Classifier Design Research Articles

Modified Tensor Distance-Based Multiview Spectral Embedding for PolSAR Land Cover Classification

This letter proposes a novel method for combining multiview features in polarimetric synthetic aperture radar (PolSAR) for land cover classification. It is well-known that feature extraction and classifier design are two significant steps in machine learning methods for PolSAR data interpretation. Each PolSAR pixel can be represented in different feature spaces, such as polarimetric data scattering, or the polarimetric target decomposition spaces. In this letter, a tensor-based multiview embedding algorithm is proposed to fuse those features from different spaces in order to obtain a distinctive set of features for the subsequent classification. Based on the pixel-based classification tasks, a modified tensor distance (MTD) is designed to accurately calculate the distance between tensors. It emphasizes the importance of the central pixel, and decreases the influence of the neighbors in the feature patch when calculating tensor distance. Furthermore, the complementary properties of different views are exploited by an MTD measured tensor multiview spectral embedding method, so as to obtain relevant low-dimensional features. Compared with state-of-the-art methods, the validation and effectiveness of the proposed method is demonstrated on two real PolSAR data sets.

Combination of Pattern Classifiers Based on Naive Bayes and Fuzzy Integral Method for Biological Signal Applications

Background: Achieving the best possible classification accuracy is the main purpose of each pattern recognition scheme. An interesting area of classifier design is to design for biomedical signal and image processing. Materials and Methods: In the current work, in order to increase recognition accuracy, a theoretical frame for combination of classifiers is developed. This method uses different pattern representations to show that a wide range of existing algorithms could be incorporated as the particular cases of compound classification where all the pattern representations are used jointly to make an accurate decision. Results: The results show that the combination rules developed under the Naive Bayes and Fuzzy integral method outperforms other classifier combination schemes. Conclusion: The performance of different combination schemes has been studied through an experimental comparison of different classifier combination plans. The dataset used in the article has been obtained from biological signals.

Fuzzy classification of OpenCL programs targeting heterogeneous systems

Heterogeneous systems featuring multiple kinds of processors are becoming increasingly attractive due to their high performance and energy savings over their homogeneous counterparts. With the OpenCL as a unified programming language providing program portability across different types of accelerators, finding the best task-to-device mapping will be the key to achieve such a high performance. We introduce in this work the design of a fuzzy logic classifier and the evaluation of its performance in classifying OpenCL workloads in a CPU-GPU-FPGA heterogeneous environment based on carefully analyzed kernel features. The classifier is designed as part of a scheduling scheme. Results demonstrate substantial improvement in accuracy when compared to other classifiers such as the K-Nearest- Neighbor (KNN), Support-Vector-Machine (SVM), Random-Forest (RF), Naïve-Bayes (NB) and the Bayes-Network (BN) with low computational complexity, facilitating run-time operation.

Dimensionality Reduction for Tensor Data Based on Local Decision Margin Maximization.

In machine learning, the idea of maximizing the margin between two classes is widely used in classifier design. Enlighted by the idea, this paper proposes a novel supervised dimensionality reduction method for tensor data based on local decision margin maximization. The proposed method seeks to preserve and protect the local discriminant information of the original data in the low-dimensional data space. Firstly, we depart the original tensor dataset into overlapped localities with discriminant information. Then, we extract the similarity and anti-similarity coefficients of each high-dimensional locality and preserve these coefficients in the embedding data space via the multilinear projection scheme. Under the combined effect of these coefficients, each dimension-reduced locality tends to be a convex set where strongly correlated intraclass points gather. Simultaneously, the local decision margin, which is defined as the shortest distance from the boundary of each locality to the nearest point of each side, will be maximized. Therefore, the local discriminant structure of the original data could be well maintained in the low-dimensional data space. Moreover, a simple iterative scheme is proposed to solve the final optimization problem. Finally, the experiment results on 6 real-world datasets demonstrate the effectiveness of the proposed method.

A detailed study about CDW-PSO, BWO and GM-CPSO methods on continuous function optimization

Gauss map based chaotic particle swarm optimization (GM-CPSO) is a state-of-the-art method involving the necessary chaotic map for PSO and has proved itself in global optimization, hybrid classifier design, etc. GM-CPSO can outperform recent techniques such as chaotic dynamic weight PSO (CDW-PSO), outclassing 20 optimization methods. However, the behavior of GM-CPSO on continuous characterized functions is unknown. As the main aim, this paper comprehensively determines the performance of GM-CPSO specifically on continuous function optimization. Black widow optimization (BWO) includes a non-stable population size that cannot be fixed. In BWO, the population can increase without any intervention during iterations, which prevents an objective comparison of the method with other methods. Thus, as the second aim, a new viewpoint on BWO population size selection is suggested for an objective comparison of the method. In various disciplines, stochastic optimization is inevitable to efficiently perform function optimization. Here, the necessary question concerns with which method the best convergence and performance can be achieved. As the third aim, we evaluate three state-of-the-art optimization methods to answer this question. To realize all of these aims, GM-CPSO is compared with CDW-PSO and BWO methods by using 10 continuous benchmark functions to perform a detailed comparison and reveal which one can achieve reliable scores on low-, middle-, and high-dimensional problems. Fitness-based comparisons, computation time analysis, and convergence-based evaluations are presented to determine the robustness of algorithms. As a result, GM-CPSO arises as the most remarkable method, especially for the middle-and high-dimensional continuous functions.

Optimal design of a general type-2 fuzzy classifier for the pulse level and its hardware implementation

. Nowadays, soft computing has been of great help in solving real-world problems and satisfying the needs in our everyday life. We require more than ever the development and implementation of models and techniques that assist in medical issues due to the recent pandemic. The aim of this work is the development and hardware implementation of a general type-2 fuzzy classifier for the pulse levels and the optimization of the general type-2 membership functions parameters using the Ant Lion Optimizer to compare results of type 1 and interval type-2 fuzzy classifiers with the patients of the Framingham database. This study additionally explains the implementation process of the general type-2 fuzzy classifier on the Jetson Nano hardware Development Board and the comparison of execution time with interval type-2, and type-1 fuzzy classifiers. This work offers a novel perspective in that the general type-2 fuzzy classifier can be implemented for embedded applications with excellent performance regarding hardware resources consumption.

IMPLEMENTASI TEORI NAIVE BAYES DALAM KLASIFIKASI CALON MAHASISWA BARU STMIK KHARISMA MAKASSAR

STMIK KHARISMA Makassar has graduated thousands of alumni since it was founded. However, the number of students registering is uncertain every year, although from 2016 to 2019 there has been an increase in the number of registrations. The problem is the percentage of the number of prospective students registering has actually decreased significantly. The purpose of this research is to implement the Naive Bayes theory in classification of STMIK KHARISMA Makassar prospective students. This research basically uses the Naive Bayes theory as a classifier, and is made using the Python programming language. At the classifier design stage, there were a total of 499 data collected from 2016 to 2019. The data was divided by a ratio of 80:20 for training data and test data. The result from the research indicate the level of accuracy of the classifier reaches 73%.

Ultrasound tissue classification: a review

Ultrasound imaging is the most widespread medical imaging modality for creating images of the human body in clinical practice. Tissue classification in ultrasound has been established as one of the most active research areas, driven by many important clinical applications. In this paper, we present a survey on ultrasound tissue classification, focusing on recent advances in this area. We start with a brief review on the main clinical applications. We then introduce the traditional approaches, where the existing research on feature extraction and classifier design are reviewed. As deep learning approaches becoming popular for medical image analysis, the recent deep learning methods for tissue classification are also introduced. We briefly discuss the FDA-cleared techniques being used clinically. We conclude with the discussion on the challenges and research focus in future.

Fault diagnosis based on the quantification of the fault features in a rotary machine

Fault diagnosis of rotary machinery is essential to fixing defective rotary machines and preventing rotary systems from breaking. Recently, fault diagnosis techniques have moved from traditional methods to artificial intelligence techniques. Many research groups have reported on the development of various artificial intelligence-based classifiers to improve diagnostic performance. In classifier design, selecting the datasets that include standard or fault features is essential to obtaining a high-performance classifier. However, there are few studies on the techniques to evaluate the quality of datasets numerically. In this study, we have developed a fault severity criterion to quantify the faultless and fault features of measured data. Using the proposed criterion, we have determined the dominant direction of representative faults in a rotary machine. The standard and fault data have been obtained, considering the dominant direction of each fault. The intrinsic mode function (IMF) that presents the fault features has been obtained by an empirical mode decomposition and a sensitive IMF selection criterion. Finally, we have designed an accurate and memory-efficient classifier using the extracted data and verified its performance by diagnosing a 7.5 kW servo motor. A conventional support vector machine has been used to verify the effects of the proposed algorithm on the classifier’s performance improvement. The developed classifier demonstrated an increase of performance up to an average of 51.9%, compared with the classifiers using training datasets measured in an arbitrary direction, with the detection rate of 99.9%. The study results suggest that the proposed classifier design technique based on the quantification of the fault features is useful for creating high-quality training datasets, machine learning, and deep learning-based classifiers.

A New Under-Sampling Method to Face Class Overlap and Imbalance

Class overlap and class imbalance are two data complexities that challenge the design of effective classifiers in Pattern Recognition and Data Mining as they may cause a significant loss in performance. Several solutions have been proposed to face both data difficulties, but most of these approaches tackle each problem separately. In this paper, we propose a two-stage under-sampling technique that combines the DBSCAN clustering algorithm to remove noisy samples and clean the decision boundary with a minimum spanning tree algorithm to face the class imbalance, thus handling class overlap and imbalance simultaneously with the aim of improving the performance of classifiers. An extensive experimental study shows a significantly better behavior of the new algorithm as compared to 12 state-of-the-art under-sampling methods using three standard classification models (nearest neighbor rule, J48 decision tree, and support vector machine with a linear kernel) on both real-life and synthetic databases.

Causative label flip attack detection with data complexity measures

A causative attack which manipulates training samples to mislead learning is a common attack scenario. Current countermeasures reduce the influence of the attack to a classifier with the loss of generalization ability. Therefore, the collected samples should be analyzed carefully. Most countermeasures of current causative attack focus on data sanitization and robust classifier design. To our best knowledge, there is no work to determinate whether a given dataset is contaminated by a causative attack. In this study, we formulate a causative attack detection as a 2-class classification problem in which a sample represents a dataset quantified by data complexity measures, which describe the geometrical characteristics of data. As geometrical natures of a dataset are changed by a causative attack, we believe data complexity measures provide useful information for causative attack detection. Furthermore, a two-step secure classification model is proposed to demonstrate how the proposed causative attack detection improves the robustness of learning. Either a robust or traditional learning method is used according to the existence of causative attack. Experimental results illustrate that data complexity measures separate untainted datasets from attacked ones clearly, and confirm the promising performance of the proposed methods in terms of accuracy and robustness. The results consistently suggest that data complexity measures provide the crucial information to detect causative attack, and are useful to increase the robustness of learning.

Neural-Network-Based Classification of Commercial Ships From Multi-Influence Passive Signatures

Monitoring the underwater environment is important for maritime security, marine conservation, and mine countermeasures. With developments in computation and artificial intelligence, it is increasingly important to measure and classify underwater ship signatures. In this work, we design an artificial neural network that classifies commercial ships based on their multi-influence signature. In total, 103 ship passages were included in the considered data set, with signatures recorded as the ship crossed a line of passive underwater sensors. The multi-influence signature was formed by feature-level sensor fusion of the hydroacoustic signature, the underwater electric potential, and the static and alternating magnetic signatures. Ships were classified according to size, or type, as broadcast on the AIS. With feature-level fusion, the neural network will optimize the relationship between different types of signatures, emphasizing features with greater predictive power. At the same time, weak features, even if not independently adequate for classification, can add information that improves accuracy further. The developed neural network achieved a classification accuracy of 87.4% when classifying according to size. With augmented data to balance the classes, 85.0% classification accuracy was achieved when classifying according to ship type. This is a large improvement on the found classification accuracy when using only hydroacoustic or electromagnetic signatures. This article verifies the value of feature-level sensor fusion in classification, and provides guidance on classifier design depending on the exact ship classification task.

Multi-View Representation Learning With Deep Gaussian Processes.

Multi-view representation learning is a promising and challenging research topic, which aims to integrate multiple data information from different views to improve the learning performance. The recent deep Gaussian processes (DGPs) have the advantages of good uncertainty estimates, powerful non-linear mapping ability and great generalization capability, which can be used as an excellent data representation learning method. However, DGPs only focus on single view data and are rarely applied to the multi-view scenario. In this paper, we propose a multi-view representation learning algorithm with deep Gaussian processes (named MvDGPs), which inherits the advantages of deep Gaussian processes and multi-view representation learning, and can learn more effective representation of multi-view data. The MvDGPs consist of two stages. The first stage is multi-view data representation learning, which is mainly used to learn more comprehensive representations of multi-view data. The second stage is classifier design, which aims to select an appropriate classifier to better employ the representations obtained in the first stage. In contrast with DGPs, MvDGPs support asymmetrical modeling depths for different views of data, resulting in better characterizations of the discrepancies among different views. Experimental results on real-world multi-view data sets verify the effectiveness of the proposed algorithm, which indicates that MvDGPs can integrate the complementary information in multiple views to discover a good representation of the data.

Hierarchical classifier design for speech emotion recognition in the mixed-cultural environment

ABSTRACT Recognition of emotion in speech is a difficult task due to many speaker factors like gender, age, and the cultural background (nationality, ethnicity, and region) as well as the acoustical environment. Among these factors, the cultural background of the speaker has a strong influence on the expression of emotion. The reason for the unsatisfactory performance of an emotion recognition engine built using mixed-cultural samples can be traced back to this. To address this issue, a two-level hierarchical engine has been designed to identify emotion from the speech of different cultural backgrounds. The first level of the hierarchical engine is a culture identification system, which identifies the corpus of an input utterance. As most of the speakers involved in the construction of a specific corpus are from the same locality and cultural background, we assume that a corpus represents the cultural background of the speakers of the corpus constructed. Based on the response of the first level classifier, the input utterance is forwarded to an appropriate corpus-specific emotion recognition engine, in the second level. Each corpus-specific emotion recognition system is a discriminative, multiclass SVM classifier, trained with the emotional utterances of that particular corpus. The system has been tested with five different corpora, collected from diverse cultural backgrounds, namely EMO-DB, SAVEE, IITKGP-SEC, Spanish corpus S0329, and CMU’s Woogles corpus. The system achieved an accuracy of 82.01% which is an improvement of 13.38% over monolithic approaches.

Classification of Medieval Documents

The present study is a comparative exploration of different classification tasks for Swedish medieval charters (transcriptions from the “SDHK” collection) and different classifier setups. In particular, we explored the identification of the issuer, place of issue, and decade of production. The experiments used features based on lowercased words and character 3- and 4-grams. We evaluated the performance of two learning algorithms: linear discriminant analysis and decision trees. For evaluation, five-fold cross-validation was performed. We report accuracy and macro-averaged F1 score. The validation made use of six labeled subsets of SDHK combining the three tasks with Old Swedish and Latin. Issuer identification for the Latin dataset (595 charters from 12 issuers) reached the highest scores, above 0.9, for the decision tree classifier using word features. The best accuracy for Old Swedish issuer identification was 0.81. Place and decade identification produced lower performance scores for both languages. Which classifier design is the best one seems to depend on peculiarities of the dataset and the classification task. The present study does however support the idea that text classification is useful also for medieval documents characterized by extreme spelling variation.

Structured feature for multi-label learning

Multi-label learning tackles the problem in which each instance is composed of a single sample and associated with multiple labels simultaneously. In the past decades, many algorithms have been proposed for this emerging machine learning paradigm. However, these methods often focus on designing new classification strategies. If the original input data with a high dimensionality, it will suffer from the curse of dimensionality and get highly biased estimates. To address the issue, this paper proposes a novel feature construction approach named Structured Feature (SF), which can achieve a discriminant representation for multi-label data by exploiting the local and global geometric structures. These structures are characterized by the reconstruction coefficients of data in manifold and sparse subspaces. Since reconstruction coefficients reflect the data similarity relationships, and the similar data usually has the common class labels, SF incorporates the similarity relationships and supervised information into the feature representation. The SF contains supervised information, thus, even though classifier design ignores the correlations and interdependences between labels, it is able to obtain satisfactory classification performance.SF combines manifold and sparse structures in the feature representation for multi-label learning. Extensive experiments validate that SF can achieve comparable to or even better results than other state-of-the-art algorithms.

Automatic Discontinuity Classification of Wind-turbine Blades Using A-scan-based Convolutional Neural Network

Recent development trends in wind power generation have increased the importance of the safe operation of wind-turbine blades (WTBs). To realize this objective, it is essential to inspect WTBs for any defects before they are placed into operation. However, conventional methods of fault inspection in WTBs can be rather difficult to implement, since complex curvatures that characterize the WTB structures must ensure accurate and reliable inspection. Moreover, it is considered useful if inspection results can be objectively and consistently classified and analyzed by an automated system and not by the subjective judgment of an inspector. To address this concern, the construction of a pressure- and shape-adaptive phased-array ultrasonic testing platform, which is controlled by a nanoengine operation system to inspect WTBs for internal defects, has been presented in this paper. An automatic classifier has been designed to detect discontinuities in WTBs by using an A-scan-imaging-based convolutional neural network (CNN). The proposed CNN classifier design demonstrates a classification accuracy of nearly 99%. Results of the study demonstrate that the proposed CNN classifier is capable of automatically classifying the discontinuities of WTB with high accuracy, all of which could be considered as defect candidates.

A Machine-Learning based generalization for an iterative Hybrid Embedded Fracture scheme

Accurately simulating fractured systems requires treating the fracture's characteristics. Here we describe a novel framework that involves coupling the Hybrid Embedded Fracture (HEF) scheme with Machine Learning. In general, HEF is more accurate than continuum medium schemes and less reliable but more efficient than the Discrete Fracture Networks (DFN) schemes. In our work, the attributes used to estimate the HEF flux exchange parameters are extracted using image processing, Machine-Learning, and Artificial-Intelligence techniques. In addition, we formulate a pure Machine-Learning classifier and Deep-Learning topology design to deal with the extraction of hierarchical fracture features from low-level to high-level based on Neural-Network layers. Computations are visualized using velocity vectors that are controlled by fractures characteristics extracted automatically from the fractured systems images. Their results provide an understanding of the flow behavior and maps of pressure distributions.

A Hybrid PSO-ANN-based Fault Classification System for EHV Transmission Lines

The paper presents a robust design of Artificial Neural Network Classifier used for the classification of faults occurring on Extra High Voltage Transmission lines with multiple series compensation. The feature selection and model parameters play a vital role in providing better accuracy as well as stability of Classifier in recognition of fault patterns. The paper describes the use of Particle Swarm Optimization (PSO)-assisted training of Artificial Neural Network (ANN). The PSO is used to find the optimal number of features to attain sufficiently high accuracy along with optimal design parameters of ANN so as to reduce the computational burden. A combined use of Multiresolution Wavelet Analysis (MRA) and Statistical Features (SF) module is shown for fault pattern analysis. A Particle Swarm optimization based Neural Network with Feature Selection (PSONNFS) Algorithm developed is applied on a large database of fault patterns with a wide range of operating conditions in series compensated system. The best features found from the PSONNFS algorithm are later used to find the optimal structure of ANN. The results obtained provide an enhanced model of ANN with a high degree of accuracy with optimality in Classifier parameters. A Genetic Algorithm based Neural Network optimization with Feature Selection (GANNFS) is also developed to verify the performance of the classifier. The results of the PSONNFS algorithm are further compared with the GANNFS algorithm. The results obtained show an enhanced model of ANN using PSONNFS which has a high degree of accuracy with optimality in Classifier parameters.

An Efficient, Parallelized Algorithm for Optimal Conditional Entropy-Based Feature Selection.

In Machine Learning, feature selection is an important step in classifier design. It consists of finding a subset of features that is optimum for a given cost function. One possibility to solve feature selection is to organize all possible feature subsets into a Boolean lattice and to exploit the fact that the costs of chains in that lattice describe U-shaped curves. Minimization of such cost function is known as the U-curve problem. Recently, a study proposed U-Curve Search (UCS), an optimal algorithm for that problem, which was successfully used for feature selection. However, despite of the algorithm optimality, the UCS required time in computational assays was exponential on the number of features. Here, we report that such scalability issue arises due to the fact that the U-curve problem is NP-hard. In the sequence, we introduce the Parallel U-Curve Search (PUCS), a new algorithm for the U-curve problem. In PUCS, we present a novel way to partition the search space into smaller Boolean lattices, thus rendering the algorithm highly parallelizable. We also provide computational assays with both synthetic data and Machine Learning datasets, where the PUCS performance was assessed against UCS and other golden standard algorithms in feature selection.