One-versus-one Research Articles

Feature optimized hybrid model for prediction of myocardial infarction

Background Cardiovascular disease is rampant worldwide and has become the leading factor in increasing the global mortality rates. According to the World Heart Federation, death toll due to CVD has increased from 12.1 million in 1990 to around 19 million in 2019. Myocardial Infarction (MI) is a condition where the heart muscle dies due to reduced or inhibited flow of oxygenated blood. It has affected approximately 3 million people worldwide, with more than 1 million deaths in the United States annually. Such unusual proliferation in global death toll due to CVD can be reduced to a great extent by predicting the risk of CVD at an early stage. Method In this paper, several feature selection techniques including Variance-based, Mutual Information (MI), Maximum Relevance Minimum Redundancy (MRMR), Boruta, and Recursive Feature Elimination (RFE) algorithms are used feature optimization. For class prediction, the Logistic Regression (LR), Support Vector Machine (SVM), Decision Tree (DT), and Adaboost algorithms were implemented in their ordinary, One-vs-Rest (OVR) and One-vs-One (OVO) methods. Result The performance of Adaboost model has significantly improved by using feature selection techniques, that is, the accuracy of 74% (without any feature selection taking 5.3 seconds) is increased to 85% (with Boruta feature selection taking only 2.17 seconds training time) and 88% (with MRMR feature selection taking 1.6 seconds training time). Similarly, the DT-OVO model’s performance has improved from 84% (without any feature selection taking 1.48 seconds training time) to 86% (with Boruta feature selection taking 0.58 training time). For other models, the performance is maintained with reduced model training times. Conclusion This research paper prioritizes on feature selection in developing machine learning models for CVD prediction. This conclusion is justified by demonstrating the significant reduction in model training times for the 72 models generated while maintaining or even improving the model’s predictive performance.

Utilizing Multi-Class Classification Methods for Automated Sleep Disorder Prediction

Even from infancy, a human’s day-life alternates from a period of wakefulness to a period of sleep at night, during the 24-hour cycle. Sleep is a normal process necessary for human physical and mental health. A lack of sleep makes it difficult to control emotions and behaviour, reduces productivity at work, and can even increase stress or depression. In addition, poor sleep affects health; when sleep is insufficient, the chances of developing serious diseases greatly increase. Researchers in sleep medicine have identified an extensive list of sleep disorders, and thus leveraged Artificial Intelligence (AI) to automate their analysis and gain a deeper understanding of sleep patterns and related disorders. In this research, we seek a Machine Learning (ML) solution that will allow for efficient classification of unlabeled instances as being Sleep Apnea, Insomnia or Normal (subjects without a specific sleep disorder) by assessing the performance of two well-established strategies for multi-class classification tasks: the One-Vs-All (OVA) and One-Vs-One (OVO). In the context of the specific strategies, two well-known binary classification models were assumed, Logistic Regression (LR) and Support Vector Machines (SVMs). Both strategies’ validity was verified upon a dataset of diverse information related to the profiles (anthropometric data, sleep metrics, lifestyle and cardiovascular health factors) of potential patients or individuals not exhibiting any specific sleep disorder. Performance evaluation was carried out by comparing the weighted average results in all involved classes that represent these two specific sleep disorders and no-disorder occurrence; accuracy, kappa score, precision, recall, f-measure, and Area Under the ROC curve (AUC) were recorded and compared to identify an effective and robust model and strategy, both class-wise and on average. The experimental evaluation unveiled that after feature selection, 2-degree polynomial SVM under both strategies was the least complex and most efficient, recording an accuracy of 91.44%, a kappa score of 84.97%, precision, recall and f-measure equal to 0.914, and an AUC of 0.927.

A novel cost-sensitive quality determination framework in hot rolling steel industry

In the hot rolling industry, a high-precision quality determination framework is the guarantee for intelligent decision-making regarding products. Conventional methods ignore the significance of faulty products, which are assigned the same attention as the qualified products. To overcome these limitations, this paper proposes a novel compensated cost-sensitive multi-layer perceptron with minority attention (CCS-MLP-MIAT) to construct a high-precision quality determination framework. We first transform an imbalanced multi-class ordinal industrial problem into multiple imbalanced binary ordinal classifications using one-versus-one (OVO) decomposition. Then, we implement a minority attention mechanism and a cost-sensitive loss for MLP. Subsequently, a novel cost compensation coefficient is introduced and determined by Artificial Hummingbird Algorithm (AHA) to further improve the identification rate of faulty products. The advantage lies in the model to allocate more attention to the faulty products, thus promoting intelligent decision-making of hot rolling production. The proposed method is validated on eight UCI benchmark datasets and two real-cases of hot-rolled strip crown by cross validation. Experimental results demonstrate that the excellent generalization and robustness of CCS-MLP-MIAT in addressing the quality determination of hot-rolled steel.

Multiple Classifier System for Handling Imbalanced and Overlapping Datasets on Multiclass Classification

The performance of classification models suffer when the dataset contains imbalanced and overlapping data. These two conditions are already challenging separately and even more complex if they occur together. In the research, an ensemble method called a Multiple Classifier System was proposed to address these issues by combining K-Nearest Neighbour and Logistic Regression. The Synthetic Minority Oversampling Technique (SMOTE) method was also applied to balance the dataset. The One Versus One (OVO) decomposition technique helped the multiclass classification process. A simulation with 18 scenarios proves that the MCS-SMOTE model can handle these problems by providing good performance. The model’s performance is also tested using empirical data on Poverty in West Java in 2021. Empirical data also show that the proposed method performs well, with an accuracy rate of 80.09%, an F1 score of 0.782, and a G-Mean of 0.242. The areas with the highest poverty rates are Bogor, Bekasi City, Bandung City, Bekasi Regency, and Depok City, located near DKI Jakarta, the capital city. Based on existing predictor variables, poor households in West Java are more likely to occur when they do not have access to credit, the number of household members is more than three, multiple families live in one building, and the head of the household has not graduated from elementary school.

Abstract PO2-07-05: Deep learning model for automated quantification of HER2 expression in invasive breast cancers from immunohistochemical whole slide images

Abstract Introduction Human epidermal growth factor receptor 2 (HER2) protein overexpression and/or HER2 gene amplification is found in about 20% of invasive breast cancers. Considering the results of the DESTINY-Breast trials confirming the remarkable efficacy of anti-HER2 antibody drug-conjugate (T-Dxd) in both HER2-overexpressed and HER2-low tumors, it is necessary to identify not only HER2 (immunohistochemistry (IHC) score 3+) overexpressing tumors, but also HER2-low tumors. The latter category, defined as IHC1+ or IHC2+ but non-amplified, has proven to be challenging even for experienced pathologists, with high inter-observer variability. Here, we validate the performance of a deep learning (DL) model at: 1) predicting the IHC score from IHC histological features and 2) identifying HER2-low tumors with high sensitivity. Methods 675 HER2 stained IHC slides from primary breast cancer patients were selected based on pathology reports across three different cohorts (KCL_GSTT_1 n=369; Cypath_Breast n=214; KCL_GSTT_2 n=92). The slides were digitized and reviewed by expert breast pathologists. Specifically, Cypath_Breast and KCL_GSTT_2 were each annotated by one expert pathologist, while KCL_GSTT_1 was annotated by 5 expert pathologists with the ground truth defined as a majority vote between the 5 annotators. Cypath_Breast was assigned as the “discovery” set; while KCL_GSTT_1 and KCL_GSTT_2 were used as two independent validation cohorts. The model was specifically trained to extract features from IHC tiles to ensure they would capture both the staining intensity and the staining location on the cells (membrane, cytoplasmic or nuclei). The one-vs-one (OVO) and one-vs-rest (OVR) AUC as well as sensitivity and specificity to HER2-low and positive tumors were used as metrics to assess the performance of our model. An expert pathologist reviewed the most predictive regions of HER2 expression. Results After review by expert pathologists, 59 slides from KCL_GSTT_1 were removed either because of folded tissue, blurriness or because there was too little tissue. The IHC scores (0/1+/2+/3+) for the 3 cohorts were distributed as follows; Cypath_Breast: 42/120/36/16; KCL_GSTT_1: 120/90/52/48; KCL_GSTT_2: 54/22/15/1. For KCL_GSTT_1, the average agreement between the 5 expert pathologists amounted to a Cohen’s Kappa score of 0.63. Following training on Cypath_Breast, the performance of the model are presented in An ablation study proved that a feature extractor trained on IHC tiles outperformed an in-house extractor solely trained on H&E tiles (OVO AUC 0.95 vs 0.93 on KCL_GSTT_1; DeLong p< 0.001). Specifically, the model performed best at distinguishing HER2-negative (HER2 IHC 0) from HER2-low and HER2-positive tumors with a sensitivity (0 vs 1+/2+/3+) of 0.95 [0.93 - 0.98] and a specificity (0 vs 1+/2+/3+) of 0.78 [0.72 - 0.84] on KCL_GSTT_1. We observed the same tendency on KCL_GSTT_2 with a sensitivity (0 vs 1+/2+/3+) of 0.97 [0.91 - 1.00] and a specificity of 0.91 [0.83 - 0.98]. More importantly, the DL model reached a Cohen’s Kappa score of 0.72 [0.67 - 0.77] with the ground truth on KCL_GSTT_1, surpassing the average agreement Kappa score between expert pathologists (0.63). The most predictive regions showed cytoplasmic staining in tumor regions. Conclusion Our model provides a path towards a fully automated workflow for identifying up to 95% of breast cancer patients who could potentially benefit from anti-HER2 targeted therapies. Additionally, we show the utility of AI-based tools to minimize discrepancies among pathologists and assist in the diagnostic patient pathway. Table 1: Results of external validation on KCL_GSTT_1 and KCL_GSTT_2 <grpahic> Cohorts OVO AUC OVR AUC IHC 0 OVR AUC IHC 1+ OVR AUC IHC 2+ OVR AUC IHC 3+ KCL_GSTT_1 0.95 0.95 0.80 0.91 0.99 [0.93 - 0.96] [0.94 - 0.95] [0.79 - 0.82] [0.90 - 0.92] [0.99 - 1.00] KCL_GSTT_2 0.92 0.97 0.78 0.84 N/A [0.90 - 0.93] [0.96 - 0.98] [0.76 - 0.81] [0.80 - 0.88] Citation Format: Pierre-Antoine Bannier, Loïc Herpin, Rémy Dubois, Lydwine Van Praet, Charles Maussion, Ellen Amonoo, Anca Mera, Jasmine Timbres, Cheryl Gillett, Elinor Sawyer, Patrycja Gazinska, Piotr Ziolkowski, Roberto Salgado, Sheeba Irshad. Deep learning model for automated quantification of HER2 expression in invasive breast cancers from immunohistochemical whole slide images [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-07-05.

Asymptotic behavior of some multicategory classification methods for high-dimensional data

We consider multicategory extensions of binary discrimination methods via one-versus-one (OVO) or one-versus-rest (OVR) methodologies, focusing on extensions of the binary classification by linear mean difference (MD), support vector machine (SVM), maximal data piling (MDP), and distance weighted discrimination (DWD) via OVO, and the multicategory extension of MD via OVR, in the context of high-dimensional and low sample size (HDLSS) data. The asymptotic behavior of OVO-MD, OVO-SVM, OVO-MDP and OVO-DWD is described when the dimension of the data increases and the sample size is fixed, in terms of the probabilities of correct classification of a new data point, finding sufficient conditions for the correct classification probabilities to converge to one as the dimension approaches infinity. As in the binary case, OVO-MD, OVO-SVM and OVO-MDP have the same asymptotic behavior while OVO-DWD could behave differently. We also consider the asymptotic behavior of the OVR-MD methodology providing necessary and sufficient conditions for a new data point of a given class to be correctly classified with probability tending to one. A simulation experiment is conducted to further compare the methodologies, and consider the four binary methods in the OVR case. We evaluate the performance of the considered methods using a microarray data set.

One-vs-One, One-vs-Rest, and a novel Outcome-Driven One-vs-One binary classifiers enabled by optoelectronic memristors towards overcoming hardware limitations in multiclass classification

Deep neural networks have achieved considerable success over the past ten years in a variety of fields. However, current state-of-the-art artificial intelligence (AI) systems require large computing hardware infrastructure and high power consumption. To overcome these hurdles, it is required to adopt new strategies such as designing novel computation architectures and developing building blocks that can mimic the low energy consumption of biological systems. On the architecture level, implementing classification tasks by splitting the problem into simpler subtasks is a way to relax hardware constraints despite the less accuracy of the approach. On the computation unit level, memristive devices are a promising technology for low power neuromorphic computation. Hereby, we combine both these two approaches and present a novel algorithmic approach for multiclass classification tasks through splitting the problem into binary subtasks while using optoelectronics memristors as synapses. Our approach leverages the core fundamentals from the One-vs-One (OvO) and the One-vs-Rest (OvR) classification strategies towards a novel Outcome-Driven One-vs-One (ODOvO) approach. The light modulation of synaptic weights, fed in our algorithm from experimental data, is a key enabling parameter that permits classification without modifying further applied electrical biases. Our approach requires at least a 10X less synapses (only 196 synapses are required) while reduces the classification time by up to N2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\frac{N}{2}$$\\end{document} compared to conventional memristors. We show that the novel ODOvO algorithm has similar accuracies to OvO (reaching over 60% on the MNIST dataset) while requiring even fewer iterations compared to the OvR. Consequently, our approach constitutes a feasible solution for neural networks where key priorities are the minimum energy consumption i.e., small iterations number, fast execution, and the low hardware requirements allowing experimental verification.

Fuzzy regular least squares twin support vector machine and its application in fault diagnosis

Objective function ignores the generalization error in LSTSVM, and training overfitting results in poor generalization performance. All sample labels are considered deterministic, however, some samples contain outliers affected by noises, which leads to low reliability. A recognition method based on fuzzy regular LSTSVM (FRLSTSVM) is proposed. Firstly, L2 norm regular term is introduced into objective function to improve the generalization performance. Secondly, outliers of samples are detected through support vector domain description (SVDD), which improves outlier detection accuracy. Then, a membership degree S3 is constructed to give the outliers a suitable membership degree, reducing the impact of outliers on results. Finally, FRLSTSVM is extended to a multi-classification model by one versus one (OVO) and binary tree (BT) strategies, and it is combined with an improved multiscale fluctuating Rényi dispersion entropy (IMFRDE) for fault diagnosis. The results show that the method has stronger generalization, lower sensitivity to parameters and higher reliability.

Abstract 997: Cell-free DNA fragmentomics distinguish between benign, pre-malignant and malignant peripheral nerve sheath tumors in neurofibromatosis type 1

Abstract Background: Malignant peripheral nerve sheath tumors (MPNST) are aggressive soft tissue sarcomas that, in the setting of neurofibromatosis type 1 (NF1), arise within pre-malignant atypical neurofibroma (AN) and benign plexiform neurofibroma (PN). Early surgical resection improves prognosis, however, early detection by imaging and tissue biopsies is challenging due to tissue heterogeneity. In this multi-institutional study we analyze fragmentomic profiles of plasma cell free DNA (cfDNA) to non-invasively distinguish between NF1 associated PN, AN and MPNST. Accurate classification would inform clinical care: standards of care for PN is observation, AN is narrow-margin resection, and MPNST is wide-margin resection. Methods: We performed whole genome sequencing of plasma cfDNA samples from healthy controls (n = 21), patients with PN (n = 113), AN (n = 39) and MPNST (n = 71). cfDNA fragment profiles were analyzed using two complementary approaches. First, we used unsupervised non-negative matrix factorization (NMF) to obtain global fragment length signatures to infer tumor fragment length distributions. The optimal cutpoint was determined after receiver operating characteristic analysis by Youden’s index in one-versus-one (OVO) disease state comparisons. Additionally, we implemented a bin-wise fragmentomic analysis based on DELFI, training a classifier on the ratios of short (100-150bp) and long (151-220bp) fragments in 5 megabase regions across the genome and arm-level features. Results: NMF accurately distinguished disease states on OVO comparisons: MPNST v AN (acc 0.71), MPNST v PN (acc 0.75), MPNST v healthy (acc 0.84), AN v PN (acc 0.70), AN v healthy (acc 0.80) and PN v healthy (acc 0.87). Accuracies were moderately improved in nearly all conditions with bin-wise fragmentomics: MPNST v AN (acc 0.62), MPNST v PN (acc 0.83), MPNST vs healthy (acc 0.86), AN v PN (acc 0.87), AN vs healthy (acc 0.72) and PN vs healthy (acc 0.88). Strikingly, the two AN with the DELFI scores most closely resembling MPNST were separately identified by independent clinical care teams to have very high-risk features and recent history warranting short-interval follow up. Conclusions: This study demonstrates that the spectrum of benign, pre-malignant and malignant peripheral nerve sheath tumors have distinct, disease state specific fragmentomic signatures. Fragmentomics alone outperformed our previously published copy number based cfDNA classifier in all conditions, most notably in low mutational burden healthy, PN and AN states. Finally, preliminary clinical vignettes suggest that this approach may be able to inform surveillance intervals by identifying higher risk premalignant lesions. Together, this work has the potential to enable earlier detection of clinically actionable AN and early-stage MPNST, thereby improving survival outcomes. Citation Format: Alex C. Pan, Russell Taylor Sundby, Jeffrey J. Szymanski, Paul A. Jones, Peter K. Harris, Aadel A. Chaudhuri, Angela C. Hirbe, Jack F. Shern. Cell-free DNA fragmentomics distinguish between benign, pre-malignant and malignant peripheral nerve sheath tumors in neurofibromatosis type 1 [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 997.

Mining online reviews for ranking products: A novel method based on multiple classifiers and interval-valued intuitionistic fuzzy TOPSIS

Online reviews can assist customers in making purchasing decisions, however, when multiple alternative products need to be compared based on online reviews, a large number of reviews information cannot be processed manually. Therefore, this study develops a novel method for product ranking based on mining online reviews and an interval-valued intuitionistic fuzzy Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS). Concretely, it contains three major steps: (1) preprocessing, (2) calculating sentiment scores, and (3) ranking alternative products. First, the results of preprocessing are vectorization representation of online reviews, which are obtained with the Bidirectional Encoder Representations from Transformers (BERT) model. Then, sentiment orientations of online reviews of alternative products regarding different attributes are determined by a multiple classifiers system, which consists of multi-class base classifiers constructed using Support Vector Machine (SVM) with One-Vs-One (OVO) strategy. Lastly, based on the sentiment scores, the ranking result is obtained by the interval-valued intuitionistic fuzzy TOPSIS method. A case study on real-world datasets is provided to illustrate the application of our proposed method, which indicates the validity of our proposal. Thus, the method proposed in this paper can effectively assist consumers in selecting products based on online reviews that meet their preferences.

A mechanical part fault diagnosis method based on improved multiscale weighted permutation entropy and multiclass LSTSVM

Multi-scale permutation entropy (MPE) has poor stability, accuracy and noise immunity, recognition accuracy of twin SVM is low, so a diagnosis method based on improved multi-scale weighted permutation entropy (IMWPE) and least squares TSVM (LSTSVM) is proposed. The innovations include that the improved variational mode decomposition (IVMD) based on energy and mutual information is used to suppress noise, and novel composite coarse-grained and permutation pattern weighting methods are introduced to construct IMWPE, after which the noise immunity, stability and accuracy of features have been improved. LSTSVM is extended to a multi-class LSTSVM through one vs one (OVO) and binary tree (BT) strategies, the performance of the proposed methodology is significantly improved. The results show that signal-to-noise ratio of signal is enhanced, the stability of IMWPE is improved, the diagnosis accuracies for bearing and check valve are increased by 17.084 % and 12.498 %, which provides a reference for fault diagnosis under noise.

Multi-Dimensional Classification via Decomposed Label Encoding

In multi-dimensional classification (MDC), a number of class variables are assumed in the output space with each of them specifying the class membership w.r.t. one heterogeneous class space. One major challenge in learning from MDC examples lies in the heterogeneity of class spaces, where the modeling outputs from different class spaces are not directly comparable. To tackle this problem, we propose a new strategy named <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">decomposed label encoding,</i> which enables modeling alignment for MDC in an encoded label space derived from one-versus-one (OvO) decomposition. Specifically, the original MDC output space is transformed into a ternary encoded label space by conducting OvO decomposition w.r.t. each class space. Then, the manifold structure in the feature space is exploited to enrich the labeling information in the encoded label space. Finally, the predictive model is induced by fitting the metric-aligned modeling outputs with enriched labeling information. Extensive experiments over twenty benchmark data sets clearly show the superiority of the proposed MDC strategy against state-of-the-art approaches.

Instance selection using one‐versus‐all and one‐versus‐one decomposition approaches in multiclass classification datasets

AbstractInstance is important in data analysis and mining; it filters out unrepresentative, redundant, or noisy data from a given training set to obtain effective model learning. Various instance selection algorithms are proposed in the literature, and their potential and applicability in data cleaning and preprocessing steps are demonstrated. For multiclass classification datasets, the existing instance selection algorithms must deal with all the instances across the different classes simultaneously to produce a reduced training set. Generally, every multiclass classification dataset can be regarded as a complex domain problem, which can be effectively solved using the divide‐and‐conquer principle. In this study, the one‐versus‐all (OVA) and one‐versus‐one (OVO) decomposition approaches were used to decompose a multiclass dataset into multiple binary class datasets. These approaches have been widely employed when constructing the classifier but have never been considered in instance selection. The results of instance selection performance obtained with the OVA, OVO, and baseline approaches were assessed and compared for 20 different domain multiclass datasets as the first study and five medical domain datasets as the validation study. Furthermore, three instance selection algorithms were compared, including IB3, DROP3, and GA. The results demonstrate that using the OVO approach to perform instance selection can make the support vector machine (SVM) and k‐nearest neighbour (k‐NN) classifiers perform significantly better than the OVA and baseline approaches in terms of the area under the ROC curve (AUC) rate, regardless of the instance selection algorithm used. Moreover, the OVO approach can provide reasonably good data reduction rates and processing times, which are all better than those of the OVA approach.

Multiclass financial distress prediction based on one‐versus‐one decomposition integrated with improved decision‐directed acyclic graph

AbstractAlthough binary financial distress prediction (FDP) is widely explored, multiclass FDP (MFDP) lacks enough investigation. This paper designs a new fusion method of improved decision‐directed acyclic graph (IDDAG) based on the principle of maximizing generalization ability and integrates it with the one‐versus‐one (OVO) decomposition method (OVO‐IDDAG) to build MFDP models. Based on data of Chinese listed companies, we categorize corporate financial status into four states (financial soundness, financial pseudosoundness, moderate financial distress, and serious financial distress) and compare the MFDP performance of OVO‐IDDAG with two benchmark methods, that is, the OVO decomposition method integrated with the max‐win fusion method (OVO‐ max‐win) and the OVO decomposition method integrated with the original DDAG fusion method (OVO‐ODDAG). The experimental results show that the OVO‐IDDAG method overall outperformances the benchmark methods. Among the six base classifiers of multivariate discriminant analysis, Logit, naive bayes, support vector machine, decision tree, and random forest and the three imbalance processing mechanisms of random oversampling, random undersampling, and synthetic minority oversampling technique, the combination of the random forest classifier with the imbalance processing mechanism of random undersampling or the support vector machine classifier with the imbalance processing mechanism of random oversampling or synthetic minority oversampling technique is more suitable for MFDP based on OVO‐IDDAG. This study not only improves artificial intelligence methodology for multiclass classification but also enriches applications of intelligent systems to MFDP.

Multi-class SVM Classification Comparison for Health Service Satisfaction Survey Data in Bahasa

This study aimed to compare the Multi-class Support Vector Machine (MSVM) classification with the One-versus-One (OvO) and One-versus-Rest (OvR) approaches using unigram and bigram features. The study used the service satisfaction survey report of Denpasar public health centers by the Center for Public Health Innovation (CPHI), Medical School, Udayana University. As Bali is known as the world's main tourism destination, it is important to know about its supporting public health service through its representative capital city, Denpasar. Moreover, this study laid the foundation for the classification process using the available methods to fit in Indonesian health service satisfaction survey data, which assists in making decisions to improve health services. Since Bali is one of the provinces in Indonesia and all of those provinces refer to the same national regulation, health service satisfaction survey data that is in the Indonesian language (Bahasa) should have the same aspects, like category, priority, word-related matters (including abbreviations, acronyms, terminology), etc. that overall make it unique and need specific processing. That work was considered a contribution since there is no such study to the best of the author's knowledge and the foundation would be useful as a part of the future vision for the integrated system of Indonesian health big data. Since in reality, satisfaction survey data tends to be unbalanced, this study also compares the developed models using unigram and bigram features without and with feature selection (FS). Those features were then processed using the OvO MSVM and OvR MSVM models. k-fold cross-validation was used to divide training data and testing data and, at the same time, validate the models. Through experiments without and with FS, the OvO MSVM and OvR MSVM models with unigram features had better performance in general than the same models with bigram features. Without FS and with unigram features, comparable differences were found where the OvO MSVM model was slightly better on accuracy and precision, while the OvR MSVM model was slightly better on recall and the F1score. Without FS and with bigram features, comparable differences were also found, where the OvR MSVM model had slightly better performance than the OvO MSVM model. With FS and with unigram and bigram features, the OvR MSVM model had better performance in general than the OvO MSVM model. Doi: 10.28991/HIJ-2022-03-04-05 Full Text: PDF

Multi-class imbalanced enterprise credit evaluation based on asymmetric bagging combined with light gradient boosting machine

Most existing research on multi-class imbalanced enterprise credit evaluation modeling has been built on data-level imbalance processing methods and single classifier approaches. Using the one-versus-one (OVO) decomposition and fusion method to dispose multi-class classification, this paper proposes two new credit evaluation ensemble models by combining the asymmetric bagging (AB) and the light gradient boosting machine (LightGBM). Based on a multi-class imbalanced dataset of Chinese enterprises that issue corporate bonds from 2014 through 2020, this study conducts a series of empirical experiments for multi-class imbalanced enterprise credit evaluation. The experimental results demonstrate that our proposed models can significantly outperform the benchmark models, which integrate the OVO and one-versus-all decomposition and fusion method respectively with the random under sampling LightGBM, random over sampling LightGBM, synthetic minority over sampling technique LightGBM and the AB decision tree. In addition, the proposed models can carry out analysis on feature importance, which provides decision-making basis for enterprise stakeholders.

Chemistry Combining Elemental Profile, Stable Isotopic Ratios, and Chemometrics for Fine Classification of a Chinese Herb Licorice (Glycyrrhiza uralensis Fisch.) from 37 Producing Area.

A method based on elemental fingerprint, stable isotopic analysis and combined with chemometrics was proposed to trace the geographical origins of Licorice (Glycyrrhiza uralensis Fisch) from 37 producing areas. For elemental fingerprint, the levels of 15 elements, including Ca, Cu, Mg, Pb, Zn, Sr, Mn, Se, Cd, Fe, Na, Al, Cr, Co, and K, were analyzed by inductively coupled plasma atomic emission spectrometry (ICP-AES). Three stable isotopes, including δ13C, δ15N, and δ18O, were measured using an isotope-ratio mass spectrometer (IRMS). For fine classification, three multiclass strategies, including the traditional one-versus-rest (OVR) and one-versus-one (OVO) strategies and a new ensemble strategy (ES), were combined with two binary classifiers, partial least squares discriminant analysis (PLSDA) and least squares support vector machines (LS-SVM). As a result, ES-PLSDA and ES-LS-SVM achieved 0.929 and 0.921 classification accuracy of GUF samples from the 37 origins. The results show that element fingerprint and stable isotope combined with chemometrics is an effective method for GUF traceability and provides a new idea for the geographical traceability of Chinese herbal medicine.

Training Up to 50 Class ML Models on 3 $ IoT Hardware via Optimizing One-vs-One Algorithm (Student Abstract)

Multi-class classifier training using traditional meta-algorithms such as the popular One-vs-One (OvO) method may not always work well under cost-sensitive setups. Also, during inference, OvO becomes computationally challenging for higher class counts K as O(K^2) is its time complexity. In this paper, we present Opt-OvO, an optimized (resource-friendly) version of the One-vs-One algorithm to enable high-performance multi-class ML classifier training and inference directly on microcontroller units (MCUs). Opt-OvO enables billions of tiny IoT devices to self learn/train (offline) after their deployment, using live data from a wide range of IoT use-cases. We demonstrate Opt-OvO by performing live ML model training on 4 popular MCU boards using datasets of varying class counts, sizes, and feature dimensions. The most exciting finding was, on the 3 $ ESP32 chip, Opt-OvO trained a multi-class ML classifier using a dataset of class count 50 and performed unit inference in super real-time of 6.2 ms.

OVASO: Integrated binary CNN models to classify COVID‐19, pneumonia and healthy lung in X‐ray images

AbstractSeveral radiologists have paid attention to computer‐aided detection (CAD) systems which assist in classifying diseases on chest x‐ray (CXR). Recently, with the outbreak of COVID‐19, CAD based on deep learning has an important role in screening COVID‐19 on CXR. However, imbalanced training datasets such as COVID‐19 datasets, COVID‐19 (473), pneumonia (5458), and normal (7966) cause difficulty in classification. In this paper, we suggest a new evaluation approach, OVASO, that selectively combines one‐versus‐all (OVA) classifier and one‐versus‐one (OVO) to overcome class imbalance caused by the lower number of COVID‐19 training datasets. In addition, as part of efforts to properly apply transfer learning, we initialized batch normalization (BN) values including γ and β from the viewpoint of transfer learning and found that appropriate initialization at all binary models, OVASO's components, usually increased the binary models' performance. As a result, the proposed OVASO model achieved improved accuracy and F1‐score of 95.33% and 95.88%, respectively. Furthermore, the suggested OVASO performed similarly to COVID‐Net, which is the current state‐of‐the‐art model for classifying COVID‐19 on CXR.

Multiscale time-frequency method for multiclass Motor Imagery Brain Computer Interface

Motor Imagery Brain Computer Interface (MI-BCI) has become a promising technology in the field of neurorehabilitation. However, the performance and computational complexity of the current multiclass MI-BCI have not been fully optimized, and the intuitive interpretation of individual differences on motor imagery tasks is seldom investigated. In this paper, a well-designed multiscale time-frequency segmentation scheme is first applied to multichannel EEG recordings to obtain Time-Frequency Segments (TFSs). Then, the TFS selection based on a specific wrapper feature selection rule is utilized to determine optimum TFSs. Next, One-Versus-One (OvO)-divCSP implemented in divergence framework is used to extract discriminative features. Finally, One-Versus-Rest (OvR)-SVM is utilized to predict the class label based on selected multiclass MI features. Experimental results indicate our method yields a superior performance on two publicly available multiclass MI datasets with a mean accuracy of 80.00% and a mean kappa of 0.73. Meanwhile, the proposed TFS selection method can significantly alleviate the computational burden with little accuracy reduction, demonstrating the feasibility of real-time multiclass MI-BCI. Furthermore, the Motor Imagery Time-Frequency Reaction Map (MI-TFRM) is visualized, contributing to analyzing and interpreting the performance differences between different subjects.