Dual-level Fuzzy Learning with Patch Guidance for Image Ordinal Regression

Many real-world datasets are labeled with natural orders, i.e., ordinal labels. Ordinal regression is a method to predict ordinal labels that finds a wide range of applications in data-rich domains, such as natural, health and social sciences. Most existing ordinal regression approaches work well for independent and identically distributed (IID) instances via formulating a single ordinal regression task. However, for heterogeneous non-IID instances with well-defined local geometric structures, e.g., subpopulation groups, multi-task learning (MTL) provides a promising framework to encode task (subgroup) relatedness, bridge data from all tasks, and simultaneously learn multiple related tasks in efforts to improve generalization performance. Even though MTL methods have been extensively studied, there is barely existing work investigating MTL for heterogeneous data with ordinal labels. We tackle this important problem via sparse and deep multi-task approaches. Specifically, we develop a regularized multi-task ordinal regression (MTOR) model for smaller datasets and a deep neural networks based MTOR model for large-scale datasets. We evaluate the performance using three real-world healthcare datasets with applications to multi-stage disease progression diagnosis. Our experiments indicate that the proposed MTOR models markedly improve the prediction performance comparing with single-task ordinal regression models.

Quadratic Weighted Kappa (QWK) is a statistic to measure the agreement between two annotators. QWK has been widely used as the evaluation measure for various medical imaging problems, where, the class labels have a natural ordering, e.g., no Diabetic Retinopathy (DR), mild DR, and severe DR. The easiest way to treat the classification problem with these ordinal labels is to consider the problem as a multiclass classification problem and apply the Cross Entropy (CE) loss. However, when applying CE loss the order of the classes becomes meaningless, i.e., the loss will be same if a healthy image is classified into mild DR or severe DR. At the same time, the QWK score will be severely affected if a healthy image is classified into severe DR than mild DR. The most appropriate way to get a better classification score is to optimize the evaluation measure itself. i.e., directly optimize the QWK statistics. However, this optimization may hinder the learning, and may lead to sub-optimal solutions, and therefore, may give lower performance than expected. On the other hand, Ordinal Regression (OR) based approaches also can be used for such problems. The main focus of this work is to investigate which loss function (CE loss, QWK loss or OR loss) is the most appropriate one to the Convolutional Neural Network-based ordinal classification problems, where, QWK is used as the evaluation measure. Experiments on two public datasets, Diabetic Retinopathy and Prostate Cancer, with two different network architectures suggest that directly optimizing QWK is the better choice when small networks are used. On the other hand, we found that for large networks OR based loss function gives better performance.

Behavioral Signal Processing aims at automating behavioral coding schemes such as those prevalent in psychology and mental health research. This paper describes a method to automatically quantify the approach-and-avoidance (AA) behavior, described by ordinal labels manually assigned by experts using either video-only or video-with-audio. We propose a novel ordinal regression (OR) algorithm and its hidden Markov model (HMM) extension for estimation of AA labels from visual motion capture based and acoustic features. The proposed algorithm transforms the OR to multiple binary classification problems, solves them by independent score-outputting classifiers and fits the cumulative logit logistic regression model with proportional odds (CLLRMP) to vectors of the classifier scores. The time series extension treats labels as states of the HMM with a likelihood function derived from the probabilistic CLLRMP output. We compare performances of the proposed algorithm applying the weighted binary SVMs in the second step (SVM-OLR), its time-series extension (HMM-SVM-OLR) and the baseline multi-class SVM. On the used dyadic interaction dataset the HMM-SVM-OLR achieves the highest estimation accuracies 71.6 % and 65.7 % for AA labels assigned respectively using video-only and video-with-audio.

A novel supervised approach to learning efficient kernel descriptors for high accuracy object recognition

In situations where the log transformation is used to make the assumptions of the ordinary linear regression model more reasonable, interval estimates for future observations are often desired in the scale of the original measurements. Common practice is to transfer the usual equal tail interval in the transformed scale back to the original scale. Conditions under which this procedure gives reasonably short intervals as compared with the shortest possible interval having the same content are derived, and a simple procedure that gives a reasonably short interval is suggested when it is not satisfactory. Primarily, a Bayesian point of view is taken.

Along with the rapid development of industries and the acceleration of urbanisation, the problem of air pollution is becoming more serious. Exploring the relevant factors affecting air quality and accurately predicting the air quality index are significant in improving the overall environmental quality and realising green economic development. Machine learning algorithms and statistical models have been widely used in air quality prediction and ranking assessment. In this paper, based on daily air quality data for the city of Xi’an, China, from 1 October 2022 to 30 September 2023, we construct support vector regression (SVR), gradient boosting decision tree (GBDT), extreme gradient boosting (XGBoost), random forests (RF), neural network (NN) and long short-term memory (LSTM) models to analyse the influence of the air quality index for Xi’an and to conduct comparative tests. The predicted values and 95% prediction intervals of the AQI for the next 15 days for Xi’an, China, are given based on the Bootstrap-XGBoost algorithm. Further, the ordinal logit regression and ordinal probit regression models are constructed to evaluate and accurately predict the AQI ranks of the data from 1 October 2023 to 15 October 2023 for Xi’an. Finally, this paper proposes some suggestions and policy measures based on the findings of this paper.

The video-EEG prolonged monitoring established the clinical, syndromic and topographic correct diagnosis in these patients; the fuzzy logic best defined a mathematical value of the correspondence between ambulatorial diagnosis of TLE and the video-EEG prolonged monitoring.

Time Series Ordinal Classification (TSOC) is yet an unexplored field of machine learning consisting in the classification of time series whose labels follow a natural order relationship between them. In this context, a well-known approach for time series nominal classification was previously used: the Shapelet Transform (ST). The exploitation of the ordinal information was included in two steps of the ST algorithm: 1) by using the Pearson’s determination coefficient (\(R^2\)) for computing the quality of the shapelets, which favours shapelets with better ordering, and 2) by applying an ordinal classifier instead of a nominal one to the transformed dataset. For this, the distance between labels was represented by the absolute value of the difference between the corresponding ranks, i.e. by the \(L_1\) norm. In this paper, we study the behaviour of different \(L_p\) norms for representing class distances in ordinal regression, evaluating 9 different \(L_p\) norms with 7 ordinal time series datasets from the UEA-UCR time series classification repository and 10 different ordinal classifiers. The results achieved demonstrate that the Pearson’s determination coefficient using the \(L_{1.9}\) norm in the computation of the difference between the shapelet and the time series labels achieves a significantly better performance when compared to the rest of the approaches, in terms of both Correct Classification Rate (CCR) and Average Mean Absolute Error (AMAE).

Recently, feature relation learning has attracted extensive attention in cross-spectral image patch matching. However, most feature relation learning methods can only extract shallow feature relations and are accompanied by the loss of useful discriminative features or the introduction of disturbing features. Although the latest multi-branch feature difference learning network can relatively sufficiently extract useful discriminative features, the multi-branch network structure it adopts has a large number of parameters. Therefore, we propose a novel two-branch feature interaction learning network (FIL-Net). Specifically, a novel feature interaction learning idea for cross-spectral image patch matching is proposed, and a new feature interaction learning module is constructed, which can effectively mine common and private features between cross-spectral image patches, and extract richer and deeper feature relations with invariance and discriminability. At the same time, we re-explore the feature extraction network for the cross-spectral image patch matching task, and a new two-branch residual feature extraction network with stronger feature extraction capabilities is constructed. In addition, we propose a new multi-loss strong-constrained optimization strategy, which can facilitate reasonable network optimization and efficient extraction of invariant and discriminative features. Furthermore, a public VIS-LWIR patch dataset and a public SEN1-2 patch dataset are constructed. At the same time, the corresponding experimental benchmarks are established, which are convenient for future research while solving few existing cross-spectral image patch matching datasets. Extensive experiments show that the proposed FIL-Net achieves state-of-the-art performance in three different cross-spectral image patch matching scenarios.

Estimation models in software engineering are used to predict some important attributes of future entities such as software development effort, software reliability and programmers productivity. Among these models, those estimating software effort have motivated considerable research in recent years. Estimation by analogy is one of the attractive techniques in software effort estimation field. However, the procedure used in estimation by analogy is not yet able to handle correctly categorical data. In this paper, we propose a new approach based on reasoning by analogy, fuzzy logic and linguistic quantifiers to estimate effort when the software project is described either by categorical or numerical data. Fuzzy logic-based cost estimation models are more appropriate when vague and imprecise information is to be accounted for. Fuzzy systems try to behave just like the processes of the brain with a rule base. In our proposed method the effort and the cost can be estimated using the Fuzzy logic based analogy method and the cost can be calculated. The performance of the proposed system is calculated using MARE (Mean Absolute Relative Error).

The analysis of microscopic images of fish gonad cells (oocytes) is a useful tool to estimate parameters of fish reproductive ecology and to analyze fish population dynamics. The study of oocyte dynamics is needed to understand ovary development and reproductive cycle of fish. Oocytes go through different developmental states in a continuum temporal sequence providing an interesting example of ordinal classification, which is not exploited by the current oocyte analysis software. This promising paradigm of machine learning known as ordinal classification or ordinal regression focus on classification problems where there exist a natural order between the classes, thus requiring specific methods and evaluation metrics. In this paper we compare 11 ordinal and 15 nominal state-of-the-art classifiers using oocytes of three fish species (Merluccius merluccius, Trisopterus luscus and Reinhardtius hippoglossoides). The best results are achieved by SVMOD, an ordinal decomposition method of the labelling space based on the Support Vector Machine, varying strongly with the number of states for each specie (about 95 and 80 % of accuracy with three and six states respectively). The classifiers designed specially for ordinal classification are able to capture the underlying nature of the state ordering much better than common nominal classifiers. This is demonstrated by several metrics specially designed to measure misclassification errors associated to states far in the ranking scale.

Developing a semi-supervised learning and ordinal classification framework for quality level prediction in manufacturing

This paper focuses on problems related to learning rules using numerical data for the Hierarchical Fuzzy Logic Systems (HFLS) described in [12]. Using hierarchical structure of Fuzzy Logic Systems (FLS) complex problems could be divided into subproblems with smaller dimensions. “Hierarchical” means that fuzzy sets produced as output of one of fuzzy logic systems are then processed as an input of another one as the sets of auxiliary variables. The main problem is to learn a rulebase with numerical data, which does not contain any data for those auxiliary variables. Learning rules for FLS in short could be accomplished by using many different approaches, building one, complex rulebase using all available input and output variables for complex problems. Our learning method based on the Wang & Mendel (W&M) method adopted for the HFLS with selective activation of unit FLS were introduced in [13]. The main scope of this paper is to extend our method applying quality measures of IF-THEN rules in the sense of Wu & Mendel (Wu&M) to remove conflicting rules. The proposal presented in this paper operates on a type-1 HFLS, built with the fuzzy logic systems (in the sense of Mamdani). An example of single-player games, i.e. where the “enemy” is controlled by agents is used. Two new problems are briefly introduced.

To improve the cognition and understanding capabilities of artificial intelligence (AI) technology, it is a tendency to explore the human brain learning processing and integrate brain mechanisms or knowledge into neural networks for inspiration and assistance. This paper concentrates on the application of AI technology in advanced driving assistance system. In this field, millimeter-wave radar is essential for elaborate environment perception due to its robustness to adverse conditions. However, it is still challenging for radar object classification in the complex traffic environment. In this paper, a knowledge-assisted neural network (KANN) is proposed for radar object classification. Inspired by the human brain cognition mechanism and algorithms based on human expertise, two kinds of prior knowledge are injected into the neural network to guide its training and improve its classification accuracy. Specifically, image knowledge provides spatial information about samples. It is integrated into an attention mechanism in the early stage of the network to help reassign attention precisely. In the late stage, object knowledge is combined with the deep features extracted from the network. It contains discriminant semantic information about samples. An attention-based injection method is proposed to adaptively allocate weights to the knowledge and deep features, generating more comprehensive and discriminative features. Experimental results on measured data demonstrate that KANN is superior to current methods and the performance is improved with knowledge assistance.

Background and objective: Arrhythmia is a widely seen cardiologic ailment worldwide, and is diagnosed using electrocardiogram (ECG) signals. The ECG signals can be translated manually by human experts, but can also be scheduled to be carried out automatically by some agents. To easily diagnose arrhythmia, an intelligent assistant can be used. Machine learning-based automatic arrhythmia detection models have been proposed to create an intelligent assistant. Materials and Methods: In this work, we have used an ECG dataset. This dataset contains 1000 ECG signals with 17 categories. A new hand-modeled learning network is developed on this dataset, and this model uses a 3D shape (prismatoid) to create textural features. Moreover, a tunable Q wavelet transform with low oscillatory parameters and a statistical feature extractor has been applied to extract features at both low and high levels. The suggested prismatoid pattern and statistical feature extractor create features from 53 sub-bands. A neighborhood component analysis has been used to choose the most discriminative features. Two classifiers, k nearest neighbor (kNN) and support vector machine (SVM), were used to classify the selected top features with 10-fold cross-validation. Results: The calculated best accuracy rate of the proposed model is equal to 97.30% using the SVM classifier. Conclusion: The computed results clearly indicate the success of the proposed prismatoid pattern-based model.