Bearings are critical components in induction motors and brushless direct current motors. Bearing failure is the most common mode these By implementing health monitoring fault diagnosis of bearings, unscheduled maintenance economic losses caused by bearing failures can be avoided. This paper introduces trace ratio linear discriminant analysis (TR-LDA) to deal with high-dimensional non-Gaussian data for dimension reduction classification. Motor single-point faults generalized-roughness used...

In this paper, we propose a novel Convolutional Neural Network (CNN) structure for general-purpose multi-task learning (MTL), which enables automatic feature fusing at every layer from different tasks. This is in contrast with the most widely used MTL CNN structures empirically or heuristically share features on some specific layers (e.g., all except last convolutional layer). The proposed layerwise scheme formulated by combining existing components way, clear mathematical interpretability...

We propose two nuclear- and L2,1-norm regularized 2D neighborhood preserving projection (2DNPP) methods for extracting representative image features. 2DNPP extracts features by minimizing a Frobenius norm-based reconstruction error that is very sensitive noise outliers in given data. To make the distance metric more reliable robust, encode accurately, we minimize L2,1-norm-based error, respectively measure it over each image. Technically, enhanced variants of 2DNPP, nuclear-norm-based sparse...

Recovering low-rank and sparse subspaces jointly for enhanced robust representation classification is discussed. Technically, we first propose a transductive principal feature coding (LSPFC) formulation that decomposes given data into component part encodes features noise-fitting error part. To well handle the outside data, then present an inductive LSPFC (I-LSPFC). I-LSPFC incorporates embedded by projection one problem direct minimization, so can effectively map both inside underlying to...

It has been witnessed that supportive learning played a crucial role in educational quality enhancement. School and family tutoring offer personalized help provide positive feedback on students' learning. Predicting performance is of much interest which reflects their understanding the subjects. Particularly it desired students to manage well fundamental knowledge order build strong foundation for post-secondary studies career. In this paper, improved conditional generative adversarial...

Visualizing similarity data of different objects by exhibiting more separate organizations with local and multimodal characteristics preserved is important in multivariate analysis. Laplacian Eigenmaps (LAE) Locally Linear Embedding (LLE) aim at preserving the embeddings all pairs close vicinity reduced output space, but they are unable to identify interclass neighbors. This paper considers semi-supervised manifold learning problems. We apply pairwise Cannot-Link Must-Link constraints...

A new graph based constrained semi-supervised learning (G-CSSL) framework is proposed. Pairwise constraints (PC) are used to specify the types (intra- or inter-class) of points with labels. Since number labeled data typically small in SSL setting, core idea this create and enrich PC sets using propagated soft labels from both unlabeled by special label propagation (SLP), hence obtaining more supervised information for delivering enhanced performance. We also propose a Two-stage Sparse...

Isomap is a well-known nonlinear dimensionality reduction (DR) method, aiming at preserving geodesic distances of all similarity pairs for delivering highly manifolds. efficient in visualizing synthetic data sets, but it usually delivers unsatisfactory results benchmark cases. This paper incorporates the pairwise constraints into and proposes marginal (M-Isomap) manifold learning. The Cannot-Link Must-Link are used to specify types neighborhoods. M-Isomap computes shortest path over...

In this paper, we propose a semisupervised label consistent dictionary learning (SSDL) framework for machine fault classification. SSDL is extension of recent fully supervised approach, since the number labeled data usually limited in practice. To enable model to use both and commonly readily available unlabeled enhancing performance, incorporate merits prediction present joint adaptive technique. setting, first employ existing estimate labels training signals transductive fashion enriching...

Recommender systems are currently utilized widely in e-commerce for product recommendations and within content delivery platforms. Previous studies usually use independent features to represent item content. As a result, the relationship hidden among is overlooked. In fact, reason that an attracts user may be attributed only few set of features. addition, these often semantically coupled. this paper, we present optimization model extracting by considering preferences. The learned feature...

The study of patient behaviours (vital sign, physical action and emotion) is crucial to improve one's quality life. only solution for handling managing millions people's health would be big data IoT technology because most the countries are lack medical professionals. In this paper, a IoT-based behaviour monitoring system have proposed. Qualitative studies carried out on selected analytics, cardiovascular disease identification fall detection. At last, authors summarised general challenges...

Learning and extracting high-level features from point cloud is the key to improving segmentation performances on clouds for many networks. At present, networks present very deep structures extract 3D perception. However, we argue that even better results can be achieved by (i) building feature vectors integrates multi-scale geometric features, (ii) exerting discriminative constraints learning of mid-levels features. In this paper, propose a Multi-scale Neighborhood Feature Extraction...

The emergence of energy storage system (ESS) enables the service provider to profit from price difference between purchasing electric utility companies and selling it customers through battery operations while more frequent charging/discharging behaviors cause degradation. However, accurate estimation ESS degradation cost is one main obstacles for participating in management. This paper considers a smart grid scenario, consisting an lithium-ion (ALBD) model, renewable generators bilateral...

This paper incorporates the group sparse representation into well-known canonical correlation analysis (CCA) framework and proposes a novel discriminant feature extraction technique named (GSCCA). GSCCA uses two sets of variables aims at preserving (GS) characteristics data within each set in addition to maximize global interset covariance. With GS weights computed prior extraction, locality, sparsity information can be adaptively determined. The are obtained from an NP-hard group-sparsity...

Although supervised single image deraining (SID) have obtained impressive results, they still cannot obtain satisfactory results on real images for the weak generalization of rain removal capacity. In this paper, we mainly discuss semi-supervised SID and propose a new GAN-based network called Semi-DerainGAN, which can use both synthetic data in uniform based two unsupervised processes. For task, streak learner termed SSRML sharing same parameters processes is derived, makes contribute more...