Soft Classification Research Articles

A 2-Tier Bengali Dataset for Evaluation of Hard and Soft Classification Approaches

Document classification is an open problem in library, information, and computer sciences towards assigning documents to one or more classes. The interest of linguistic researchers in this domain has increased day by day due to interesting applications like language identification, readability assessment, sentiment analysis, spam filtering, etc. However, researchers focussing on natural language processing of resource-scaring languages have faced many hurdles due to the absence of benchmark datasets. Bengali is among the most-spoken resource-scaring or low-resource language. Although Bengali NLP researchers have endeavoured towards creating their own datasets, they are only useful for performance evaluation of their proposed document classification techniques only. Therefore, there is a gap in the literature on the availability of benchmark datasets. To overcome this barrier, this paper presents a benchmark dataset for Bengali document classification, which is publicly accessible and freely available. This dataset consists of a two-tier architecture, the first-tier for hard classification and the second-tier for soft classification techniques. Hard classification techniques follow supervised learning based models for the classification of documents, while on the other hand, soft classification techniques follow unsupervised learning based models for the clustering of documents. The proposed dataset consists of thirteen unique characteristics. This paper also introduces four new feature sets to evaluate the performance of the proposed dataset, namely: location revealing factor, part of speech tagging factor, relative frequency, and prominence factor.

Data-Driven Fault Early Warning Model of Automobile Engines Based on Soft Classification

Since automobile engine fault is the main factor leading to a vehicle breaking down, engine fault diagnosis has captured a lot of attention. Fault diagnosis identifies fault types to facilitate maintenance. However, the method of the warning before the fault occurs is more attractive to users and is more challenging. Therefore, this study would like to explore the feasibility of implementing automobile engine fault early warning based on the fault diagnosis model. First, the theoretical method of a fault domain is established, and the state of the engine is regarded as a point in n-dimensional space. The normal or fault of the engine will correspond to different state domains in this space. Second, to diagnose multiple fault types at the same time, an ensemble model based on multiple machine learning methods is established. The probability outputs by the ensemble model measure the distance between the point and each fault domain in the space. Finally, considering the temporal factor, an early warning threshold is established based on the probability, and a fault warning model is established by using the dual probability structure. Comparative experiments show that the proposed method can greatly reduce the calculation time based on ensuring the accuracy of early warning and is suitable for real-time early warning of multiple faults.

MOLECULE-BASED SPARSE REPRESENTATION STRATEGY AS A REPLACEMENT SOLUTION OF ORTHOGONAL MATCHING PURSUIT TECHNIQUE FOR DECOMPOSITION OF TIME SERIES OF VEGETATION INDICES

Abstract. This paper proposes a new strategy based on introducing the molecules as a replacement alternative to the well-known OMP method for a sparse representation of the time series of vegetation indices. This method, by preventing the occurrence of leakage problems, has led to improving the accuracy of the soft classification of agricultural products. To do so, a library of molecules (combination of atoms) is firstly generated by the representative ground truth data as the probable cultivation patterns, and then each time series of vegetation index is decomposed based on molecules. The efficiency of each molecule is measured through three different criteria 1- Maximizing the accuracy of reconstructing the input signal, 2- Minimizing the number of contributed atoms in the molecule, and 3- Minimizing the estimated negative abundances. Then, considering the uncertainties of representative atoms, an iterative imposition of constraints has been used to balance the estimated abundances. The proposed method has improved by 25.21% compared to the common OMP method in the soft classification of temporal signals of vegetation indices.

An evidential combination method with multi-color spaces for remote sensing image scene classification

Remote sensing image scene classification aims to commit the semantic labels according to the content of images. Convolutional Neural Network (CNN) is often used here to extract deep discriminative feature of remote sensing images for classification. In practice, CNN is usually trained by images in the Red Green Blue (RGB) color space. Whereas, CNN also can be trained by images in some other color spaces, e.g., Hue Saturation Value. The CNN models trained by images in diverse color spaces will perform differently because different color spaces often emphasize diverse color information. Thus, we present an Evidential Combination method with Multi-color Spaces (ECMS) to integrate the complementary information of different color spaces for classification performance improvement. In ECMS, labeled remote sensing images in the RGB color space are first converted into other color spaces, and then they are used to train CNN models, respectively. The soft classification results (of query images) yielded by these CNN models are combined by evidence theory. During fusion, the reliabilities/weights of these outputs of different CNN models are usually different, so they should not be equally treated for combination. In our approach, the weights are learnt by minimizing the mean squared error between the combination results and ground truth on labeled images. By doing this, weighted evidence combination of soft classification results is employed to make scene class decision. We conducted experiments on several datasets to verify the effectiveness of ECMS, and the results show ECMS can significantly improve classification accuracy compared with many existing methods.

Estimation of water quality parameters of GF-1 WFV in turbid water based on soft classification

Estimation of water quality parameters of GF-1 WFV in turbid water based on soft classification

Early detection of Parkinson's disease by using neuroimaging and biomarkers through hard and soft classifiers

Early detection of Parkinson's disease by using neuroimaging and biomarkers through hard and soft classifiers

Early detection of Parkinson's disease by using neuroimaging and biomarkers through hard and soft classifiers

Early detection of Parkinson's disease by using neuroimaging and biomarkers through hard and soft classifiers

Facial Age Estimation Models for Embedded Systems: A Comparative Study

Automated age estimation from face images is the process of assigning either an exact age or a specific age range to a facial image. In this paper a comparative study of the current techniques suitable for this task is performed, with an emphasis on lightweight models suitable for embedded implementation. We investigate both the suitable modern deep learning architectures for feature extraction and the variants of framing the problem itself as either classification, regression or soft label classification. The models are evaluated on Audience dataset for age group classification and FG-NET dataset for exact age estimation. To gather in-depth insights into automated age estimation and in contrast to existing studies, we additionally compare the performance of both classification and regression on the same dataset. We propose a novel loss function that combines regression and classification approaches and show that it outperforms other considered approaches. At the same time, with a lightweight backbone, such an architecture is suitable for implementation on embedded devices.

Fuzzy-Inspired Nanobiosensing for Multi-Feature Soft Classification of Tumor

This paper proposes a novel fuzzy-inspired nanobiosensing framework for soft classification of tumor by using multiple features of the tumor, where interrelationships between different features are accounted for. For illustration purpose, two features of breast cancer, namely, tumor tissue abnormality and tumor foci separation, are utilized to determine the disease state. The sensing performance is then evaluated through the fuzzy relation transfer analysis. This analysis enables an intuitive yet systematic way to characterize the variation of classification fuzziness occurred during the sensing process facilitated by nanoscale contrast agents. Subsequently, the mean absolute error of the pre- and post-contrast membership functions is employed to evaluate the sensing integrity. Finally, numerical results are presented to demonstrate the principles of the proposed framework.

Deep Learning Approach for Emotion Recognition Analysis in Text Streams

Social media sites employ various approaches to track feelings, including diagnosing neurological problems, including fear, in people or assessing a population public sentiment. One essential obstacle for automatic emotion recognition principles is variable with fluctuating limitations, language, and interpretation shifts. Therefore, in this paper, a deep learning-based emotion recognition (DL-EM) system has been proposed to describe the various relational effects in emotional groups. A soft classification method is suggested to quantify the tendency and allocate a message to each emotional class. A supervised framework for emotions in text streaming messages is developed and tested. Two of the major activities are offline teaching assignments and interactive emotion classification techniques. The first challenge offers templates in text responses to describe sentiment. The second activity includes implementing a two-stage framework to identify live broadcasts of text messages for dedicated emotion monitoring.

An automated system for the assessment and grading of adolescent delinquency using a machine learning-based soft voting framework

Adolescent (or juvenile) delinquency is defined as the habitual engagement in unlawful behavior of a minor under the age of majority. According to studies, the likelihood of acquiring a deviant personality increases significantly during adolescence. As a result, identifying deviant youth early and providing proper medical counseling makes perfect sense. Due to the scarcity of qualified clinicians, human appraisal of individual adolescent behavior is subjective and time-consuming. As a result, a machine learning-based intelligent automated system for assessing and grading delinquency levels in teenagers at an early stage must be devised. To solve this problem, a soft voting-based ensemble classification model has been developed that includes a Decision Tree, Multi-layer Perceptron, and Support Vector Machine as base classifiers to accurately classify teenagers into three groups based on severity levels, viz., low, medium, and high. Over the normalized structured behavioral data, the proposed soft voting-based model outperforms all other individual classifiers with 87.50% accuracy, an AUC of 0.94, 0.81 Kappa value, and an F-score of 0.88.

A Comparative Analysis of Methods of Endmember Selection for Use in Subpixel Classification: A Convex Hull Approach.

Mixed pixels in aerial and satellite images are common, especially near the boundaries of two or more discrete classes; that is, they tend to occur at the transitional region between two classes. Ideally, to decipher the mixed pixel, a soft classification is performed compared to a hard- or a per-pixel classification. Soft or subpixel classification is carried out where the fractional cover of the LULC contained within a pixel is derived. Endmembers are extracted for three VNIR bands of ASTER data for two image datasets using three approaches, namely, principal component analysis (PCA), pixel purity index (PPI), and convex hull-Graham scan (CHGS). On comparing the DN values of the identified endmembers, it is observed that the CHGS method provides the most appropriate end members than the PCA-derived and PPI-derived end members. This is based on deriving the endmembers from two different image conditions. Convex hull implemented using the Graham scan algorithm delineates the pure pixel and pinpoints the exact number of endmembers. These accurate end members would result in accurate proportions of the land cover for better modeling of the terrain.

An Integrated Framework for Multi-State Driver Monitoring Using Heterogeneous Loss and Attention-Based Feature Decoupling.

Multi-state driver monitoring is a key technique in building human-centric intelligent driving systems. This paper presents an integrated visual-based multi-state driver monitoring framework that incorporates head rotation, gaze, blinking, and yawning. To solve the challenge of head pose and gaze estimation, this paper proposes a unified network architecture that tackles these estimations as soft classification tasks. A feature decoupling module was developed to decouple the extracted features from different axis domains. Furthermore, a cascade cross-entropy was designed to restrict large deviations during the training phase, which was combined with the other features to form a heterogeneous loss function. In addition, gaze consistency was used to optimize its estimation, which also informed the model architecture design of the gaze estimation task. Finally, the proposed method was verified on several widely used benchmark datasets. Comprehensive experiments were conducted to evaluate the proposed method and the experimental results showed that the proposed method could achieve a state-of-the-art performance compared to other methods.

Dimensionality Reduction and Classification of Hyperspectral Remote Sensing Image Feature Extraction

Terrain classification is an important research direction in the field of remote sensing. Hyperspectral remote sensing image data contain a large amount of rich ground object information. However, such data have the characteristics of high spatial dimensions of features, strong data correlation, high data redundancy, and long operation time, which lead to difficulty in image data classification. A data dimensionality reduction algorithm can transform the data into low-dimensional data with strong features and then classify the dimensionally reduced data. However, most classification methods cannot effectively extract dimensionality-reduced data features. In this paper, different dimensionality reduction and machine learning supervised classification algorithms are explored to determine a suitable combination method of dimensionality reduction and classification for hyperspectral images. Soft and hard classification methods are adopted to achieve the classification of pixels according to diversity. The results show that the data after dimensionality reduction retain the data features with high overall feature correlation, and the data dimension is drastically reduced. The dimensionality reduction method of unified manifold approximation and projection and the classification method of support vector machine achieve the best terrain classification with 99.57% classification accuracy. High-precision fitting of neural networks for soft classification of hyperspectral images with a model fitting correlation coefficient (R2) of up to 0.979 solves the problem of mixed pixel decomposition.

Soft-margin classification of object manifolds.

A neural population responding to multiple appearances of a single object defines a manifold in the neural response space. The ability to classify such manifolds is of interest, as object recognition and other computational tasks require a response that is insensitive to variability within a manifold. Linear classification of object manifolds was previously studied for max-margin classifiers. Soft-margin classifiers are a larger class of algorithms and provide an additional regularization parameter used in applications to optimize performance outside the training set by balancing between making fewer training errors and learning more robust classifiers. Here we develop a mean-field theory describing the behavior of soft-margin classifiers applied to object manifolds. Analyzing manifolds with increasing complexity, from points through spheres to general manifolds, a mean-field theory describes the expected value of the linear classifier's norm, as well as the distribution of fields and slack variables. By analyzing the robustness of the learned classification to noise, we can predict the probability of classification errors and their dependence on regularization, demonstrating a finite optimal choice. The theory describes a previously unknown phase transition, corresponding to the disappearance of a nontrivial solution, thus providing a soft version of the well-known classification capacity of max-margin classifiers. Furthermore, for high-dimensional manifolds of any shape, the theory prescribes how to define manifold radius and dimension, two measurable geometric quantities that capture the aspects of manifold shape relevant to soft classification.

Airborne imaging spectrometer dataset for spectral characterization and predictive mineral mapping using sub-pixel based classifier in parts of Udaipur, Rajasthan, India

Imaging spectroscopy is used for the mapping of different minerals using their characteristics absorption depths. The present study covers the parts of the Udaipur region for mineral mapping with the help of a Mixture Tuned Matched Filtering (MTMF) soft classifier using the Airborne Visible Infrared Imaging Spectrometer – Next Generation dataset. The preprocessed and classified dataset by MTMF soft classifier identified carbonates, serpentine minerals, and alteration products of ultramafics, clay and mica minerals, and intimate mixture of kaolinite and muscovite in the region. These minerals show the characteristic absorption features at different wavelength positions of the SWIR region. The comparison of the image spectrum was carried out with the reference (mineral library) and laboratory collected reflectance spectral curves. The derived mineral map accurately mapped the minerals and discriminated the minerals belonging to a similar mineral class. Validation was done using field evidences, photomicrographs, XRD, laboratory spectral curves, and already published maps. The accuracy for the generated mineral map is 83.93%.

Aeroelastic Design and Comprehensive Analysis of Composite Rotor Blades Through Cluster-Based Kriging

Excessive vibration during high-speed flights is a critical issue in modern rotorcraft. The aeroelastic design of the composite rotor blades can relieve the hub vibratory loads of the rotor that are a dominant source of these vibrations. However, it has many technical problems, such as a high-dimensional design space, narrow and discontinuous feasible regions, and highly nonlinear design indices. Although the surrogate-based design is known to improve the design results, the conventional surrogate models have an insufficient capacity to represent the nonlinearity of aeroelastic responses. Because the inaccuracy of the surrogate model is combined with the discontinuity of feasible regions during the optimization process, it is likely to converge to local optima. Cluster kriging could accurately represent the complicated physical relations using machine learning techniques, such as soft clustering and classification algorithms. The accurate surrogate model results in a global optimization with 56% vibration reduction, whereas the global approximation using a single kriging yields local optima with 39% vibration reduction. Furthermore, a comprehensive design analysis, conducted to investigate the underlying physics of the vibration reduction, confirms the global optimization. The proposed design method can fully use the multidisciplinary nature of aeroelastic responses, which is not considered in conventional design methods.

Diagnosing COVID-19, Prioritizing Treatment, and Planning Vaccination Priority via Fuzzy Parameterized Fuzzy Soft Matrices

In the fight against the COVID-19 pandemic, it is vital to rapidly diagnose possible contagions, treat patients, plan follow-up procedures with correct and effective use of resources and ensure the formation of herd immunity. The use of machine learning and statistical methods provides great convenience in dealing with too many data produced during research. Since access to the PCR test used for the diagnosis of COVID-19 may be limited, the test is relatively too slow to yield results, the cost is high, and its reliability is controversial; thus, making a symptomatic classification before the PCR is timesaving and far less costly. In this study, by modifying a state-of-the-art classification method, namely Comparison Matrix-Based Fuzzy Parameterized Fuzzy Soft Classifier (FPFS-CMC), an effective method is developed for a rapid diagnosis of COVID-19. The paper then presents the accuracy, sensitivity, specificity, and F1-score values that represent the diagnostic performances of the modified method. The results show that the modified method can be adopted as a competent and accurate diagnosis procedure. Afterwards, a tirage study is performed by calculating the patients’ risk scores to manage inpatient overcrowding in healthcare institutions. In the subsequent section, a vaccine priority algorithm is proposed to be used in the case of a possible crisis until the supply shortage of a newly developed vaccine is over if a possible variant of COVID-19 that is highly contagious is insensitive to the vaccine. The accuracy of the algorithm is tested with real-life data. Finally, the need for further research is discussed.

Incorporating Measurement Error in Astronomical Object Classification

Most general-purpose classification methods, such as support-vector machine (SVM) and random forest (RF), fail to account for an unusual characteristic of astronomical data: known measurement error uncertainties. In astronomical data, this information is often given in the data but discarded because popular machine learning classifiers cannot incorporate it. We propose a simulation-based approach that incorporates heteroscedastic measurement error into an existing classification method to better quantify uncertainty in classification. The proposed method first simulates perturbed realizations of the data from a Bayesian posterior predictive distribution of a Gaussian measurement error model. Then, a chosen classifier is fit to each simulation. The variation across the simulations naturally reflects the uncertainty propagated from the measurement errors in both labeled and unlabeled data sets. We demonstrate the use of this approach via two numerical studies. The first is a thorough simulation study applying the proposed procedure to SVM and RF, which are well-known hard and soft classifiers, respectively. The second study is a realistic classification problem of identifying high-z (2.9 ≤ z ≤ 5.1) quasar candidates from photometric data. The data are from merged catalogs of the Sloan Digital Sky Survey, the Spitzer IRAC Equatorial Survey, and the Spitzer-HETDEX Exploratory Large-Area Survey. The proposed approach reveals that out of 11,847 high-z quasar candidates identified by a random forest without incorporating measurement error, 3146 are potential misclassifications with measurement error. Additionally, out of 1.85 million objects not identified as high-z quasars without measurement error, 936 can be considered new candidates with measurement error.

COLAM: Co-Learning of Deep Neural Networks and Soft Labels via Alternating Minimization

Softening labels of training datasets with respect to data representations has been frequently used to improve the training of deep neural networks. While such a practice has been studied as a way to leverage “privileged information” about the distribution of the data, a well-trained learner with soft classification outputs should be first obtained as a prior to generate such privileged information. To solve such a “chicken-and-egg” problem, we propose COLAM framework that Co-Learns DNNs and soft labels through Alternating Minimization of two objectives—(a) the training loss subject to soft labels and (b) the objective to learn improved soft labels—in one end-to-end training procedure. We performed extensive experiments to compare our proposed method with a series of baselines. The experiment results show that COLAM achieves improved performance on many tasks with better testing classification accuracy. We also provide both qualitative and quantitative analyses that explain why COLAM works well.