An Outlier Detection Using Autoencoder for Ocean Observation Data

Outlier detection research in ocean data has traditionally been performed using statistical and distance-based machine learning algorithms. Recently, AI-based methods have received a lot of attention and so-called supervised learning methods that require classification information for data are mainly used. This supervised learning method requires a lot of time and costs because classification information (label) must be manually designated for all data required for learning. In this study, an autoencoder based on unsupervised learning was applied as an outlier detection to overcome this problem. For the experiment, two experiments were designed: one is univariate learning, in which only SST data was used among the observation data of Deokjeok Island and the other is multivariate learning, in which SST, air temperature, wind direction, wind speed, air pressure, and humidity were used. Period of data is 25 years from 1996 to 2020, and a pre-processing considering the characteristics of ocean data was applied to the data. An outlier detection of actual SST data was tried with a learned univariate and multivariate autoencoder. We tried to detect outliers in real SST data using trained univariate and multivariate autoencoders. To compare model performance, various outlier detection methods were applied to synthetic data with artificially inserted errors. As a result of quantitatively evaluating the performance of these methods, the multivariate/univariate accuracy was about 96%/91%, respectively, indicating that the multivariate autoencoder had better outlier detection performance. Outlier detection using an unsupervised learning-based autoencoder is expected to be used in various ways in that it can reduce subjective classification errors and cost and time required for data labeling.

A Feature Selection Algorithm Integrating Maximum Classification Information and Minimum Interaction Feature Dependency Information

Feature selection is the key step in the analysis of high-dimensional small sample data. The core of feature selection is to analyse and quantify the correlation between features and class labels and the redundancy between features. However, most of the existing feature selection algorithms only consider the classification contribution of individual features and ignore the influence of interfeature redundancy and correlation. Therefore, this paper proposes a feature selection algorithm for nonlinear dynamic conditional relevance (NDCRFS) through the study and analysis of the existing feature selection algorithm ideas and method. Firstly, redundancy and relevance between features and between features and class labels are discriminated by mutual information, conditional mutual information, and interactive mutual information. Secondly, the selected features and candidate features are dynamically weighted utilizing information gain factors. Finally, to evaluate the performance of this feature selection algorithm, NDCRFS was validated against 6 other feature selection algorithms on three classifiers, using 12 different data sets, for variability and classification metrics between the different algorithms. The experimental results show that the NDCRFS method can improve the quality of the feature subsets and obtain better classification results.

A Deformable Convolutional Neural Network with Spatial-Channel Attention for Remote Sensing Scene Classification

Remote sensing scene classification converts remote sensing images into classification information to support high-level applications, so it is a fundamental problem in the field of remote sensing. In recent years, many convolutional neural network (CNN)-based methods have achieved impressive results in remote sensing scene classification, but they have two problems in extracting remote sensing scene features: (1) fixed-shape convolutional kernels cannot effectively extract features from remote sensing scenes with complex shapes and diverse distributions; (2) the features extracted by CNN contain a large number of redundant and invalid information. To solve these problems, this paper constructs a deformable convolutional neural network to adapt the convolutional sampling positions to the shape of objects in the remote sensing scene. Meanwhile, the spatial and channel attention mechanisms are used to focus on the effective features while suppressing the invalid ones. The experimental results indicate that the proposed method is competitive to the state-of-the-art methods on three remote sensing scene classification datasets (UCM, NWPU, and AID).

LRSCnet: Local Reference Semantic Code learning for breast tumor classification in ultrasound images

Purpose This study proposed a novel Local Reference Semantic Code (LRSC) network for automatic breast ultrasound image classification with few labeled data. Methods In the proposed network, the local structure extractor is firstly developed to learn the local reference which describes common local characteristics of tumors. After that, a two-stage hierarchical encoder is developed to encode the local structures of lesion into the high-level semantic code. Based on the learned semantic code, the self-matching layer is proposed for the final classification. Results In the experiment, the proposed method outperformed traditional classification methods and AUC (Area Under Curve), ACC (Accuracy), Sen (Sensitivity), Spec (Specificity), PPV (Positive Predictive Values), and NPV(Negative Predictive Values) are 0.9540, 0.9776, 0.9629, 0.93, 0.9774 and 0.9090, respectively. In addition, the proposed method also improved matching speed. Conclusions LRSC-network is proposed for breast ultrasound images classification with few labeled data. In the proposed network, a two-stage hierarchical encoder is introduced to learn high-level semantic code. The learned code contains more effective high-level classification information and is simpler, leading to better generalization ability.

KinasePhos 3.0: Redesign and Expansion of the Prediction on Kinase-specific Phosphorylation Sites

The purpose of this work is to enhance KinasePhos, a machine-learning-based kinase-specific phosphorylation site prediction tool. Experimentally verified kinase-specific phosphorylation data were collected from PhosphoSitePlus, UniProt, GPS 5.0, and Phospho.ELM. In total, 41,421 experimentally verified kinase-specific phosphorylation sites were identified. A total of 1380 unique kinases were identified, including 753 with existing classification information from KinBase and the remaining 627 annotated by building a phylogenetic tree. Based on this kinase classification, a total of 771 predictive models were built at the individual, family, and group levels, using at least 15 experimentally verified substrate sites in positive training datasets. The improved models were observed to be more effective than other prediction tools. For example, the prediction of sites phosphorylated by the Akt, CKT, and PKA families had accuracies of 94.5%, 92.5%, and 90.0%, respectively. The average prediction accuracy for all 771 models was 87.2%. For enhancing interpretability, the Shapley additive explanations (SHAP) method was employed to assess feature importance. The web interface of KinasePhos 3.0 has been redesigned with the goal of providing comprehensive annotations of kinase-specific phosphorylation sites on multiple proteins. Additionally, considering the large scale of phosphoproteomic data, a downloadable prediction tool is available at https://awi.cuhk.edu.cn/KinasePhos/index.html or https://github.com/tom-209/KinasePhos-3.0-executable-file.

Using Ontologies in Failure Analysis

Fault analysis is a complex task that requires electrical engineers to perform various analyses to detect and localize a physical defect. The analysis process is very knowledge-intensive and must be precisely documented to report the issue to customers as well as to ensure the best possible reuse of the acquired experience in similar future analyses. However, writing unambiguous documentation can be complicated for many reasons, such as selecting details and results to be presented in a report, or the naming of terms and their definition. To avoid some of these issues, FA engineers must agree on a clearly defined terminology specifying methods, physical faults and their electrical signatures, tools, and relations between them. Moreover, to allow FA software systems to use this terminology, it must be stored in a format that can be interpreted similarly by both engineers and software. This paper presents an approach that solves these challenges by using an ontology describing FA-relevant terminology using a logic-based representation. The latter guarantees the same interpretation of the defined terms by engineers and software systems, which can use it to perform various tasks like text classification, information retrieval, or workflow verification.

FEF-Net: A Deep Learning Approach to Multiview SAR Image Target Recognition

Synthetic aperture radar (SAR) is an advanced microwave imaging system of great importance. The recognition of real-world targets from SAR images, i.e., automatic target recognition (ATR), is an attractive but challenging issue. The majority of existing SAR ATR methods are designed for single-view SAR images. However, multiview SAR images contain more abundant classification information than single-view SAR images, which benefits automatic target classification and recognition. This paper proposes an end-to-end deep feature extraction and fusion network (FEF-Net) that can effectively exploit recognition information from multiview SAR images and can boost the target recognition performance. The proposed FEF-Net is based on a multiple-input network structure with some distinct and useful learning modules, such as deformable convolution and squeeze-and-excitation (SE). Multiview recognition information can be effectively extracted and fused with these modules. Therefore, excellent multiview SAR target recognition performance can be achieved by the proposed FEF-Net. The superiority of the proposed FEF-Net was validated based on experiments with the moving and stationary target acquisition and recognition (MSTAR) dataset.

J-Net: Asymmetric Encoder-Decoder for Medical Semantic Segmentation

With the development of deep learning, breakthroughs have been made in the field of semantic segmentation. However, it is difficult to generate a fine mask on the same medical images because medical images have low contrast, high resolution, and insufficient semantic information. In most scenarios, existing approaches mostly use a pooling layer to reduce the resolution of feature maps. Therefore, it is difficult for them to consider the whole image features, resulting in information loss and performance degradation. In this paper, a multiscale asymmetric encoder-decoder semantic segmentation network is proposed. The network consists of two parts, which perform feature extraction and image restoration on the input, respectively. The encoder network obtains multiscale feature information by connecting multiple ASPP modules to form a feature pyramid. Meanwhile, the upsampling layer of each decoder can be connected to the feature map generated by the corresponding ASPP module. Finally, the classification information of each pixel is obtained through the sigmoid function. The performance of the proposed method can be verified on publicly available datasets. The experimental evidence shows that the proposed method can take full advantage of multiscale feature information and achieve superior performance with less inference computational cost.

CORPORATE BANKRUPTCY AND INSOLVENCY PREDICTION MODEL

In any competitive economy, the risk of bankruptcy is pervasive. The research aims to contribute in improving the predictive power of bankruptcy and insolvency risk among companies by introducing new methods of processing and validation. This paper investigates the extensive application of the Z score model for predicting the economic-financial stability of Romanian companies in the manufacturing and extractive industries. A list of 37 financial indicators determined on the basis of the balance sheet data of 80 companies for the period 2015–2018 was used. Stepwise Least Squares Estimation through the Forward method allowed the identification of the most relevant ones. Canonical discriminant analysis and sensitivity analyzes were introduced to test the predictive power of the model. The new model identified allows both the prediction of bankruptcy and insolvency risk. This study contributes to the literature by testing variables in relation to financial difficulties and by including other classification information. The robustness of the determined canonical discriminant function was verified by testing the model on two other samples.