Raw Data Space Research Articles

ECGVEDNET: A Variational Encoder-Decoder Network for ECG Delineation in Morphology Variant ECGs.

Electrocardiogram (ECG) delineation to identify the fiducial points of ECG segments, plays an important role in cardiovascular diagnosis and care. Whilst deep delineation frameworks have been deployed within the literature, several factors still hinder their development: (a) data availability: the capacity of deep learning models to generalise is limited by the amount of available data; (b) morphology variations: ECG complexes vary, even within the same person, which degrades the performance of conventional deep learning models. To address these concerns, we present a large-scale 12-leads ECG dataset, ICDIRS, to train and evaluate a novel deep delineation model-ECGVEDNET. ICDIRS is a large-scale ECG dataset with 156,145 QRS onset annotations and 156,145 T peak annotations. ECGVEDNET is a novel variational encoder-decoder network designed to address morphology variations. In ECGVEDNET, we construct a well-regularized latent space, in which the latent features of ECG follow a regular distribution and present smaller morphology variations than in the raw data space. Finally, a transfer learning framework is proposed to transfer the knowledge learned on ICDIRS to smaller datasets. On ICDIRS, ECGVEDNET achieves accuracy of 86.28%/88.31% within 5/10 ms tolerance for QRS onset and accuracy of 89.94%/91.16% within 5/10 ms tolerance for T peak. On QTDB, the average time errors computed for QRS onset and T peak are -1.86 ± 8.02 ms and -0.50 ± 12.96 ms, respectively, achieving state-of-the-art performances on both large and small-scale datasets. We will release the source code and the pre-trained model on ICDIRS once accepted.

Robust Reflection Removal With Flash-Only Cues in the Wild.

We propose a simple yet effective reflection-free cue for robust reflection removal from a pair of flash and ambient (no-flash) images. The reflection-free cue exploits a flash-only image obtained by subtracting the ambient image from the corresponding flash image in raw data space. The flash-only image is equivalent to an image taken in a dark environment with only a flash on. This flash-only image is visually reflection-free and thus can provide robust cues to infer the reflection in the ambient image. Since the flash-only image usually has artifacts, we further propose a dedicated model that not only utilizes the reflection-free cue but also avoids introducing artifacts, which helps accurately estimate reflection and transmission. Our experiments on real-world images with various types of reflection demonstrate the effectiveness of our model with reflection-free flash-only cues: our model outperforms state-of-the-art reflection removal approaches by more than 5.23dB in PSNR. We extend our approach to handheld photography to address the misalignment between the flash and no-flash pair. With misaligned training data and the alignment module, our aligned model outperforms our previous version by more than 3.19dB in PSNR on a misaligned dataset. We also study using linear RGB images as training data. Our source code and dataset are publicly available at https://github.com/ChenyangLEI/flash-reflection-removal.

Heterogeneous Regularization-Based Tensor Subspace Clustering for Hyperspectral Band Selection.

Band selection (BS) reduces effectively the spectral dimension of a hyperspectral image (HSI) by selecting relatively few representative bands, which allows efficient processing in subsequent tasks. Existing unsupervised BS methods based on subspace clustering are built on matrix-based models, where each band is reshaped as a vector. They encode the correlation of data only in the spectral mode (dimension) and neglect strong correlations between different modes, i.e., spatial modes and spectral mode. Another issue is that the subspace representation of bands is performed in the raw data space, where the dimension is often excessively high, resulting in a less efficient and less robust performance. To address these issues, in this article, we propose a tensor-based subspace clustering model for hyperspectral BS. Our model is developed on the well-known Tucker decomposition. The three factor matrices and a core tensor in our model encode jointly the multimode correlations of HSI, avoiding effectively to destroy the tensor structure and information loss. In addition, we propose well-motivated heterogeneous regularizations (HRs) on the factor matrices by taking into account the important local and global properties of HSI along three dimensions, which facilitates the learning of the intrinsic cluster structure of bands in the low-dimensional subspaces. Instead of learning the correlations of bands in the original domain, a common way for the matrix-based models, our model learns naturally the band correlations in a low-dimensional latent feature space, which is derived by the projections of two factor matrices associated with spatial dimensions, leading to a computationally efficient model. More importantly, the latent feature space is learned in a unified framework. We also develop an efficient algorithm to solve the resulting model. Experimental results on benchmark datasets demonstrate that our model yields improved performance compared to the state-of-the-art.

Joint sparse latent representation learning and dual manifold regularization for unsupervised feature selection

As an effective dimensionality reduction method, unsupervised feature selection (UFS) focuses on the mutual correlations between high-dimensional data features but often overlooks the intrinsic relationships between instances. We also utilize pseudo-labels learned from the data to guide feature selection in UFS. However, the raw data space may contain noise and outliers, leading to a lower accuracy of the learned pseudo-label matrix. We propose a minimum-redundant UFS approach to tackle these problems through jointing sparse latent representation learning with dual manifold regularization (SLRDR). Firstly, SLRDR learns a subspace of latent representation by exploring the interconnection of original data. To enhance subspace sparsity, ℓ2,1-norm is applied to the residual matrix of latent representation learning. Pseudo-label matrix learning is then carried out in the high-quality latent space, resulting in effective pseudo-label information that can provide more useful guidance for sparse regression. Secondly, based on the manifold learning hypothesis, SLRDR exploits features’ local structural properties in feature space and explores the association between data and labels, allowing the model to learn richer and more accurate structural information. In addition, ℓ2,1/2-norm is imposed on the weight matrix to obtain a minimum-redundant solution and select more discriminative features. Finally, an alternating iterative method is used for SLRDR to solve the optimization problem of the objective function, and the convergence of the model is theoretically analyzed. Besides, a series of comparative experiments with ten existing algorithms on nine benchmark datasets are used to verify the model’s effectiveness.

Measurement-induced criticality as a data-structure transition

We employ unsupervised learning tools to identify different phases and their transition in quantum systems subject to the combined action of unitary evolution and stochastic measurements. Specifically, we consider principal component analysis and intrinsic dimension estimation to reveal a measurement-induced structural transition in the data space. We test our approach on a 1+1D stabilizer circuit and find the quantities of interest furnish novel order parameters defined directly in the raw data space. Our results provide a first use of unsupervised tools in dynamical quantum phase transitions.

Abnormal event detection using adversarial predictive coding for motion and appearance

In this paper, we propose adversarial predictive coding (APC), a novel method for detecting abnormal events. Abnormal event detection (AED) is to identify unobserved events from a given training dataset consisting of normal events, and it is considered as one of the most important objectives in developing intelligent surveillance systems. Given videos and motion flows of normal events, the APC derives a normal event model by applying an adversarial prediction approach on the jointly learnt latent feature space from the videos and motion flows. Since latent space requires more abstracted and noise-free information than the raw data space, the APC can derive a more discriminative model for normal events compared with other deep learning-based AED methods which directly apply uni-modal losses such as mean square error and cross-entropy to low-level data such as video frames. We demonstrate the effectiveness of our method in detecting abnormal events using UCSD-Ped, Avenue, and UCF-Crime datasets. The experimental results show that the APC surpass the existing state-of-the-art AED methods by deriving a more discriminative model for normal events.

Self-supervised graph convolutional clustering by preserving latent distribution

Node clustering has attracted widespread concern due to the fact that graph structure helps to seek the geometric structure of data. Existing graph convolutional networks (GCNs) based node clustering methods usually contain two steps. First, learning latent nodes representations. Second, inferring the unknown labels. Despite the promising preliminary results, it cannot take full advantage of the information embedded in current pseudo clustering labels, resulting in suboptimal performance. In fact, it is the truth we do not know the true label of the node in clustering, but the pseudo clustering label of the node generated by each iteration is known. Under this condition, although the pseudo labels of some nodes are incorrect, the partial correct label information is useful and significant. Moreover, GCNs based node clustering aims to learn satisfactory low dimensional node representations which are usually used for clustering. However, existing GCNs based clustering methods fail to take into account distribution consistency between the raw data space and the latent space of nodes representations, resulting in insufficient representations. To address the above problems, we propose a novel GCN-based node clustering method. By introducing a self-supervision module, it can employ the good property of pseudo clustering labels to self-supervise the learning of node representations. Moreover, a latent distribution preserving term, which is measured by KL divergence, is employed to help the latent representations of the same sample to further have a consistent distribution in dimension space as well as in the original dimension space. We formulate the above two concerns into a unified optimization framework. Experimental results on several public datasets indicate our method is effective compared with the state-of-the-art node clustering methods.

Sketch-then-Edit Generative Adversarial Network

Generative Adversarial Network (GAN) has been widely used to generate impressively plausible data. However, it is a non-trivial task to train the original GAN model in practice due to the vanishing gradient problem. This is because the JS divergence could be a constant (i.e., log2) when original data distribution and generated data distribution hold a negligible overlapping area. Under such a scenario, the gradient of generator is 0. Most efforts have been devoted to designing a more proper difference measure while few attentions have been paid to the former aspect of the issue.In this paper, we propose a new method to design a noise distribution having a guaranteed non-negligible overlapping area with raw data distribution. The key idea is to transform the noise from the randomized space into the raw data space. We propose to obtain the transformation as the basis matrix in non-negative matrix factorization because the basis matrix has the underlying features of the raw data. The proposed idea is instantiated as Sketch-then-Edit GAN (SEGAN) where sketches are the noises after transformation and are adopted as the name since they contains basic features of the raw data. Moreover, a new generator for editing the sketches into realistic-like data is designed. We mathematically prove that SEGAN solves the gradient vanishing problem, and conduct extensive experiments on the MNIST, CIFAR10, SVHN and Celeba datasets to demonstrate the effectiveness of SEGAN.

A Flexible Architecture for Extracting Metro Tunnel Cross Sections from Terrestrial Laser Scanning Point Clouds

This paper presents a novel framework to extract metro tunnel cross sections (profiles) from Terrestrial Laser Scanning point clouds. The entire framework consists of two steps: tunnel central axis extraction and cross section determination. In tunnel central extraction, we propose a slice-based method to obtain an initial central axis, which is further divided into linear and nonlinear circular segments by an enhanced Random Sample Consensus (RANSAC) tunnel axis segmentation algorithm. This algorithm transforms the problem of hybrid linear and nonlinear segment extraction into a sole segmentation of linear elements defined at the tangent space rather than raw data space, significantly simplifying the tunnel axis segmentation. The extracted axis segments are then provided as input to the step of the cross section determination which generates the coarse cross-sectional points by intersecting a series of straight lines that rotate orthogonally around the tunnel axis with their local fitted quadric surface, i.e., cylindrical surface. These generated profile points are further refined and densified via solving a constrained nonlinear least squares problem. Our experiments on Nanjing metro tunnel show that the cross sectional fitting error is only 1.69 mm. Compared with the designed radius of the metro tunnel, the RMSE (Root Mean Square Error) of extracted cross sections’ radii only keeps 1.60 mm. We also test our algorithm on another metro tunnel in Shanghai, and the results show that the RMSE of radii only keeps 4.60 mm which is superior to a state-of-the-art method of 6.00 mm. Apart from the accurate geometry, our approach can maintain the correct topology among cross sections, thereby guaranteeing the production of geometric tunnel model without crack defects. Moreover, we prove that our algorithm is insensitive to the missing data and point density.

Representation Space-Based Discriminative Graph Construction for Semisupervised Hyperspectral Image Classification

Graph-based semisupervised learning methods have been successfully applied in hyperspectral image (HSI) classification with limited labeled samples. The critical step of graph-based methods is to learn a similarity graph, and numerous graph construction methods have been developed in recent years. However, existing approaches usually return a similarity matrix from the raw data space. In this letter, we propose a representation space-based discriminative graph for semisupervised HSI classification, which can learn the representations of samples and the similarity matrix of representations simultaneously. Moreover, we explicitly incorporate the probabilistic class relationship between sample and class, which can be estimated by the partial label information, into the above model to further boost the discriminability of graph. The experimental results on Hyperion and AVIRIS hyperspectral data demonstrate the effectiveness of the proposed approach.

Raw data queries during data-intensive parallel workflow execution

Computer simulations consume and produce huge amounts of raw data files presented in different formats, e.g., HDF5 in computational fluid dynamics simulations. Users often need to analyze domain-specific data based on related data elements from multiple files during the execution of computer simulations. In a raw data analysis, one should identify regions of interest in the data space and retrieve the content of specific related raw data files. Existing solutions, such as FastBit and RAW, are limited to a single raw data file analysis and can only be used after the execution of computer simulations. Scientific Workflow Management Systems (SWMS) can manage the dataflow of computer simulations and register related raw data files at a provenance database. This paper aims to combine the advantages of a dataflow-aware SWMS and the raw data file analysis techniques to allow for queries on raw data file elements that are related, but reside in separate files. We propose a component-based architecture, named as ARMFUL (Analysis of Raw data from Multiple Files) with raw data extraction and indexing techniques, which allows for a direct access to specific elements or regions of raw data space. ARMFUL innovates by using a SWMS provenance database to add a dataflow access path to raw data files. ARMFUL facilitates the invocation of ad-hoc programs and third party tools (e.g., FastBit tool) for raw data analyses. In our experiments, a real parallel computational fluid dynamics is executed, exploring different alternatives of raw data extraction, indexing and analysis.

A Bayesian Classification Approach Using Class-Specific Features for Text Categorization

In this paper, we present a Bayesian classification approach for automatic text categorization using class-specific features. Unlike conventional text categorization approaches, our proposed method selects a specific feature subset for each class. To apply these class-specific features for classification, we follow Baggenstoss's PDF Projection Theorem (PPT) to reconstruct the PDFs in raw data space from the class-specific PDFs in low-dimensional feature subspace, and build a Bayesian classification rule. One noticeable significance of our approach is that most feature selection criteria, such as Information Gain (IG) and Maximum Discrimination (MD), can be easily incorporated into our approach. We evaluate our method's classification performance on several real-world benchmarks, compared with the state-of-the-art feature selection approaches. The superior results demonstrate the effectiveness of the proposed approach and further indicate its wide potential applications in data mining.

Semi-Supervised Affine Alignment of Manifolds

High dimensional data is usually produced by the source that only enjoys a limited number of degrees of freedom. Manifold leaning technique plays an important part in finding the correlation among the high dimensional data datasets. By making use of manifold alignment, the paired mapping relationship can be explored easily. However, the common manifold alignment algorithm can only give the mapping result of the training set, and cannot deal with a new coming point. A new manifold alignment algorithm is proposed in this paper. The benefit of our algorithm is two fold: First, the method is a semi-supervised approach, which makes better use of the local geometry information of the unpaired points and improves the learning effect when the labeled proportion is very low. Second, an extended spectral aggression trick is used in the algorithm, which can produce a linear mapping between the raw data space and the aligned space. The experiments result shows that, the correlation mapping can be precisely obtained, the hidden space can be aligned effectively, and the cost of mapping a coming point is very low. DOI : http://dx.doi.org/10.11591/telkomnika.v12i4.4808

Multivariate Spatial Outlier Detection Using Robust Geographically Weighted Methods

Outlier detection is often a key task in a statistical analysis and helps guard against poor decision-making based on results that have been influenced by anomalous observations. For multivariate data sets, large Mahalanobis distances in raw data space or large Mahalanobis distances in principal components analysis, transformed data space, are routinely used to detect outliers. Detection in principal components analysis space can also utilise goodness of fit distances. For spatial applications, however, these global forms can only detect outliers in a non-spatial manner. This can result in false positive detections, such as when an observation’s spatial neighbours are similar, or false negative detections such as when its spatial neighbours are dissimilar. To avoid mis-classifications, we demonstrate that a local adaptation of various global methods can be used to detect multivariate spatial outliers. In particular, we account for local spatial effects via the use of geographically weighted data with either Mahalanobis distances or principal components analysis. Detection performance is assessed using simulated data as well as freshwater chemistry data collected over all of Great Britain. Results clearly show value in both geographically weighted methods to outlier detection.

Development, implementation and evaluation of a dedicated metal artefact reduction method for interventional flat-detector CT

The purpose of this study was to develop, implement and evaluate a dedicated metal artefact reduction (MAR) method for flat-detector CT (FDCT). The algorithm uses the multidimensional raw data space to calculate surrogate attenuation values for the original metal traces in the raw data domain. The metal traces are detected automatically by a three-dimensional, threshold-based segmentation algorithm in an initial reconstructed image volume, based on twofold histogram information for calculating appropriate metal thresholds. These thresholds are combined with constrained morphological operations in the projection domain. A subsequent reconstruction of the modified raw data yields an artefact-reduced image volume that is further processed by a combining procedure that reinserts the missing metal information. For image quality assessment, measurements on semi-anthropomorphic phantoms containing metallic inserts were evaluated in terms of CT value accuracy, image noise and spatial resolution before and after correction. Measurements of the same phantoms without prostheses were used as ground truth for comparison. Cadaver measurements were performed on complex and realistic cases and to determine the influences of our correction method on the tissue surrounding the prostheses. The results showed a significant reduction of metal-induced streak artefacts (CT value differences were reduced to below 22 HU and image noise reduction of up to 200%). The cadaver measurements showed excellent results for imaging areas close to the implant and exceptional artefact suppression in these areas. Furthermore, measurements in the knee and spine regions confirmed the superiority of our method to standard one-dimensional, linear interpolation.

Preprocessing of SAW Sensor Array Data and Pattern Recognition

This paper discusses data preprocessing methods for feature extraction in polymer coated surface acoustic wave (SAW) vapor sensor array. The role of sensor response modeling is explored in developing an appropriate preprocessing strategy. The preprocessor should be designed to transform the experimentally measured data variables (sensor signals) into a format that relates the characteristic features of the object for recognition (vapor) linearly with the transformed data variables. This facilitates generation of greater dispersion in feature space defined by linear feature extraction method such as principal component analysis. Considering solvation parameters of the vapor molecules to be their characteristic descriptors (features) and prompted by the equilibrium response model of SAW sensors, it is demonstrated that by transforming the raw data space logarithmically generates greater dispersion in feature space defined by the principal component analysis, and enhances classification efficiency of the backpropagation neural network substantially.

Class-specific classifier: avoiding the curse of dimensionality

This article describes a new probabilistic method called the "class-specific method" (CSM). CSM has the potential to avoid the "curse of dimensionality" which plagues most classifiers which attempt to determine the decision boundaries in a high-dimensional feature space. In contrast, in CSM, it is possible to build classifiers without a common feature space. Separate low-dimensional features sets may be defined for each class, while the decision functions are projected back to the common raw data space. CSM effectively extends the classical classification theory to handle multiple feature spaces. It is completely general, and requires no simplifying assumption such as Gaussianity or that data ties in linear subspaces.

The class-specific method for classification

This talk describes a new probabilistic method for classification called the ‘‘class-specific method’’ (CSM). CSM is able to avoid the ‘‘curse of dimensionality’’ which plagues most classifiers which attempt to determine the decision boundaries in a high-dimensional feature space. Using CSM, it is possible to build a theoretically optimum classifier without a common feature space. Separate low-dimensional features sets may be defined for each class, while the decision functions are projected back to the common raw data space. CSM effectively extends classical classification theory to handle multiple feature spaces. It is completely general, and requires no simplifying assumption such as Gaussianity or that data lies in linear subspaces. In real-data problems, CSM has shown orders of magnitude reductions in the false-alarm rate. CSM achieves this gain because it is able to make use of partial prior knowledge about the data classes. In contrast, the existing theory can only make use of full knowledge—that is when the parametric forms of the data probability density functions (PDFs) are known.

The PDF projection theorem and the class-specific method

We present the theoretical foundation for optimal classification using class-specific features and provide examples of its use. A new probability density function (PDF) projection theorem makes it possible to project probability density functions from a low-dimensional feature space back to the raw data space. An M-ary classifier is constructed by estimating the PDFs of class-specific features, then transforming each PDF back to the raw data space where they can be fairly compared. Although statistical sufficiency is not a requirement, the classifier thus constructed becomes equivalent to the optimal Bayes classifier if the features meet sufficiency requirements individually for each class. This classifier is completely modular and avoids the dimensionality curse associated with large complex problems. By recursive application of the projection theorem, it is possible to analyze complex signal processing chains. We apply the method to feature sets, including linear functions of independent random variables, cepstrum, and Mel cepstrum. In addition, we demonstrate how it is possible to automate the feature and model selection process by direct comparison of log-likelihood values on the common raw data domain.

Neural-morphological approach for pattern classification

In cluster analysis, the mode boundaries are a very important part of the hierarchy of structures that link raw data with their interpretation. The existing mode boundary detection approaches for clustering are conditioned by the adjustment of some parameters, which become critical for large dimensionality data sets. Mode boundary detection can be greatly facilitate by mapping, as a first step of process understanding, a reduction of data dimensionality. Under this assumption, an approach is discussed, based on both neural network and mathematical morphology. It requires neither a starting classification, nor an a priori number of clusters or their distribution. Data projection mapping is done using a multilayer neural network with a fast training rule based on a conjugate gradient. Mode boundaries of the underlying probability density function, estimated from the patterns in the projection space, are then easily obtained by making concepts of morphological watershed transformations suitable for their detection. The observations in the raw data space corresponding to those falling in the so-detected mode boundaries are taken as prototypes for classification. The clustering scheme, illustrated using an artificial simulation, has been applied to determine the clusters inside a set of biometrical six-dimensional data of the Guadeloupe honeybee's races.