a sentence, document, and tweet etc.) as the bag of words. In this research work, we have proposed a deep learning model for personality traits classification based on text. In this research paper, we emphasis on pre-processing over input data, meant for dimension reduction which may improve the classification accuracy and will reduce the computation time. In result section we will show the input data dimension reduced after pre-processing.

Designing deep CNN models based on sparse coding for aerial imagery: a deep-features reduction approach

ABSTRACT Traditional methods focus on low-level handcrafted features representations and it is difficult to design a comprehensive classification algorithm for remote sensing scene classification problems. Recently, convolutional neural networks (CNNs) have obtained remarkable performance outcomes, setting several remote sensing benchmarks. Furthermore, direct applications of UAV remote sensing images that use deep convolutional networks are extremely challenging given high input data dimensionality with relatively small amounts of available labelled data. We, therefore, propose a CNN approach to scene classification that architecturally incorporates sparse coding (SC) technique for dimension reduction to minimize overfitting. Outcomes were compared with principal component analysis (PCA) and global average pooling (GAP) alternatives that use fully connected layer(s) in pre-trained CNN architecture(s) to minimize overfitting. SC was used to encode deep features extracted from the last convolutional layer of pre-trained CNN models by using different features maps in which deep features had been converted into low-dimensional SC features. These same sparse-coded features were concatenated by means of different pooling techniques to obtain global image features for scene classification. The proposed algorithm outperformed current state-of-the-art algorithms based on handcrafted features. When using our own UAV-based dataset and existing datasets, it was also exceptionally efficient computationally when learning data representations, producing a 93.64% accuracy rate..

A mix-pooling CNN architecture with FCRF for brain tumor segmentation

MR technique is prevalent for doctor to diagnose and assess glioblastomas which are the most lethal form of brain tumors. Although Convolutional Neural Networks (CNN) has been applied in automatic brain tumor segmentation and is proved useful and efficient, traditional one-pathway CNN architecture with convolutional layers and max pooling layers has limited receptive fields representing the local context information. Such mindset in traditional CNN may dismiss useful global context information. In this paper, we design a two-pathway model with average and max pooling layers in different paths. Besides, 1 × 1 kernels are followed input layers to add the non-linearity dimensions of input data. Finally, we combine the CNN architecture with fully connected CRF(FCRF) as a mixture model to introduce the global context information to optimize prediction results. Our experiments proved that the mixture model improved segmentation and labeling accuracy.

An Efficient Neural Network for Shape from Focus with Weight Passing Method

In this paper, we suggest an efficient neural network model for shape from focus along with weight passing (WP) method. The neural network model is simplified by reducing the input data dimensions and eliminating the redundancies in the conventional model. It helps for decreasing computational complexity without compromising on accuracy. In order to increase the convergence rate and efficiency, WP method is suggested. It selects appropriate initial weights for the first pixel randomly from the neighborhood of the reference depth and it chooses the initial weights for the next pixel by passing the updated weights from the present pixel. WP method not only expedites the convergence rate, but also is effective in avoiding the local minimization problem. Moreover, this proposed method may also be applied to neural networks with diverse configurations for better depth maps. The proposed system is evaluated using image sequences of synthetic and real objects. Experimental results demonstrate that the proposed model is considerably efficient and is able to improve the convergence rate significantly while the accuracy is comparable with the existing systems.

Towards Spatial Composite Indicators: A Case Study on Sardinian Landscape

Composite Indicators (CIs) recently earned popularity as decision-support tool in policy-making for their ability to give concise measures of complex phenomena. Despite growing diffusion of the use of CI in policy-making, current research has barely addressed the issue of the spatial dimension of input data and of final indicator scores. Nowadays the spatial dimension of data plays a crucial role in analysis, thanks to recent developments in spatial data infrastructures which has enabled seamless access to a large amount of geographic information. In addition, recent developments in spatial statistical techniques are facilitating the understanding of the presence of spatial effects among data, spatial dependence and spatial heterogeneity. These advances are improving our ability to understand the spatial dimension of information, which is crucial to obtain a more robust representation of the territorial reality and insights of territorial dynamics in order to inform decisions in spatial planning and policy-making. This paper proposes an original method for the integration of spatial multivariate analysis and the use of spatial data to extend existing state of the art methods for CIs, as a step towards the construction of Spatial Composite Indicators. The method was successfully tested on a landscape planning case study.

Ensemble committee-based data intelligent approach for generating soil moisture forecasts with multivariate hydro-meteorological predictors

Soil moisture (SM) is a key component of the global energy cycle that regulates all domains of the natural environmental and the agricultural system. In this research, the challenge is to develop a low-cost data-intelligent SM forecasting model using climate dynamics (i.e., the climate indices, atmospheric and hydro-meteorological parameters) as the model inputs. A newly designed, multi-model ensemble committee machine learning approach based on the artificial neural network (ANN-CoM) is developed to forecast monthly upper layer (∼0.2 m from the surface) and the lower layer (∼0.2–1.5 m deep) SM at four agricultural sites in Australia’s Murray-Darling Basin. ANN-CoM model is validated with respect to non-tuned second-order Volterra, M5 model tree, random forest, and an extreme learning machine (ELM) models. To construct the ANN-CoM model, the input variables comprised of the hydro-meteorological data from the Australian Water Availability Project, large-scale climate indices and atmospheric parameters derived from the Interim ERA European Centre for Medium-Range Weather Forecasting ECMWF reanalysis fields leads to a total of 60 potential predictors used for SM forecasting. To reduce the model input data dimensionality for accurate forecasts, the Neighborhood Component Analysis (NCA) based feature selection algorithm for regression purposes (fsrnca) is applied to determine the relative feature weights related to the targeted variable. The optimal predictor variables are then screened with an ELM model as the fitness function of the fsrnca algorithm to identify the set of most pertinent model variables. Extensive performance evaluation using statistical score metrics with visual and diagnostic plots show that the ensemble committee based, ANN-CoM model is able to effectively capture the nonlinear dynamics involved in the modeling of monthly upper and lower layer SM levels. Therefore, the ANN-CoM multi-model ensemble-based approach can be considered to be a superior SM forecasting tool, portraying as an amicable, integrated (or ensemble) machine learning stratagem that can be explored for soil moisture modeling and applications in agriculture and other hydro-meteorological phenomena.

Building unconditional forecast model of Stock Market Indexes using combined leading indicators and principal components: application to Vietnamese Stock Market

Objectives: The goal of this paper is to propose a new approach to build unconditional forecast models of stock market indexes in the context of high dimensional input data set. Methods: The methodology of building these models is to combine the method of leading indicators and the Principal Component Analysis techinique (PCA) for dimensionality reduction and to use the multiple regression method on the new data set of reduced dimension. The real data set collected by month of 293 economic-financial variables were used to build a forecast model of Vietnamese stock market index VNINDEX. Findings: The absolute error percentage of the out-of-sample forecasts for the next 4 periods of the model built under the proposed methodology is no more than 1.6%. Applications: The methodology proposed can also be applied to build the unconditional forecast models for the stock prices as well as many other financial-economic indicators. Keywords: Dimensionality Reduction, High Dimension Data, Leading Indicator, PCA, Unconditional Forecast Model, Stock Market

A comparison of the discrete cosine and wavelet transforms for hydrologic model input data reduction

Abstract. The treatment of input data uncertainty in hydrologic models is of crucial importance in the analysis, diagnosis and detection of model structural errors. Data reduction techniques decrease the dimensionality of input data, thus allowing modern parameter estimation algorithms to more efficiently estimate errors associated with input uncertainty and model structure. The discrete cosine transform (DCT) and discrete wavelet transform (DWT) are used to reduce the dimensionality of observed rainfall time series for the 438 catchments in the Model Parameter Estimation Experiment (MOPEX) data set. The rainfall time signals are then reconstructed and compared to the observed hyetographs using standard simulation performance summary metrics and descriptive statistics. The results convincingly demonstrate that the DWT is superior to the DCT in preserving and characterizing the observed rainfall data records. It is recommended that the DWT be used for model input data reduction in hydrology in preference over the DCT.

A SURVEY ON THE CURES FOR THE CURSE OF DIMENSIONALITY IN BIG DATA

Dimensionality reduction techniques are used to reduce the complexity for analysis of high dimensional data sets. The raw input data set may have large dimensions and it might consume time and lead to wrong predictions if unnecessary data attributes are been considered for analysis. So using dimensionality reduction techniques one can reduce the dimensions of input data towards accurate prediction with less cost. In this paper the different machine learning approaches used for dimensionality reductions such as PCA, SVD, LDA, Kernel Principal Component Analysis and Artificial Neural Network have been studied.

Unsupervised feature selection via low-rank approximation and structure learning

Feature selection is an important research topic in machine learning and computer vision in that it can reduce the dimensionality of input data and improve the performance of learning algorithms. Low-rank approximation techniques can well exploit the low-rank property of input data, which coincides with the internal consistency of dimensionality reduction. In this paper, we propose an efficient unsupervised feature selection algorithm, which incorporates low-rank approximation as well as structure learning. First, using the self-representation of data matrix, we formalize the feature selection problem as a matrix factorization with low-rank constraints. This matrix factorization formulation also embeds structure learning regularization as well as a sparse regularized term. Second, we present an effective technique to approximate low-rank constraints and propose a convergent algorithm in a batch mode. This technique can serve as an algorithmic framework for general low-rank recovery problems as well. Finally, the proposed algorithm is validated in twelve publicly available datasets from machine learning repository. Extensive experimental results demonstrate that the proposed method is capable to achieve competitive performance compared to existing state-of-the-art feature selection methods in terms of clustering performance.

A Hybrid Climate Model for Rainfall Forecasting Based on Combination of Self-Organizing Map and Analog Method

A hybrid climate model (HCM) is a novel proposed model based on the combination of self-organizing map (SOM) and analog method (AM). The main purpose was to improve the accuracy in rainfall forecasting using HCM. In combination process of HCM, SOM algorithm classifies high dimensional input data to low dimensional of several disjointed clusters in which similar input is grouped. AM searches the future day that has similar property with the day in the past. Consequently, the analog day is mapped to each cluster of SOM to investigate rainfall. In this study, the input data, geopotential height at 850 hPa from the Climate Forecast System Reanalysis (CFSR) are training set data and also the complete rainfall data at 30-meteorological stations from Thai meteorological department (TMD) are observed. To improve capability of rainfall forecasting, three different measures were evaluated. The experimental results showed that the performance of HCM is better than the traditional AM. It is illustrated that the HCM can forecast rainfall proficiently.

Seismic Failure Probability of a Curved Bridge Based on Analytical and Neural Network Approaches

This study focuses on seismic fragility assessment of horizontal curved bridge, which has been derived by neural network prediction. The objective is the optimization of structural responses of metaheuristic solutions. A regression model for the responses of the horizontal curved bridge with variable coefficients is built in the neural networks simulation environment based on the existing NTHA data. In order to achieve accurate results in a neural network, 1677 seismic analysis was performed in OpenSees. To achieve better performance of neural network and reduce the dimensionality of input data, dimensionality reduction techniques such as factor analysis approach were applied. Different types of neural network training algorithm were used and the best algorithm was adopted. The developed ANN approach is then used to verify the fragility curves of NTHA. The obtained results indicated that neural network approach could be used for predicting the seismic behavior of bridge elements and fragility, with enough feature extraction of ground motion records and response of structure according to the statistical works. Fragility curves extracted from the two approaches generally show proper compliance.

Enhanced hybrid method of divide-and-conquer and RBF neural networks for function approximation of complex problems

This paper provides an enhanced method focused on reducing the computational complexity of function approximation problems by dividing the input data vectors into small groups, which avoids the curse of dimensionality. The computational complexity and memory requirements of the approximated problem are higher when the input data dimensionality increases. A divide-and-conquer algorithm is used to distribute the input data of the complex problem to a divided radial basis function neural network (Div-RBFNN). Under this algorithm, the input data variables are typically distributed to different RBFNNs based on whether the number of the input data dimensions is odd or even. In this paper, each Div-RBFNN will be executed independently and the resulting outputs of all Div-RBFNNs will be grouped using a system of linear combination function. The parameters of each Div-RBFNN (centers, radii, and weights) will be optimized using an efficient learning algorithm, which depends on an enhanced clustering algorithm for function approximation, which clusters the centers of the RBFs. Compared to traditional RBFNNs, the proposed methodology reduces the number of executing parameters. It further outperforms traditional RBFNNs not only with respect to the execution time but also in terms of the number of executing parameters of the system, which produces better approximation error.

An unsupervised two-stage clustering approach for forest structure classification based on X-band InSAR data — A case study in complex temperate forest stands

Forest structure at stand level plays a key role for sustainable forest management, since the biodiversity, productivity, growth and stability of the forest can be positively influenced by managing its structural diversity. In contrast to field-based measurements, remote sensing techniques offer a cost-efficient opportunity to collect area-wide information about forest stand structure with high spatial and temporal resolution. Especially Interferometric Synthetic Aperture Radar (InSAR), which facilitates worldwide acquisition of 3d information independent from weather conditions and illumination, is convenient to capture forest stand structure. This study purposes an unsupervised two-stage clustering approach for forest structure classification based on height information derived from interferometric X-band SAR data which was performed in complex temperate forest stands of Traunstein forest (South Germany). In particular, a four dimensional input data set composed of first-order height statistics was non-linearly projected on a two-dimensional Self-Organizing Map, spatially ordered according to similarity (based on the Euclidean distance) in the first stage and classified using the k-means algorithm in the second stage. The study demonstrated that X-band InSAR data exhibits considerable capabilities for forest structure classification. Moreover, the unsupervised classification approach achieved meaningful and reasonable results by means of comparison to aerial imagery and LiDAR data.