Search

Augmented reality (AR) technologies provide new scenarios for analysis and medical data representation, by combining the standard image-based analysis using 2D-slices with 3D virtual models of patient-specific structures of the body overlapped in the view of the real world in real-time. In this paper, we present the development of a generic AR-based and wireless gestures platform for image-guided surgery planning. The system was assessed with two minimally invasive surgery scenarios: heart mitral valve repair and endovascular brain aneurysm repair using CT Angiography (CTA) images. Twenty-one experiments with heart surgeons and neurosurgeons were carried out, testing the clinical pathology, anatomy exploration, interaction, and perceived usability. The tasks and scenarios, perceived illusion related to the realism of the 3D virtual models and hand manipulation, and the navigation experience were all well evaluated. Prospective applications were recognized as potentially useful and feasible for enhancing planning and trans-operative interventions for all the participants. Experiments performed showed that surgeons found the application a suitable alternative to understand better the previous steps involved in surgical planning, especially for minimally invasive surgery.

The popularity of mobile devices has led to an explosive growth in the number of mobile apps in which Android mobile apps are the mainstream. Android mobile apps usually undergo frequent update due to new requirements proposed by users. Just-in-time (JIT) defect prediction is appropriate for this scenario for quality assurance because it can provide timely feedback by determining whether a new code commit will introduce defects into the apps. As defect-prediction performance usually relies on the quality of the data representation and the used classification model, in this work, we propose a model, called Simplified Deep Forest (SDF), to conduct JIT defect prediction for Android mobile apps. SDF modifies a state-of-the-art deep forest model by removing the multigrained scanning operation that is designed for data with a high-dimensional feature space. It uses a cascade structure with ensemble forests for representation learning and classification. We conduct experiments on 10 Android mobile apps and experimental results show that SDF performs significantly better than comparative methods in terms of 3 performance indicators.

Graph Convolutional Networks (GCNs) recently have been adopted in several feature representation studies for different classification tasks. While many of these methods are used to work with irregular structure data, they are rarely used to learn regular structure data. It is crucial to construct an excellent graph representation to traditional classification tasks for obtaining the sufficient data representation, including attribute representation and relative representation. In this context, we propose a novel method to construct a reasonable graph representation by capturing the relations in low dimensional space among the data. In order to get the well-represented, we introduce the low-rank constraint and L2-norm regularization to the graph learning framework simultaneously. Experiments demonstrate that the proposed method to learn graph representation is helpful for classification task, and leads to improved performance when compared to state-of-the-art graph learning methods on twelve data sets.

Body movements, from a short smile to a marathon run, are driven by muscle activity. Despite the fact that measuring muscle activity with electromyography (EMG) is technically well established, it is highly complex and its use in interfaces has been limited. Easy access to muscle sensing can offer new opportunities to Human-Computer Interaction (HCI) research. Off-the-shelf sensors often only provide low-level access, hence requiring expertise in signal processing and widening the gulf of execution for users without engineering skills. To address this challenge, we introduce EMBody, a data-centric toolkit for EMG-based interface prototyping and experimentation. EMBody offers multiple levels of prototyping fidelity for EMG sensing, signal processing, and data interpretation. Our data-centric toolkit encapsulates the different data representation stages, offering a wide range of customization opportunities to experts while also allowing non-technical designers to focus on creating new interaction techniques. EMBody features a lightweight form factor and wireless connectivity. Additionally, the system leverages an exploration-centered workflow by allowing rapid access to measurement data via the accompanying software. Users define a set of motions to be recognized and interactively provide example data points. The toolkit then handles signal processing and classification. The recognized movements are streamed on the local network, ready to be used by interactive applications. This paper reports on how to use EMBody and its implementation. We iteratively developed the toolkit in a series of workshops and example applications. Users who had none or very limited knowledge of EMG could rapidly create engaging functional prototypes, while experts appreciated the modularity of the software component allowing for a high degree of customization. We contribute the software and hardware components of EMBody as a tool for the research community to stimulate creative exploration of EMG systems.

Food Recommender Systems (FRS) have the potential to support informed and satisfying food choices. However, to realize their full potential, FRS must engage with the complexity of the choices people make around food. For example, while taste and ingredients are important, contextual and practical factors also play a critical role in food choice. Much of the previous literature on FRS has focused on ingredient-based recommendations, often in limited food datasets. Here we describe a broader approach, focusing on the use of Ensemble Topic Modelling (EnsTM) to support personalized recipe recommendations that implicitly capture and account for multi-domain food preferences in any food-corpus. EnsTM has the additional advantage of enabling a reduced data representation format that facilitates efficient user-modelling and recommendation. This article describes the results of two studies. The first investigated EnsTM based recommendation in a cold-start scenario. We investigated three different EnsTM based variations using a large-scale, real-world corpus of 230,876 recipes, and compared them with a conventional content-based approach. In a user study with 48 participants, EnsTM-based recommenders significantly outperformed the content-based approach. Alongside excellent coverage, they enabled an implicit understanding of users’ food preference across multiple food domains. The second study investigated the use of EnsTM in a long-term or regular-use scenario. We implemented multiple variations of feature and/or topic based hybrid recipe recommenders, which updated users’ profiles in real-time and predicted their preferences for new recipes. When compared against the current state of the art EnsTM-based recommenders performed significantly better, providing higher accuracy and coverage.

The spread of multidrug resistant organisms (MDRO) is a global healthcare challenge. Nosocomial outbreaks caused by MDRO are an important contributor to this threat. Computer-based applications facilitating outbreak detection can be essential to address this issue. To allow application reusability across institutions, the various heterogeneous microbiology data representations needs to be transformed into standardised, unambiguous data models. In this work, we present a multi-centric standardisation approach by using openEHR as modelling standard. Data models have been consented in a multicentre and international approach. Participating sites integrated microbiology reports from primary source systems into an openEHR-based data platform. For evaluation, we implemented a prototypical application, compared the transformed data with original reports and conducted automated data quality checks. We were able to develop standardised and interoperable microbiology data models. The publicly available data models can be used across institutions to transform real-life microbiology reports into standardised representations. The implementation of a proof-of-principle and quality control application demonstrated that the new formats as well as the integration processes are feasible. Holistic transformation of microbiological data into standardised openEHR based formats is feasible in a real-life multicentre setting and lays the foundation for developing cross-institutional, automated outbreak detection systems.

With the progress of financial technology (FinTech), real-time information from FinTech is huge and complicated. For various fields of research, identifying intrinsic features of complex data is important, not limited to financial big data. Reviewing previous studies, there are no suitable methods to deal with complex financial data. General methods are traditionally developed from statistics and machine learning. They are usually in some shallow model forms, which cannot fully represent complex, compositional, and hierarchical financial data features. Due to above drawbacks, this study tries to address the problem by advanced deep learning (DL) methods. In DL more layers will increase the power for abstract data representation. Recently, DL has achieved state-of-the-art performance in a wide range of tasks including speech, image, and vision. DL is effective in learning increasingly more abstract representations in a layer-wise manner. That just meets the characteristic of financial data. This study applies newly developed DL networks, the deep canonically correlation analysis and deep canonically correlated autoencoders to perform FinTech data mining. To test the proposed model, this study employed financial statement data regarding many listed high technology companies in Taiwan stock markets. The computation of deep learning is leveraged by multiple graphics processing unit. Our systems and traditional methods are compared by the same data. Empirical results showed that our systems outperform traditional techniques from statistics and machine learning.

Stream mining considers the online arrival of examples at high speed and the possibility of changes in its descriptive features or class definitions compared with past knowledge (i.e., concept drifts). The fast detection of drifts is essential to keep the predictive model updated and stable in changing environments. For many applications, such as those related to smart sensors, the high number of features is an additional challenge in terms of memory and time for stream processing. This paper presents an unsupervised and model-independent concept drift detector suitable for high-speed and high-dimensional data streams. We propose a straightforward two-dimensional data representation that allows the faster processing of datasets with a large number of examples and dimensions. We developed an adaptive drift detector on this visual representation that is efficient for fast streams with thousands of features and is accurate as existing costly methods that perform various statistical tests considering each feature individually. Our method achieves better performance measured by execution time and accuracy in classification problems for different types of drifts. The experimental evaluation considering synthetic and real data demonstrates the method’s versatility in several domains, including entomology, medicine, and transportation systems.

This article introduces a scheme for semi-supervised learning by estimating a flexible non-linear data representation that exploits Spectral Graph Convolutions structure. Structured data are exploited in order to determine non-linear and linear models. The introduced scheme takes advantage of data-driven graphs at two levels. First, it incorporates manifold smoothness that is naturally encoded by the graph itself. Second, the regression model is built on the convolved data samples that are derived from the data and their associated graph. The proposed semi-supervised embedding can tackle the problem of over-fitting on neighborhood structures for image data. The proposed Graph Convolution-based Semi-supervised Embedding paves the way to new theoretical and application perspectives related to the non-linear embedding. Indeed, building flexible models that adopt convolved data samples can enhance both the data representation and the final performance of the learning system. Several experiments are conducted on six image datasets for comparing the introduced scheme with some state-of-the-art semi-supervised approaches. This empirical evaluation shows the effectiveness of the proposed embedding scheme.

Data representation plays a crucial role in semi-supervised learning. This paper proposes a framework for semi-supervised data representation. It introduces a flexible nonlinear embedding model that integrates graph-based data convolutions. The proposed approach exploits structured data in order to estimate a nonlinear data representation as well as a linear transformation, enabling an inductive semi-supervised model. The introduced approach exploits data graphs at two different levels. First, it integrates manifold regularization that is encoded by the graph itself. Second, it optimizes a flexible linear transformation that maps the convolved data samples to their nonlinear representations. These convolved data are generated by the joint use of the graph and data. The proposed semi-supervised model overcomes some challenges related to some samples distributions in the original spaces. The proposed Graph Convolution based Semi-supervised Embedding (GCSE) provides flexible models which can improve both the data representation and the final performance of the learning model. Experiments are run on six image datasets for comparing the proposed approach with several state-of-art semi-supervised methods. These results show the effectiveness of the proposed framework.

With the advent of the era of big data, data mining has become one of the key technologies in the field of research and business. In order to improve the efficiency of data mining, this paper studies data mining based on the intelligent recommendation system. Firstly, this paper makes mathematical modeling of the intelligent recommendation system based on association rules. After analyzing the requirements of the intelligent recommendation system, Java 2 Platform, Enterprise Edition, technology is used to divide the system architecture into the presentation layer, business logic layer, and data layer. Recommendation module is divided into three substages: data representation, model learning, and recommendation engine. Then, the fuzzy clustering algorithm is used to optimize the system. After the system is built, the performance of the system is evaluated, and the evaluation indexes include accuracy, coverage, and response time. Finally, the system is put into a trial operation of an e-commerce platform. The click-through rate and purchase conversion rate of recommended products before and after the operation are compared, and a questionnaire survey is randomly launched to the platform users to analyze the user satisfaction. The experimental data show that the MAE of this system is the lowest, maintained at about 0.73, and its accuracy is the highest; before the recommended threshold exceeds 0.5, the average coverage rate of this system is the highest: 0.75; in Q1–Q5 subsets, the shortest response time of the system is 0.2 s. Before and after the operation of the system, the average click-through rate increased by 11.04%, and the average purchase rate increased by 9.35%. Among the 1216 users, 43% of the users were satisfied with 4 and 9% with 1. This shows that the system algorithm convergence speed is fast; it can recommend products more in line with user needs and interests and promote higher click-through rate and purchase rate, but user satisfaction can be further improved.

The development of wireless sensor networks is currently very rapid in various sectors. One of the implementations of the wireless sensor network is for monitoring temperature and humidity. In monitoring a room or a large area, many nodes are installed at a certain point. With the number of senses by each node so that one data that can represent the area is needed. To get one data that can represent the data in that area, we need a consensus method. In this study, the average distributed consensus method was used to obtain data representation from an area's wireless sensor network. After testing and simulating the ideal range between WSN nodes is 30 meters where the average percentage of detection errors reaches 0%. And the maximum capability is 33 meters with an average percentage of error detection of 46%. If given a barrier between WSN nodes, the distance that can be reached becomes shorter, with an ideal distance of 16 meters with an average percentage of error detection of 0% and the farthest distance reaching 21 meters with a detection error of 50%. The results of the consensus method simulation show that the more slots are filled with consensus data about the results, the faster and easier convergence will be achieved, in this case the monitoring results represent the real situation.

This article provides a detailed and visual presentation of the derivations of and relationships between many of the commonly employed functional forms of real- and reciprocal-space data employed by the worldwide total scattering community., The total scattering method is the simultaneous study of both the real- and reciprocal-space representations of diffraction data. While conventional Bragg-scattering analysis (employing methods such as Rietveld refinement) provides insight into the average structure of the material, pair distribution function (PDF) analysis allows for a more focused study of the local atomic arrangement of a material. Generically speaking, a PDF is generated by Fourier transforming the total measured reciprocal-space diffraction data (Bragg and diffuse) into a real-space representation. However, the details of the transformation employed and, by consequence, the resultant appearance and weighting of the real-space representation of the system can vary between different research communities. As the worldwide total scattering community continues to grow, these subtle differences in nomenclature and data representation have led to conflicting and confusing descriptions of how the PDF is defined and calculated. This paper provides a consistent derivation of many of these different forms of the PDF and the transformations required to bridge between them. Some general considerations and advice for total scattering practitioners in selecting and defining the appropriate choice of PDF in their own research are presented. This contribution aims to benefit people starting in the field or trying to compare their results with those of other researchers.

In the recent past, deep learning has received remarkable acceptance in real-world applications. Social computing expands the existing notion of cyber space and physical space to a more advance cyber–physical–social system (CPSS). Therefore, deep learning provides a propitious technique for accurate mining of information from CPSS, thus facilitates CPSS to offer services of exceptional quality efficiently. However, most of the current deep learning methods are struggling to keep up with the ever-increasing heterogeneous and highly nonlinear dissemination of data. Furthermore, the advancement of deep learning presents privacy concerns. This article proposes a deep private tensor autoencoder (dPTAE), where tensors are used for data representation, and differential privacy guarantees strong privacy. The core idea of our work is to enforce differential privacy through noise injection into the objective functions instead of the results they produce. In addition, the proposed method preserves the privacy of information shared amongst CPSS in smart environments. We applied dPTAE on three representative data sets. Rigorous experimental evaluations and theoretical analysis demonstrate that dPTAE is significantly effective and efficient.

To address the problem of ever-growing scientific data sizes making data movement a major hindrance to analysis, we introduce a novel encoding for scalar fields: a unified tree of resolution and precision, specifically constructed so that valid cuts correspond to sensible approximations of the original field in the precision-resolution space. Furthermore, we introduce a highly flexible encoding of such trees that forms a parameterized family of data hierarchies. We discuss how different parameter choices lead to different trade-offs in practice, and show how specific choices result in known data representation schemes such as zfp [52] , idx [58] , and jpeg 2000 [76] . Finally, we provide system-level details and empirical evidence on how such hierarchies facilitate common approximate queries with minimal data movement and time, using real-world data sets ranging from a few gigabytes to nearly a terabyte in size. Experiments suggest that our new strategy of combining reductions in resolution and precision is competitive with state-of-the-art compression techniques with respect to data quality, while being significantly more flexible and orders of magnitude faster, and requiring significantly reduced resources.

The training of Deep Neural Networks (DNNs) brings enormous memory requirements and computational complexity, which makes it a challenge to train DNN models on resource-constrained devices. Training DNNs with reduced-precision data representation is crucial to mitigate this problem. In this article, we conduct a thorough investigation on training DNNs with low-bit posit numbers, a Type-III universal number (Unum). Through a comprehensive analysis of quantization with various data formats, it is demonstrated that the posit format shows great potential to be employed in the training of DNNs. Moreover, a DNN training framework using 8-bit posit is proposed with a novel tensor-wise scaling scheme. The experiments show the same performance as the state-of-the-art (SOTA) across multiple datasets (MNIST, CIFAR-10, ImageNet, and Penn Treebank) and model architectures (LeNet-5, AlexNet, ResNet, MobileNet-V2, and LSTM). We further design an energy-efficient hardware prototype for our framework. Compared to the standard floating-point counterpart, our design achieves a reduction of 68, 51, and 75 percent in terms of area, power, and memory capacity, respectively.

Sparse coding (SC), due to its thorough theoretical property and outstanding effectiveness, is attracting more and more attention in various data representation and data mining applications. However, the optimization of most existing sparse coding algorithms are non-convex and thus prone to become stuck into bad local minima under the framework of alternative optimization, especially when there are many outliers and noisy data. To enhance the learning robustness, in this study, we will present an unified framework named Self-Paced Sparse Coding (SPSC), which gradually includes data into the learning process of SC from easy ones to complex ones by incorporating self-paced learning methodology. It implements a soft instance selection accordingly rather than a heuristic hard strategy sample selection. We also generalize the self-paced learning schema into different levels of dynamic selection on instances, features and elements respectively. Further, we show an optimization algorithm to solve it and a theoretical explanation to analyze the effectiveness of it. Extensive experimental results on the real-world clean image datasets and images with two kinds of corruptions demonstrate the remarkable robustness of the proposed method for high dimensional data representation on image clustering and reconstruction tasks over the state-of-the-arts.

Biological development is often described as a dynamic, emergent process. Yet beyond the observation of gene expression in individual cells, it is hard to conceptualize large-scale patterns that confirm this description. We provide an example of combining theoretical insights with a data science approach. The availability of quantitative data allows us to examine aggregate trends across development, from the spatial organization of embryo cells to the temporal trends as they differentiate. The first half of this paper lays out alternatives to the gene-centric view of development: namely, the view that developmental genes and their expression determine the complexity of the developmental phenotype. Caenorhabditis elegans biology provides us with a highly-deterministic developmental cell lineage and clear linkage between zygote and cells of the adult phenotype. These properties allow us to examine time-dependent properties of the embryonic phenotype. We utilize the unique life-history properties of C. elegans to demonstrate how these emergent properties can be linked together by relational processes and data analysis. The second half of this paper focuses on the process of developmental cell terminal differentiation, and how terminally-differentiated cells contribute to structure and function of the adult phenotype. An analysis is conducted for cells that were present during discrete time intervals covering 200 to 400 minutes of embryogenesis, providing us with basic statistics on the tempo of the process in addition to the appearance of specific cell types and their order relative to developmental time. As with ideas presented in the first section, these data may also provide clues as to the timing for the initial onset of stereotyped and autonomic behaviors of the developing animal. Taken together, these overlapping approaches can provide critical links across life-history, anatomy and function.

Since the time series data have the characteristics of a large amount of data and non-stationarity, we usually cannot obtain a satisfactory result by a single-model-based method to detect anomalies in time series data. To overcome this problem, in this paper, a combination-model-based approach is proposed by combining a similarity-measurement-based method and a model-based method for anomaly detection. First, the process of data representation is performed to generate a new data form to arrive at the purpose of reducing data volume. Furthermore, due to the anomalies being generally caused by changes in amplitude and shape, we take both the original time series data and their amplitude change data into consideration of the process of data representation to capture the shape and morphological features. Then, the results of data representation are employed to establish a model for anomaly detection. Compared with the state-of-the-art methods, experimental studies on a large number of datasets show that the proposed method can significantly improve the performance of anomaly detection with higher data anomaly resolution.

In modern transportation systems, an enormous amount of traffic data is generated every day. This has led to rapid progress in short-term traffic prediction (STTP), in which deep learning methods have recently been applied. In traffic networks with complex spatiotemporal relationships, deep neural networks (DNNs) often perform well because they are capable of automatically extracting the most important features and patterns. In this study, we survey recent STTP studies applying deep networks from four perspectives. 1) We summarize input data representation methods according to the number and type of spatial and temporal dependencies involved. 2) We briefly explain a wide range of DNN techniques from the earliest networks, including Restricted Boltzmann Machines, to the most recent, including graph-based and meta-learning networks. 3) We summarize previous STTP studies in terms of the type of DNN techniques, application area, dataset and code availability, and the type of the represented spatiotemporal dependencies. 4) We compile public traffic datasets that are popular and can be used as the standard benchmarks. Finally, we suggest challenging issues and possible future research directions in STTP.

This paper proposes a new fiber Bragg grating central wavelength interrogation system by combining evolutionary algorithm and machine learning techniques integrated with an unsupervised autoencoder (AE) pre-training strategy. The proposed unsupervised AE pre-training convolution neural network (CNN) allows training of the convolutional layers independently from a regression task in order to learn a new data representation and give better generalization. It is also used to improve the system accuracy by four times without extra-labeled data. Moreover, AE is combined with a differential evolutionary (DE) algorithm to automate the human labeling task. The proposed autoencoder pre-training convolution neural network and differential evolutionary (AECNNDE) interrogation system achieve good accuracy and can speed-up the computational time by a maximum of 30 times than DE algorithm.

In the last decade, data generated from different digital devices has posed a remarkable challenge for data representation and analysis. Because of the high-dimensional datasets and the rapid growth of data volume, a lot of challenges have been encountered in various fields such as data mining and data science. Conventional machine learning classifiers are of limited ability to handle the problems of high dimensionality that includes memory limitation, computational cost, and low accuracy performance. Consequently, there is a need to reduce the dimension of datasets by choosing the most significant features that would represent the data efficiently with minimum volume. This study proposes an improved binary version of the equilibrium optimizer algorithm (IBEO) to mitigate features selection problem. Two main enhancements are added to the original equilibrium optimizer (EO) to strengthen its performance. Opposition based learning is the first advancement added to the initialization stage of EO to enhance the diversity of the population in the search space. Local search algorithm is the second advancement added to enhance the exploitation of EO. Wrapper approaches can offer premium solutions. Thus, we used k-nearest neighbour classifier and support vector machine classifiers as the most popular wrapper methods. Moreover, dealing with the problem of over-fitting is an essential task that urges on applying k-fold cross-validation to split each dataset into training and testing data. Comparative tests with different well-known algorithms such as grey wolf optimization, grasshopper optimization, particle swarm optimization, whale optimization, dragonfly, and improved salp swarm algorithms are considered. The proposed algorithm is applied to the most commonly datasets used in the field to validate the performance. Statistical analysis studies demonstrate the effectiveness of the IBEO.

Variational autoencoders (VAEs) are deep latent space generative models that have been immensely successful in multiple exciting applications in biomedical informatics such as molecular design, protein design, medical image classification and segmentation, integrated multi-omics data analyses, and large-scale biological sequence analyses, among others. The fundamental idea in VAEs is to learn the distribution of data in such a way that new meaningful data with more intra-class variations can be generated from the encoded distribution. The ability of VAEs to synthesize new data with more representation variance at state-of-art levels provides hope that the chronic scarcity of labeled data in the biomedical field can be resolved. Furthermore, VAEs have made nonlinear latent variable models tractable for modeling complex distributions. This has allowed for efficient extraction of relevant biomedical information from learned features for biological data sets, referred to as unsupervised feature representation learning. In this article, we review the various recent advancements in the development and application of VAEs for biomedical informatics. We discuss challenges and future opportunities for biomedical research with respect to VAEs.

Numerous technologies exist for promoting a healthier lifestyle. These technologies collectively referred to as “Behavior Change Support Systems”. However, the majority of existing apps use quantitative data representation. Since it is difficult to understand the meaning behind quantitative data, this approach has been suggested to lower users’ motivation and fail to promote behavior change. Therefore, an interpretation of quantitative data needs to be provided as a supplement. However, different descriptions of the same data may lead to different outcomes. In this paper, we explore the impact of different communication styles for interpretations of quantitative data on behavior change by developing and evaluating Walkeeper – a web-based app that provides interpretations of the users’ daily step counts using different levels of elaborateness and indirectness with the aim of promoting walking. Through the quantitative analysis and results of a user study, we contribute new knowledge on designing such interpretations for quantitative data.

This paper presents a further development of Grafana project using LoRa media and chat software in the frame of control system. The combination of open SCADA for four process values monitoring and chat programs is the new. In this way a bot is created for sending predefined notifications as soon as occur in system and published. The so-called notification channel has been created for temperature, humidity, pressure and gas sensor for broad range of gases. The values are received by the BME 680 as a feedback in control system structure. The algorithm and programing to ensure the date exchange in real system are done. The verification, response and obtained messages, are presented as a result. The benefit is that instant messages with convenient data representation are achieved for chosen relevant time interval. This an example of integration between Grafana and messenger apps in new way. Due to integration of Grafana, using LoRa, and chats, the alarms can be set and notification can be transmitted over huge distances.

Nonnegative Matrix Factorization (NMF) is a popular technique for machine learning. Its power is that it can decompose a nonnegative matrix into two nonnegative factors whose product well approximates the nonnegative matrix. However, the nonnegative constraint of the data matrix limits its application. Additionally, the representations learned by NMF fail to respect the intrinsic geometric structure of the data. In this paper, we propose a novel unsupervised matrix factorization method, called Robust Structured Convex Nonnegative Matrix Factorization (RSCNMF). RSCNMF not only achieves meaningful factorizations of the mixed-sign data, but also learns a discriminative representation by leveraging local and global structures of the data. Moreover, it introduces the L2,1-norm loss function to deal with noise and outliers, and exploits the L2,1-norm feature regularizer to select discriminative features across all the samples. We develop an alternate iterative scheme to solve such a new model. The convergence of RSCNMF is proven theoretically and verified empirically. The experimental results on eight real-world data sets show that our RSCNMF algorithm matches or outperforms the state-of-the-art methods.

Data is the key to the operation of manufacturing workshop Digital Twin System (DTS) and also supports the top-level applications of DTS. However, manufacturing data has the characteristics of coupling and large-amount, which lead to inefficient and inaccurate operations of applications. In order to provide stable and efficient data support for the applications of DTS, this paper proposed a data construction method. The framework of data construction is designed based on the functional requirements, which are analyzed according to the characteristics of manufacturing data. Then, module implementation of the framework, including data representation module, data organization module and data management module, is introduced in detail. Finally, application of cutting tool wear prediction is taken as a case study to show the feasibility and effectiveness of the proposed data construction method.

In this era handling uncertainty its significance is considered as one of the crucial tasks for the dark data analysis. In this process problem arises that the data and its uncertainty exists beyond fourth dimensions. To deal with it “Turiyam set” is introduced at Int'l Conference on Operations Research and Applications (ORA 2021) which held at Guilin, China. In this talk the precise representation of fourth dimension data in turiyam context and its application for data analysis will be discussed. This logic is motivated from Yin-Yang theory of Chinese Taoism as well as Sanskrit Yamma-Yammi. One of the most suitable examples is medical diagnoses data set. Let us suppose COVID 19 data which contain fourth dimensions uncertainty for the analysis. The people who got recovered can be considered as true regions (t), people who died due to COVID 19 can be con-sidered as false regions (f), people who are still active can be considered as indeterminacy (i), people who got vaccinated can be considered as Turiya or Liberated state (l). The refusal degree means people who still did not come under these regions can be represented as 1- ( t + i + f + l ). This speech will discuss the analysis of fourth dimension data using the Turiyam context.

With the breakthrough of edge intelligence, we are witnessing a booming increase in distributed applications on edge nodes. These distributed applications need to apply a novel data representation algorithm to support data-information exchanging and data-information based decision among different edge nodes. As the most efficient data compact representation algorithm, Counting Bloom Filter (CBF) is an extension of Bloom filter, which enables updating data representation as well as inserting data into a representation. To facilitate distributed applications on edge nodes, edge nodes need to exchange and summarize the information of the data collected from different edge nodes. Impossible to merge with other CBFs, the existing CBF and its variants thus cannot be used for representing and exchanging data information among edge nodes. To handle this problem, we design a novel mergeable CBF, mergeCBF. Based on an insight about the counting processing of a CBF, we unfold the counter array of the conventional CBF to a group of bit arrays, and in order to support merging multiple filters, map each inputted item to the cells in this group of cuckoo-scheduled bit arrays instead of the counters in CBF. Experiments on real-world datasets demonstrate that mergeCBF can support conventional operations and merging operations in an efficient way without degrading the quality of the representation results.

Data management systems impose structure on data via a static representation schema or data structure. Information from the data is extracted by executing queries based on predefined operators. This paradigm restricts the searchability of the data to concepts and relationships that are known or assumed to exist among the objects. While this is an effective and efficient means of retrieving simple information, we propose that such a structure severely limits the ability to derive breakthrough knowledge that exists in data under the guise of “unknown unknowns.” A dynamic system will alleviate this dependence, allowing theoretically infinite projections of the data to reveal discoverable relationships that are hidden by traditional use case-driven, static query systems. In this paper, we propose a framework for a data-responsive query algebra based on a dynamic hyperbolic surface model. Such a model could provide more intuitive access to analytics and insights from massive, aggregated datasets than existing methods. This model will significantly alter the means of addressing the underlying data by representing it as an arrangement on a dynamic, hyperbolic plane. Consequently, querying the data can be viewed as a process similar to quantum annealing, in terms of characterizing data representation as an energy minimization problem with numerous minima.

Electronic text stylometry is concerned with analyzing the writing styles of input electronic texts to extract information about their authors. For example, such extracted data could be the authors’ identity or other aspects, such as their gender and age group. This survey paper presents the following contributions: 1) A description of all stylometry problems in probability terms, under a unified notation. 2) A survey of data representation (or feature extraction) methods. 3) A comprehensive evaluation of 23, 760 feature extraction methods followed by a thorough discussion of the results. This extensive evaluation is critical since the known data representation methods are often not evaluated under the same unified testbed.

LIDAR and depth cameras have gone through a profound technological evolution, making large-scale recording of 3D point cloud data possible which raises new challenges for data processing. Most of the existing 3D point cloud processing methods were developed to work properly when the entire data set fits into the memory of a single server. When point clouds are significantly larger than the main memory and data are only available on slow storage, new approaches are necessary. In this paper, we propose a DBMS-based point cloud processing pipeline that solves the template matching problem, i.e., finding the – potentially multiple – occurrences of a small query point cloud in an extensive scene data set that is preprocessed and stored in a database. The storage layer uses a compact and novel data representation to exploit the benefits of efficient indexing structures whereas the query algorithm consists of a novel combination of existing point cloud processing and matching methods. To the best of our knowledge, this is the first template matching proposal in the literature that exploits the benefits of databases.

There has recently been a rapid progress in computational methods for determining protein targets of small molecule drugs, which will be termed as compound protein interaction (CPI). In this review, we comprehensively review topics related to computational prediction of CPI. Data for CPI has been accumulated and curated significantly both in quantity and quality. Computational methods have become powerful ever to analyze such complex the data. Thus, recent successes in the improved quality of CPI prediction are due to use of both sophisticated computational techniques and higher quality information in the databases. The goal of this article is to provide reviews of topics related to CPI, such as data, format, representation, to computational models, so that researchers can take full advantages of these resources to develop novel prediction methods. Chemical compounds and protein data from various resources were discussed in terms of data formats and encoding schemes. For the CPI methods, we grouped prediction methods into five categories from traditional machine learning techniques to state-of-the-art deep learning techniques. In closing, we discussed emerging machine learning topics to help both experimental and computational scientists leverage the current knowledge and strategies to develop more powerful and accurate CPI prediction methods.

This paper describes a new library for learning Bayesian networks from data containing discrete and continuous variables (mixed data). In addition to the classical learning methods on discretized data, this library proposes its algorithm that allows structural learning and parameters learning from mixed data without discretization since data discretization leads to information loss. This algorithm based on mixed MI score function for structural learning, and also linear regression and Gaussian distribution approximation for parameters learning. The library also offers two algorithms for enumerating graph structures - the greedy Hill-Climbing algorithm and the evolutionary algorithm. Thus the key capabilities of the proposed library are as follows: (1) structural and parameters learning of a Bayesian network on discretized data, (2) structural and parameters learning of a Bayesian network on mixed data using the MI mixed score function and Gaussian approximation, (3) launching learning algorithms on one of two algorithms for enumerating graph structures - Hill-Climbing and the evolutionary algorithm. Since the need for mixed data representation comes from practical necessity, the advantages of our implementations are evaluated in the context of solving approximation and gap recovery problems on synthetic data and real datasets.

Coordinating of energy networks to form a city-level multidimensional integrated energy system becomes a new trend in Energy Internet (EI). The collaborating in the information layer is a core issue to achieve smart integration. However, the heterogeneity of multiagent data, the volatility of components, and the real-time analysis requirement in EI bring significant challenges. To solve these problems, in this article we propose a stream processing framework based on linked data for information collaboration among multiple energy networks. The framework provides a universal data representation based on linked data and semantic relation discovery approach to model and semantically fuse heterogeneous data. Semantics-based information transmission contracts and channels are automatically generated to adapt to structural changes in EI. A multimodel-based dynamic adjusting stream processing is implemented using data semantics. A real-world case study is implemented to demonstrate the adaptability, feasibility, and flexibility of the proposed framework.

In this essay, we discuss how tables can be used to ensure—and reassure about—trustworthiness in qualitative research. We posit that in qualitative research, tables help not only increase transparency about data collection, analysis, and findings, but also—and no less importantly—organize and analyze data effectively. We present some of the tables most frequently used by qualitative researchers, explain their uses, discuss how they enhance trustworthiness, and provide illustrative examples to inspire readers in their use of tables in their own research.

Semantic segmentation of point cloud data, an efficient 3-D scattered point representation, is a fundamental task for various applications, such as autonomous driving and 3-D telepresence. In recent years, deep learning techniques have achieved significant progress in semantic segmentation, especially in the 2-D image setting. However, due to the irregularity of point clouds, most of them cannot be applied to this special data representation directly. While recent works are able to handle the irregularity problem and maintain the permutation invariance, most of them fail to capture the valuable high-dimensional local feature in fine granularity. Inspired by fuzzy mathematical methods and the analysis on the drawbacks of current state-of-the-art works, in this article, we propose a novel deep neural model, Fuzzy3DSeg, that is able to directly feed in the point clouds while maintaining invariant to the permutation of the data feeding order. We deeply integrate the learning of the fuzzy neighborhood feature of each point into our network architecture, so as to perform operations on high-dimensional features. We demonstrate the effectiveness of this network architecture level integration, compared with methods of the fuzzy data preprocessing cascading neural network. Comprehensive experiments on two challenging datasets demonstrate that the proposed Fuzzy3DSeg significantly outperforms the state-of-the-art methods.

Data preprocessing is an important component of machine learning pipelines, which requires ample time and resources. An integral part of preprocessing is data transformation into the format required by a given learning algorithm. This paper outlines some of the modern data processing techniques used in relational learning that enable data fusion from different input data types and formats into a single table data representation, focusing on the propositionalization and embedding data transformation approaches. While both approaches aim at transforming data into tabular data format, they use different terminology and task definitions, are perceived to address different goals, and are used in different contexts. This paper contributes a unifying framework that allows for improved understanding of these two data transformation techniques by presenting their unified definitions, and by explaining the similarities and differences between the two approaches as variants of a unified complex data transformation task. In addition to the unifying framework, the novelty of this paper is a unifying methodology combining propositionalization and embeddings, which benefits from the advantages of both in solving complex data transformation and learning tasks. We present two efficient implementations of the unifying methodology: an instance-based PropDRM approach, and a feature-based PropStar approach to data transformation and learning, together with their empirical evaluation on several relational problems. The results show that the new algorithms can outperform existing relational learners and can solve much larger problems., Accepted in MLJ

This paper introduces a scheme for semi-supervised data representation. It proposes a flexible nonlinear embedding model that imitates the principle of spectral graph convolutions. Structured data are exploited in order to determine nonlinear and linear models. The introduced scheme takes advantage of data graphs at two different levels. First, it incorporates manifold regularization that is naturally encoded by the graph itself. Second, the regression model is built on the convolved data samples that are obtained by the joint use of the data and their associated graph. The proposed semi-supervised embedding can tackle challenges related to over-fitting in image data spaces. The proposed graph convolution-based semi-supervised embedding paves the way to new theoretical and application perspectives related to the nonlinear embedding. Indeed, building flexible models that adopt convolved data samples can enhance both the data representation and the final performance of the learning system. Several experiments are conducted on six image datasets for comparing the introduced scheme with many state-of-art semi-supervised approaches. These experimental results show the effectiveness of the introduced data representation scheme.

For many machine learning algorithms, their success heavily depends on data representation. In this paper, we present an ${\ell }_{2,1}$ -norm constrained canonical correlation analysis (CCA) model, that is, $ {L}_{{2,1}}$ -CCA, toward discovering compact and discriminative representation for the data associated with multiple views. To well exploit the complementary and coherent information across multiple views, the ${\ell }_{{2,1}}$ -norm is employed to constrain the canonical loadings and measure the canonical correlation loss term simultaneously. It enables, on the one hand, the canonical loadings to be with the capacity of variable selection for facilitating the interpretability of the learned canonical variables, and on the other hand, the learned canonical common representation keeps highly consistent with the most canonical variables from each view of the data. Meanwhile, the proposed ${L}_{{2,1}}$ -CCA can also be provided with the desired insensitivity to noise (outliers) to some degree. To solve the optimization problem, we develop an efficient alternating optimization algorithm and give its convergence analysis both theoretically and experimentally. Considerable experiment results on several real-world datasets have demonstrated that ${L}_{{2,1}}$ -CCA can achieve competitive or better performance in comparison with some representative approaches for multiview representation learning.

Analysing the data of a deadly pandemic that has created a mess in this wonderful world and caused a lot of deaths is a need of this hour such that we can easily take preventive measures and hold this pandemic growth and eradicate it with certain measures and proper planning and a study is needed to analyse whether any inter-mediate hosts have facilitated the transmission of the virus to humans or vice versa and this could only be done if precise data is analysed [1][6]. The main objective behind writing this paper is to present an idea that how knowledge to an emerging frontend technology like React JS can lead to a wonderful user interface that can serve as a data analyser web software for this deadly pandemic. The paper revolves around a project created with a bunch of features of React JS from exciting frontend components with Material UI to writing CSS, JSX and making API calls to collect worldwide data related to Corona virus [9]. The major highlights of the project entitled in this paper are no page reloads, Responsiveness, body-parsing, API calls for data collection, JSX (HTML inside JS), sorting algorithms, dynamic data representation, REDUX, data representation in dynamic graphs/pie charts. In this a large number of npm packages like react, react-dom, @material-ui etc along with React JS and CSS library like Material UI has been used in order to code the website such that it results in an attractive, responsive and a beautiful project with a single technology that is JS. The web software entitled in this paper is a complete package of features of React JS integrated with some other frontend technologies like CSS and material UI going through which any user can analyse the data related to corona virus geographically as well as graphically.

The data representation as well as naming conventions used in commercial screen files by different companies make the automated analysis of crystallization experiments difficult and time-consuming. In order to reduce the human effort required to deal with this problem, we present an approach for computationally matching elements of two schemas using linguistic schema matching methods and then transform the input screen format to another format with naming defined by the user. This approach is tested on a number of commercial screens from different companies and the results of the experiments showed an overall accuracy of 97 percent on schema matching which is significantly better than the other two matchers we tested. Our tool enables mapping a screen file in one format to another format preferred by the expert using their preferred chemical names.

The problems caused by network dimension disasters and computational complexity have become an important issue to be solved in the field of social network research. The existing methods for network feature learning are mostly based on static and small-scale assumptions, and there is no modified learning for the unique attributes of social networks. Therefore, existing learning methods cannot adapt to the dynamic and large-scale of current social networks. Even super large scale and other features. This paper mainly studies the feature representation learning of large-scale dynamic social network structure. In this paper, the positive and negative damping sampling of network nodes in different classes is carried out, and the dynamic feature learning method for newly added nodes is constructed, which makes the model feasible for the extraction of structural features of large-scale social networks in the process of dynamic change. The obtained node feature representation has better dynamic robustness. By selecting the real datasets of three large-scale dynamic social networks and the experiments of dynamic link prediction in social networks, it is found that DNPS has achieved a large performance improvement over the benchmark model in terms of prediction accuracy and time efficiency. When the α value is around 0.7, the model effect is optimal.

Clustering is an essential data analysis technique and has been studied extensively over the last decades. Previous studies have shown that data representation and data structure information are two critical factors for improving clustering performance, and it forms two important lines of research. The first line of research attempts to learn representative features, especially utilizing the deep neural networks, for handling clustering problems. The second concerns exploiting the geometric structure information within data for clustering. Although both of them have achieved promising performance in lots of clustering tasks, few efforts have been dedicated to combine them in a unified deep clustering framework, which is the research gap we aim to bridge in this work. In this paper, we propose a novel approach, Manifold regularized Deep Embedded Clustering (MDEC), to deal with the aforementioned challenge. It simultaneously models data generating distribution, cluster assignment consistency, as well as geometric structure of data in a unified framework. The proposed method can be optimized by performing mini-batch stochastic gradient descent and back-propagation. We evaluate MDEC on three real-world datasets (USPS, REUTERS-10K, and MNIST), where experimental results demonstrate that our model outperforms baseline models and obtains the state-of-the-art performance.

In the era of data-intensive computing, large-scale applications, in both scientific and the BigData communities, demonstrate unique I/O requirements leading to a proliferation of different storage devices and software stacks, many of which have conflicting requirements. Further, new hardware technologies and system designs create a hierarchical composition that may be ideal for computational storage operations. In this article, we investigate how to support a wide variety of conflicting I/O workloads under a single storage system. We introduce the idea of a Label , a new data representation, and, we present LABIOS: a new, distributed, Label- based I/O system. LABIOS boosts I/O performance by up to 17× via asynchronous I/O, supports heterogeneous storage resources, offers storage elasticity, and promotes in situ analytics and software defined storage support via data provisioning. LABIOS demonstrates the effectiveness of storage bridging to support the convergence of HPC and BigData workloads on a single platform.

Deep learning methods are characterized by their capacity to learn data representation compared to the traditional machine learning algorithms. However, these methods are prone to overfitting on small volumes of data. The objective of this research is to overcome this limitation by improving the generalization in the proposed deep learning framework based on various techniques: data augmentation, small models, optimizer selection, and ensemble learning. For ensembling, the authors used selected models from different checkpoints and both voting and unweighted average methods for combination. The experimental study on the lymphomas histopathological dataset highlights the efficiency of the MobileNet2 network combined with the stochastic gradient descent (SGD) optimizer in terms of generalization. The best results have been achieved by the combination of the best three checkpoint models (98.67% of accuracy). These findings provide important insights into the efficiency of the checkpoint ensemble learning method for histopathological image classification.

Grinding is a key process in machining, which has direct impact on the accuracy, performance and service life of the finished workpiece. With the rapid development and popularization of computers and information technology, the intelligentization of grinding technology has become an important research topic. Especially with the explosive growth of industry data, grinding process data has become the core element in intelligent manufacturing. To understand the connotation, status quo and development trends of intelligent technology in the grinding process, this paper takes the relationship between grinding process data and intelligent manufacturing as perspective. A research structure of data-driven mode is built in the grinding process, followed by the basic concepts and characteristics of grinding data. Relevant researches that maybe beneficial to the proposed research structure are reviewed in five layers, which include data acquisition layer, data processing and fusion layer, data mining and analysis layer, data representation and evolution layer, and data service layer. Finally, the current data-driven mode and future trends in this field are summarized.

An increasing number of vehicles are now equipped with GPS devices to facilitate fleet management and send their GPS locations continuously, generating a huge volume of trajectory data. Sending and storing such vehicle trajectory data cause sustainable communication and storage overheads. Trajectory data compression becomes a promising way to alleviate overhead issues. However, previous solutions are commonly carried out at the side of the data center after data having been received, thus saving the storage cost only. Here, we bring the idea of mobile edge computing and transfer the computation-intensive data compression task to the mobile devices of drivers. As a result, the trajectory data is reduced at the side of data generators before being sent out; thus, it can lower data communication and storage costs simultaneously. We propose DAVT , an error-bounded trajectory data representation, and a compression framework. Specifically, the trajectory data is reformatted into three parts (i.e., D istance, A cceleration & V elocity, and T ime), and three compressors are wisely devised to compress each part. For D and AV parts, a similar Huffman tree-forest structure is exploited to encode data elements effectively, but with quite different rationales. For the T part, the large absolute timestamps are transformed to small time intervals firstly, and different encoding techniques are adopted based on the data quality. We evaluate our proposed system using a large-scale taxi trajectory dataset collected from the city of Beijing, China. Our results show that our compressors outperform other baselines.

Graphical abstract, Highlights • Generalization of the periodic table of elements. • Inclusion of fractional order elements. • 2- and 3-dimensional maps of elements organized accordingly to their features., This paper combines the synergies of three mathematical and computational generalizations. The concepts of fractional calculus, memristor and information visualization extend the classical ideas of integro-differential calculus, electrical elements and data representation, respectively. The study embeds these notions in a common framework, with the objective of organizing and describing the "continuum" of fractional order elements (FOE). Each FOE is characterized by its behavior, either in the time or in the frequency domains, and the differences between the FOE are captured by a variety of distinct indices, such as the Arccosine, Canberra, Jaccard and Sørensen distances. The dissimilarity information is processed by the multidimensional scaling (MDS) computational algorithm to unravel possible clusters and to allow a direct pattern visualization. The MDS yields 3-dimensional loci organized according to the FOE characteristics both for linear and nonlinear elements. The new representation generalizes the standard Cartesian 2-dimensional periodic table of elements.

The organisational data models that support the information needs of utility network managers are proprietary and domain-specific, while the emerging national standards in this field often lack lifecycle data representation capabilities. However, multiple types of utility networks can be comprehensively represented with the free and open-source Utility Network Application Domain Extension (ADE) of the international standard CityGML. The Operation & Maintenance (O&M) Domain Ontology is a proposed extended version of the Utility Network ADE that allows for consistent and comprehensive processing, storage and exchange of O&M-related utility network data. So far, this ontology has not yet been implemented in a spatial-relational database. Consequently, the support it offers during routine utility asset management tasks has remained untested. This paper, therefore, tests the support of the O&M domain ontology for asset management and proposes a database implementation of this data model. To this end, it models and loads two utility networks from the campus of the University of Twente, the Netherlands. It tests the ontology’s support for asset management by simulating a street reconstruction project and retrieving necessary project information in relation to a utility’s (a) maintenance history and performance, and (b) site conditions and valve locations. Results show that the implemented model supports projects with rapid, comprehensive, and consistent information about semantic details of utilities. Such data needs yet to be collected and registered systematically to enable future data-driven asset management practices.