Feature Analysis Model Research Articles

Research on Block Image Texture Retrieval Method Based on Depth Hash

In order to improve the retrieval ability of super-resolution multi-space block images, a texture retrieval method of block images based on depth hash is proposed. A texture feature analysis model of super-resolution multi-dimensional partitioned images is constructed, which combines texture spatial structure mapping method to realize depth information fusion of partitioned images, adopts edge feature detection and texture sparse feature clustering to realize texture hierarchical structure feature decomposition of super-resolution multi-dimensional partitioned images, and adopts deep image parameter analysis method to construct pixel structure recombination model of multi-dimensional partitioned images. Multi-dimensional texture parameter structure analysis and information clustering are realized for the collected partitioned images in multi-dimensional space. According to the information clustering results, the texture retrieval and extraction of partitioned images are realized by using the deep hash fusion algorithm, and the information detection and feature recognition capabilities of partitioned images in multi-dimensional space are improved. Simulation results show that this method has higher precision and better feature resolution in texture retrieval of partitioned images in multidimensional space, which improves the texture retrieval and recognition ability of partitioned images.

Dynamic slow feature analysis and random forest for subway indoor air quality modeling

The indoor air quality of a metro station is closely related to the health of passengers. In order to monitor the concentration of particulate matter in the air more stably and reduce the monitoring cost, a soft sensing model based on dynamic slow feature analysis (DSFA) and random forest (RF) is proposed in this paper. First, the augmented matrix technique is embedded into the slow feature analysis model to create a DSFA model to cope with the time lag inherent in the actual measurement process. Then, the constructed DSFA model is used to extract slowly varying latent variables to capture the dynamic trend within the input data. Finally, taking the latent variables as the input and aiming at the nonlinearity in the data, RF is used to predict the particle concentration. The results show that the prediction performance of the DSFA-RF model is better than some traditional soft sensor models, including partial least squares, support vector regression, single RF, and a hybrid model. The root mean square error value of DSFA-RF is improved by 21.93% compared with the hybrid model based on principal component analysis and support vector regression. In addition, DSFA also outperforms in health risk assessment.

Development and technical validation of an artificial intelligence model for quantitative analysis of histopathologic features of eosinophilic esophagitis

BackgroundIn an attempt to provide quantitative, reproducible, and standardized analyses in cases of eosinophilic esophagitis (EoE), we have developed an artificial intelligence (AI) digital pathology model for the evaluation of histologic features in the EoE/esophageal eosinophilia spectrum. Here, we describe the development and technical validation of this novel AI tool. MethodsA total of 10 726 objects and 56.2 mm2 of semantic segmentation areas were annotated on whole-slide images, utilizing a cloud-based, deep learning artificial intelligence platform (Aiforia Technologies, Helsinki, Finland). Our training set consisted of 40 carefully selected digitized esophageal biopsy slides which contained the full spectrum of changes typically seen in the setting of esophageal eosinophilia, ranging from normal mucosa to severe abnormalities with regard to each specific features included in our model. A subset of cases was reserved as independent “test sets” in order to assess the validity of the AI model outside the training set. Five specialized experienced gastrointestinal pathologists scored each feature blindly and independently of each other and of AI model results. ResultsThe performance of the AI model for all cell type features was similar/non-inferior to that of our group of GI pathologists (F1-scores: 94.5–94.8 for AI vs human and 92.6–96.0 for human vs human). Segmentation area features were rated for accuracy using the following scale: 1. “perfect or nearly perfect” (95%–100%, no significant errors), 2. “very good” (80%–95%, only minor errors), 3. “good” (70%–80%, significant errors but still captures the feature well), 4. “insufficient” (less than 70%, significant errors compromising feature recognition). Rating scores for tissue (1.01), spongiosis (1.15), basal layer (1.05), surface layer (1.04), lamina propria (1.15), and collagen (1.11) were in the “very good” to “perfect or nearly perfect” range, while degranulation (2.23) was rated between “good” and “very good”. ConclusionOur newly developed AI-based tool showed an excellent performance (non-inferior to a group of experienced GI pathologists) for the recognition of various histologic features in the EoE/esophageal mucosal eosinophilia spectrum. This tool represents an important step in creating an accurate and reproducible method for semi-automated quantitative analysis to be used in the evaluation of esophageal biopsies in this clinical context.

A novel kernel dynamic inner slow feature analysis method for dynamic nonlinear process concurrent monitoring of operating point deviations and process dynamics anomalies

A novel nonlinear dynamic inner slow feature analysis method is proposed for dynamic nonlinear process concurrent monitoring of operating point deviations and process dynamics anomalies. In this method, the nonlinear correlation and serial autocorrelation are considered meanwhile to extract the serial auto-correlated latent slow features with explicit dynamic representation. In order to approve slow features (SFs) with explicit dynamic representation from nonlinear dynamic process data, a new multi-goal optimization question is formulized with constraints of the extract latent variable catch some variation information and mutually orthogonal. After the nonlinear dynamic inner slow feature analysis model is trained from data, a corresponding detection strategy is also developed to perform process condition concurrent monitoring. Finally, the superiority and effectiveness of the proposed monitoring method are demonstrated by a numerical simulation case and an actual cold rolling mill case.

Early Differentiation of Irreversible Electroporation Ablation Regions With Radiomics Features of Conventional MRI

Irreversible electroporation (IRE) is a promising non-thermal ablation technique for the treatment of patients with hepatocellular carcinoma. Early differentiation of the IRE zone from surrounding reversibly electroporated (RE) penumbra is vital for the evaluation of treatment response. In this study, an advanced statistical learning framework was developed by evaluating standard MRI data to differentiate IRE ablation zones, and to correlate with histological tumor biomarkers. Fourteen rabbits with VX2 liver tumors were scanned following IRE ablation and forty-six features were extracted from T1w and T2w MRI. Following identification of key imaging variables through two-step feature analysis, multivariable classification and regression models were generated for differentiation of IRE ablation zones, and correlation with histological markers reflecting viable tumor cells, microvessel density, and apoptosis rate. The performance of the multivariable models was assessed by measuring accuracy, receiver operating characteristics curve analysis, and Spearman correlation coefficients. The classifiers integrating four radiomics features of T1w, T2w, and T1w+T2w MRI data distinguished IRE from RE zones with an accuracy of 97%, 80%, and 97%, respectively. Also, pixelwise classification models of T1w, T2w, and T1w+T2w MRI labeled each voxel with an accuracy of 82.8%, 66.5%, and 82.9%, respectively. Regression models obtained a strong correlation with behavior of viable tumor cells (0.62 ≤ r2 ≤ 0.85, p < 0.01), apoptosis (0.40 ≤ r2 ≤ 0.82, p < 0.01), and microvessel density (0.48 ≤ r2 ≤ 0.58, p < 0.01). MRI radiomics features provide descriptive power for early differentiation of IRE and RE zones while observing strong correlations among multivariable MRI regression models and histological tumor biomarkers.

Double-Layer Distributed Monitoring Based on Sequential Correlation Information for Large-Scale Industrial Processes in Dynamic and Static States

Due to the complex static, dynamic, and large-scale characteristics for modern industrial processes, in this article, we propose a double-layer distributed monitoring approach based on multiblock slow feature analysis and multiblock independent component analysis. To this end, the processed dataset is divided into the static and dynamic blocks on the basis of the sequential information of each variable in the first layer. Considering the correlations between the variables in the large-scale processes, the sequential correlation matrices in two blocks are calculated, which serves as the second-layer block division rule. Then, the static and dynamic blocks are further divided into several static and dynamic subblocks in which the variables in each subblock are strongly correlated and in the same state. The slow feature analysis and independent component analysis monitoring models are, respectively, generated for the dynamic and static subblocks. Finally, the monitoring results in each subblock are integrated by Bayesian inference to get the final statistics. The average fault detection rate of the proposed method for the Tennessee Eastman process is 0.842, while those of the other traditional methods are lower than 0.75, which shows the advantages of the proposed method.

Regional Differentiation Characteristics of the Development Ability of Coastal Villages and Towns under the Condition of Market Economy in the Reform and Transformation

Guan, Z. and Li, P., 2020. Regional differentiation characteristics of the development ability of coastal villages and towns under the condition of market economy in the reform and transformation. In: Yang, Y.; Mi, C.; Zhao, L., and Lam, S. (eds.), Global Topics and New Trends in Coastal Research: Port, Coastal and Ocean Engineering. Journal of Coastal Research, Special Issue No. 103, pp. 120–124. Coconut Creek (Florida), ISSN 0749-0208.Before the reform and transformation, the coupling rate of each factor of the regional differentiation characteristics of the development ability of coastal villages and towns is relatively low under the market economy condition. Therefore, the research on the regional differentiation characteristics of the development ability of coastal villages and towns under the market economy condition in the reform and transformation is proposed. Through the establishment of the coastal villages and towns development capacity regional analysis feature analysis model, extract the regional development features, and complete the regional analysis feature evaluation of development capacity. The simulation experiment is designed to compare the coupling rate of each factor of the regional differentiation characteristics of the development capacity of coastal villages and towns under the market economy condition before and after the reform, which proves the effectiveness of the research.

Projection pursuit random forest using discriminant feature analysis model for churners prediction in telecom industry

A major and demanding issue in the telecommunications industry is the prediction of churn customers. Churn describes the customer who is attrite from one Telecom service provider to competitors searching for better services offers. Companies from the Telco sector frequently have customer relationship management offices it is the main objective in how to win back defecting clients because preserve long-term customers can be much more beneficial to a company than gain newly recruited customers. Researchers and practitioners are paying great attention and investing more in developing a robust customer churn prediction model, especially in the telecommunication business by proposed numerous machine learning approaches. Many approaches of Classification are established, but the most effective in recent times is a tree-based method. The main contribution of this research is to predict churners/non-churners in the Telecom sector based on project pursuit Random Forest (PPForest) that uses discriminant feature analysis as a novelty extension of the conventional Random Forest approach for learning oblique Project Pursuit tree (PPtree). The proposed methodology leverages the advantage of two discriminant analysis methods to calculate the project index used in the construction of PPtree. The first method used Support Vector Machines (SVM) as a classifier in the construction of PPForest to differentiate between churners and non-churners customers. The second method is a Linear Discriminant Analysis (LDA) to achieve linear splitting of variables node during oblique PPtree construction to produce individual classifiers that are robust and more diverse than classical Random Forest. It found that the proposed methods enjoy the best performance measurements e.g. Accuracy, hit rate, ROC curve, Gini coefficient, Kolmogorov-Smirnov statistic and lift coefficient, H-measure, AUC. Moreover, PPForest based on direct applied of LDA on the raw data delivers an effective evaluator for the customer churn prediction model.

Developing global image feature analysis models to predict cancer risk and prognosis

In order to develop precision or personalized medicine, identifying new quantitative imaging markers and building machine learning models to predict cancer risk and prognosis has been attracting broad research interest recently. Most of these research approaches use the similar concepts of the conventional computer-aided detection schemes of medical images, which include steps in detecting and segmenting suspicious regions or tumors, followed by training machine learning models based on the fusion of multiple image features computed from the segmented regions or tumors. However, due to the heterogeneity and boundary fuzziness of the suspicious regions or tumors, segmenting subtle regions is often difficult and unreliable. Additionally, ignoring global and/or background parenchymal tissue characteristics may also be a limitation of the conventional approaches. In our recent studies, we investigated the feasibility of developing new computer-aided schemes implemented with the machine learning models that are trained by global image features to predict cancer risk and prognosis. We trained and tested several models using images obtained from full-field digital mammography, magnetic resonance imaging, and computed tomography of breast, lung, and ovarian cancers. Study results showed that many of these new models yielded higher performance than other approaches used in current clinical practice. Furthermore, the computed global image features also contain complementary information from the features computed from the segmented regions or tumors in predicting cancer prognosis. Therefore, the global image features can be used alone to develop new case-based prediction models or can be added to current tumor-based models to increase their discriminatory power.

Complex dynamic process monitoring method based on slow feature analysis model of multi-subspace partitioning

This study presents an ensemble monitoring strategy based on slow feature analysis (SFA) model of multi-subspace partitioning for dynamic large-scale process. SFA can effectively extract the various dynamics of process data, where the relationship between process data and slow features (SFs) can be revealed by transformation matrix. The similar projecting directions represent similar importance of variables, and corresponding latent variables (LVs) will show similar monitoring behavior. Several LV subspaces are obtained by dividing the transformation vectors with higher similarity into the same sub-block automatically based on the defined process variable related index and hierarchical clustering, which can avoid the problems of information loss and the selection of SFs. Then, the S2 statistics constructed in each subspaces are integrated by support vector data description to show an intuitive detection results. Experiments on Tennessee Eastman benchmark process and wastewater treatment process have validated the proposed strategy’s effectiveness and excellence.

Model of Controlled Drug Release from the Polymer Implant

A mathematical model for analysis of features of the drug release , previously introduced into a polymer implant, into a biological tissue is proposed. A carbon nanolayer obtained as a result of plasma-immersion ion implantation was created to improve biocompatibility with biological tissue on the surface of the implant. The medicine can go through micro-ruptures in this layer. Calculations show that the carbon layer allows a uniform release of the drug.

Big Data Aided Vehicular Network Feature Analysis and Mobility Models Design

Vehicular networks play a pivotal role in intelligent transportation system (ITS) and smart city (SC) construction, especially with the coming of 5G. Mobility models are crucial parts of vehicular network, especially for routing policy evaluation as well as traffic flow management. The big data aided vehicle mobility analysis and design attract researchers a lot with the acceleration of big data technology. Besides, complex network theory reveals the intrinsic temporal and spatial characteristics, considering the dynamic feature of vehicular network. In the following content, a big GPS dataset in Beijing, and its complex features verification are introduced. Some novel vehicle and location collaborative mobility schemes are proposed relying on the GPS dataset. We evaluate their performance in terms of complex features, such as duration distribution, interval time distribution and temporal and spatial characteristics. This paper elaborates upon mobility design and graph analysis of vehicular networks.

Prediction Based on the Acquisition of Sina Weibo Data and Users’ Forwarding Behavior

Microblogging, also known as Weibo, has currently become the prevalent social networking platform, and its function of disseminating information has been increasingly recognized and valued. One of the main approaches of information communication on Weibo is forwarding activities among users. Therefore, the key interests have focused on the microblogs that are easily recognized and forwarded. Through predicting the possibility of a micro-blog being forwarded among users, it is able to effectively improve the efficiency of information spreading on the Weibo platform. On the basis of Sina Weibo data, this paper intends to study 13 features of Sina Weibo including user and content, so as to conduct machine learning, to establish a feature analysis model, to identify key influence factors on Weibo forwarding, to study the combination of feature attributes for the first time, to explore the correlation among features, then to predict the probability of a micro-blog being forwarded by users. Meanwhile, concerning different features held by different data, different machine learning algorithms are applied in model training, in order to identify the algorithm with the highest prediction accuracy.

Abstract P4-02-06: Improving efficacy of applying breast MRI to detect mammography-occult breast cancer

Abstract Background: Mammography is the only clinically acceptable imaging modality in current population-based breast cancer screening, but it has a relatively low sensitivity in detecting early cancers. Although breast magnetic resonance imaging (MRI) is much more sensitive in detecting early or mammography-occult cancers, the low cancer detection yield in the general screening population (∼1%) prohibits breast MRI as a screening tool with wide applicability. In order to solve this clinical dilemma, in this preliminary study, we tested the potential of applying a new mammographic image feature analysis model that serves as a short-term breast cancer risk prediction model to identify women at high risk of harboring mammography-occult breast cancers, which can be detected by breast MRI. Methods: An image dataset involving 30 women who had both mammography and breast MRI screening examinations was retrospectively assembled. All mammograms were interpreted as negative by the radiologists during the original image interpretation. When applying breast MRI examinations to these women immediately following the negative mammography, 5 women were positive with cancer detected and 25 remained negative. We developed a computer-aided detection scheme to process the bilateral CC view mammograms of the left and right breasts. Ten (10) texture-based mammographic image features were computed and compared from the bilateral mammograms. These features were then fused to build a new artificial neural network (ANN) based risk model. Based on the threshold of 0.5 applying to the ANN-generated risk prediction scores (ranging from 0 to 1), we stratified these 30 women into two groups with high and low risk of harboring mammography-occult cancer. Results: Using the ANN-generated risk scores, 9 and 21 cases were assigned to high-risk and low-risk group, respectively. All 5 breast MRI-detected cancer cases were classified into the high-risk group (with 100% sensitivity), while 4 negative cases were also classified into the high-risk group resulting in a 16% false-positive rate (4/25). All 21 cases in the low-risk group were negative cases. Hence, our risk prediction model yielded an overall prediction accuracy of 86.7% in which 26 of 30 cases were correctly classified. The potential clinical impact is that based on the case stratification result, the maximum cancer detection yield of using breast MRI is 56% (5/9) in this dataset. Meanwhile, it can also eliminate 84% (21/25) unnecessary (negative) breast MRI screening examinations. Conclusions: This is the first study, which demonstrated that by computing and analyzing the variation of mammographic image features, we are able to develop a new quantitative image feature analysis model that can be applied to predict the short-term risk of a woman having a mammography-occult cancer that can be detected by breast MRI. This new strategy has the potential to significantly increase cancer detection yield of breast MRI and thus make breast MRI a more cost-effective imaging modality in breast cancer screening. Citation Format: Zheng B, Hollingsworth AB, Tan MY, Stough RG, Liu H. Improving efficacy of applying breast MRI to detect mammography-occult breast cancer. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P4-02-06.

Protein remote homology detection by combining Chou's distance-pair pseudo amino acid composition and principal component analysis.

Protein remote homology detection is one of the important tasks in computational proteomics, which is important for basic research and practical application. Currently, the SVM-based discriminative methods have shown superior performance. However, the existing feature vectors still cannot suitably represent the protein sequences, and often lack an interpretable model for analysis of characteristic features. Previous studies showed that sequence-order effects and physicochemical properties are important for representing protein sequences. However, how to use these kinds of information for constructing predictors is still a challenging problem. In this study, in order to incorporate the sequence-order information and physicochemical properties into the prediction, a method called disPseAAC is proposed, in which the feature vector is constructed by combining the occurrences of amino acid pairs within the Chou's pseudo amino acid composition (PseAAC) approach. The predictive performance and computational cost are further improved by employing the principal component analysis strategy. Various experiments are conducted on a benchmark dataset. Experimental results show that disPseAAC achieves an ROC score of 0.922, outperforming some existing state-of-the-art methods. Furthermore, the learnt model can easily be analyzed in terms of discriminative features, and the computational cost of the proposed method is much lower than that of other profile-based methods.

A Radar Target Recognition Method Based on Nonparametric Feature Analysis and Backward Cloud Model

When applying Parameter Discriminant Analysis (PDA) in extracting features of radar target High-Resolution Range Profile (HRRP), the construction of scatter matrices relies on the assumption that HRRPs in all classes satisfy the Gaussian distribution with the same covariance matrix. However, the distribution of HRRP is actually complex. In order to tackle this problem, a radar target recognition approach based on nonparametric feature analysis and back cloud model is proposed in this paper. Compared with PDA, nonparametric feature analysis (NFA) estimates the contribution of the K nearest neighbors (KNN) points to calculate the between-class scatter matrix. NFA makes use of class boundary information and relaxes the requirement of Gaussian distribution assumption in PDA. Moreover, back cloud model better describes the complex distribution of the HRRP NFA subspace due to the representation of signal’s randomness and fuzziness. Simulation results based on a HRRP dataset of five aircraft models demonstrate the effectiveness of the proposed approach.

The Readers Project: Procedural Agents and Literary Vectors

The Readers Project is an aesthetically oriented system of software entities designed to explore the culture of human reading. These entities, or “readers,” navigate texts according to specific reading strategies based upon linguistic feature analysis and real-time probability models harvested from search engines. As such, they function as autonomous text generators, writing machines that become visible within and beyond the typographic dimension of the texts on which they operate. Thus far the authors have deployed the system in a number of interactive art installations at which audience members can view the aggregate behavior of the readers on a large screen display and also subscribe, via mobile device, to individual reader outputs. As the structures on which these readers operate are culturally and aesthetically implicated, they shed critical light on a range of institutional practices - particularly those of reading and writing - and explore what it means to engage with the literary in digital media.

Typological feature analysis models linguistic geography

Dunn and colleagues (2008) describe and exemplify the use of sophisticated analyses of abstract structural features to reconstruct language histories. The techniques that they use do show some clustering in the groups of languages that they examine; Dunn et al. state that they 'tend to favor a phylogenetic origin for the signal of relatedness' (p. 748), and that the results of their test case 'show a close degree of correspondence to the existing linguistic classification based on sound-meaning correspondences' (p. 747). We argue that a more parsimonious explanation for the results obtained by Dunn et al.'s methodology is that it accurately maps linguistic geography, the network of contact and diffusion that postdates a proto-language, in most cases corresponding to geographic distance.

Word correlation matrices for protein sequence analysis and remote homology detection

BackgroundClassification of protein sequences is a central problem in computational biology. Currently, among computational methods discriminative kernel-based approaches provide the most accurate results. However, kernel-based methods often lack an interpretable model for analysis of discriminative sequence features, and predictions on new sequences usually are computationally expensive.ResultsIn this work we present a novel kernel for protein sequences based on average word similarity between two sequences. We show that this kernel gives rise to a feature space that allows analysis of discriminative features and fast classification of new sequences. We demonstrate the performance of our approach on a widely-used benchmark setup for protein remote homology detection.ConclusionOur word correlation approach provides highly competitive performance as compared with state-of-the-art methods for protein remote homology detection. The learned model is interpretable in terms of biologically meaningful features. In particular, analysis of discriminative words allows the identification of characteristic regions in biological sequences. Because of its high computational efficiency, our method can be applied to ranking of potential homologs in large databases.

Remote homology detection based on oligomer distances

Remote homology detection is among the most intensively researched problems in bioinformatics. Currently discriminative approaches, especially kernel-based methods, provide the most accurate results. However, kernel methods also show several drawbacks: in many cases prediction of new sequences is computationally expensive, often kernels lack an interpretable model for analysis of characteristic sequence features, and finally most approaches make use of so-called hyperparameters which complicate the application of methods across different datasets. We introduce a feature vector representation for protein sequences based on distances between short oligomers. The corresponding feature space arises from distance histograms for any possible pair of K-mers. Our distance-based approach shows important advantages in terms of computational speed while on common test data the prediction performance is highly competitive with state-of-the-art methods for protein remote homology detection. Furthermore the learnt model can easily be analyzed in terms of discriminative features and in contrast to other methods our representation does not require any tuning of kernel hyperparameters. Normalized kernel matrices for the experimental setup can be downloaded at www.gobics.de/thomas. Matlab code for computing the kernel matrices is available upon request. thomas@gobics.de, peter@gobics.de.