Automatic Classification Task Research Articles

A Method for the Automatic Selection of Training Tasks in Learning Environment for IT Students

Purpose of research. The research, the results of which are presented in this article, was carried out in order to activate and improve the efficiency of independent work of students in the information environment of learning by rational individual selection of training tasks. In the process of the research, a method for automatically selecting tasks for self-completion was developed and implemented in the educational process, based on predicting the difficulty and learning effect of the task for a specific student, taking into account the complexity of the task and the student’s readiness to perform this task. Methods and materials. The article provides a distinction between the concepts of complexity, difficulty, and the learning effect of training tasks. On this basis, the task of predicting the level of difficulty of the task for the student is set as a task of automatic classification of “student-task" pairs, which represent a set of characteristics of the student and the task that are available in the database of the e-learning system. The result of the classification is a forecast of the level of difficulty of the task for the student, on the basis of which a decision is made about the learning effect of this task.The classification problem is one of the well-developed machine learning tasks “with a lecturer". Decision trees were selected from several well-known trained classification models for implementation, since they, unlike neural networks, represent prediction rules in a visual form, while highlighting significant features. The learning phase of the model consists of building a decision tree based on a training sample containing data on precedents for students to complete tasks. As a result of the computational experiment, decision trees were built for several disciplines that practice automatic verification of students’ decisions, i.e. there is data for forming a training sample.Results. The article provides an example of a decision tree based on a training sample, which is formed on the basis of data from an electronic workshop on the discipline “Foreign language ". The quality of the predictive model was determined on the exam sample by the criteria of accuracy and generalizing ability (the degree of severity of the “retraining effect”). The obtained values of these indicators allow us to recognize the quality as acceptable. The first results ofpractical application of the proposed method of selecting tasks in the educational process are analyzed. The software developed in the process of the research can be considered as the basis of a recommendation system that can not replace live communication between the student and the lecturer, but is their smart assistant in the learning process. Conclusion. In general, the results of the research show that the capabilities of artificial intelligence technologies, in particular, machine learning, allow us to put into practice the principle of individualized learning, to adapt the learning process to the individual characteristics of each student in order to effectively develop their professional competencies. The proposed method is implemented and tested in the information environment of training students of IT areas of Vologda State University, however, this approach is quite universal, and it can be extended to other subject areas and forms of training.

State-of-the-art methods in healthcare text classification system: AI paradigm.

Machine learning has shown its importance in delivering healthcare solutions and revolutionizing the future of filtering huge amountd of textual content. The machine intelligence can adapt semantic relations among text to infer finer contextual information and language processing system can use this information for better decision support and quality of life care. Further, a learnt model can efficiently utilize written healthcare information in knowledgeable patterns. The word-document and document-document linkage can help in gaining better contextual information. We analyzed 124 research articles in text and healthcare domain related to the ML paradigm and showed the mechanism of intelligence to capture hidden insights from document representation where only a term or word is used to explain the phenomenon. Mostly in the research, document-word relations are identified while relations with other documents are ignored. This paper emphasizes text representations and its linage with ML, DL, and RL approaches, which is an important marker for intelligence segregation. Furthermore, we highlighted the advantages of ML and DL methods as powerful tools for automatic text classification tasks.

Classification of authors for an automatic recommendation process for criminal responsibility

Abstract One problem in classifying tasks is the handling of features that characterize classes. When the list of features is long, a noise resistant algorithm of irrelevant features can be used, or these features can be reduced. Authorship attribution is a task that assigns an anonymous text to a subject on a list of possible authors, has been widely addressed as an automatic text classification task. In it, n-grams can produce long lists of features even in small corpora. Despite this, there is a lack of research exposing the effects of using noise-resistant algorithms, reducing traits, or combining both options. This paper responds to this lack by using contributions to discussion forums related to organized crime. The results show that the classifiers evaluated, in general, benefit from feature reduction, and that, thanks to such reduction, even classical algorithms outperform state-of-the-art classifiers considered highly noise resistant.

Automatic modulation recognition of compound signals using a deep multi-label classifier: A case study with radar jamming signals

The modern battlefield is getting more complicated due to the increasing number of different radiation sources as well as their fierce contention (interference) and confrontations (jamming) in the frequency spectrum. A radar, or a communication system usually has to struggle with multiple overlapped signals injected into its receiver to ensure desired system performance. Thus, the requirement for recognition of the modulation type of each constituent signal in a compound signal has emerged as a multiuser automatic modulation classification (mAMC) task in a signal processing field. This paper proposes a deep multi-label based mAMC framework (MLAMC) for compound signals which includes three serial steps, the time-frequency representation image (TFRI) extraction for signal preprocessing, multi-label convolutional neural network (MLCNN) construction for multi-label classification, and multi-decision thresholds optimization for output label decision. By applying the proposed MLAMC method on the compound radar jamming signals as a case study, the effectiveness and superiority of our proposed method are validated in four aspects of a smaller model size, better total performance, good extensibility for unseen signal combinations, and fine-grained analysis for recognition results.

Application of Natural Language Processing Algorithms to the Task of Automatic Classification of Russian Scientific Texts

This work is devoted to the study of applicability of modern methods of machine learning to the task of automatic classification of scientific articles and abstracts. For this purpose, the study of such models of machine learning as artificial neural networks, random forest, logistic regression, and support vector machine was carried out with taking into account such a feature of scientific texts as a large number of terms specific for various categories. Separately, the stages of data collection and extraction of text characteristics are considered. The results of research are used in development of a decision support system for assignment of scientific texts to the code of the department or abstract journal of All-Russian Institute of Scientific and Technical Information of Russian Academy of Sciences.

Automatic object classification according to 3d-lidar data based on single-photon counting technology

Subject of Research. We compare the effectiveness of the procedure for automatic object classification according to 3d-lidar data, built on the basis of single-photon counting technology, with the efficiency of the classification procedure with the usage of the conventional linear lidar data. Lidars based on the single-photons counting technology (SPCT) significantly exceed the ordinary linear lidars for a whole range of target parameters, including: spatial resolution - by more than an order of magnitude; the performance of the land surface scanning process - by an order of magnitude; by weight and size indicators - few times less. Unlike linear lidars, each laser point in the case of a SPCT-lidar is described not only by its coordinates and the intensity of the reflected signal, but by coordinates and an additional data block that characterizes the surface relief of the probed object in the direction of the probing photon flux. The presence of this additional data block for each laser point makes it possible to consider the received images as 3d-images, that simplifies the solution of not only the photogrammetric problem, but also the task of automatic objects classification on SPCT-image. Method. We consider the automatic object classification task solution based on SPCT-data with the use of the machine learning methods (XGBoost and multilayered neural networks - ANN). Main Results. The results of the numerical simulation carried out within the framework of the present study showed that, other things being equal, the efficiency of solving the classification problem based on SPCT-data, in practically important cases, increases up to 20% compared with the case where the linear-lidar data was used. Practical Relevance. The obtained results can be used in the design of the mobile SPCT-lidars enabling the detection and classification of objects on the Earth's surface in real time with high reliability.

Improving random forests by neighborhood projection for effective text classification

In this article, we propose a lazy version of the traditional random forest (RF) classifier (called LazyNN_RF), specially designed for highly dimensional noisy classification tasks. The LazyNN_RF “localized” training projection is composed by examples that better resemble the examples to be classified, obtained through nearest neighborhood training set projection. Such projection filters out irrelevant data, ultimately avoiding some of the drawbacks of traditional random forests, such as overfitting due to very complex trees, especially in high dimensional noisy datasets. In sum, our main contributions are: (i) the proposal and implementation of a novel lazy learner based on the random forest classifier and nearest neighborhood projection of the training set that excels in automatic text classification tasks, as well as (ii) a throughout and detailed experimental analysis that sheds light on the behavior, effectiveness and feasibility of our solution. By means of an extensive experimental evaluation, performed considering two text classification domains and a large set of baseline algorithms, we show that our approach is highly effective and feasible, being a strong candidate for consideration for solving automatic text classification tasks when compared to state-of-the-art classifiers.

Maya Codical Glyph Segmentation: A Crowdsourcing Approach

This paper focuses on the crowd-annotation of an ancient Maya glyph dataset derived from the three ancient codices that survived up to date. More precisely, nonexpert annotators are asked to segment glyph-blocks into their constituent glyph entities. As a means of supervision, available glyph variants are provided to the annotators during the crowdsourcing task. Compared to object recognition in natural images or handwriting transcription tasks, designing an engaging task and dealing with crowd behavior is challenging in our case. This challenge originates from the inherent complexity of Maya writing and an incomplete understanding of the signs and semantics in the existing catalogs. We elaborate on the evolution of the crowdsourcing task design, and discuss the choices for providing supervision during the task. We analyze the distributions of similarity and task difficulty scores, and the segmentation performance of the crowd. A unique dataset of over 9000 Maya glyphs from 291 categories individually segmented from the three codices was created and will be made publicly available thanks to this process. This dataset lends itself to automatic glyph classification tasks. We provide baseline methods for glyph classification using traditional shape descriptors and convolutional neural networks.

Model and technology of lingware support in tasks of the nuclear knowledge management system

Model and technology of lingware application in the tasks of the knowledge management are presented. The used model of thesauri paradigmatic relationships allows measuring semantic similarity of documents. An algorithm for compiling a semantic intersection of conceptual patterns based on the coinciding maximum principle is described. The semantic neighborhood construction in accordance with the value of semantic similarity measure and descriptors ranking allows using in the expansion of the search query the descriptors, whose semantics are closer due to the links, which reflect the common semantics of genus-species and associative relationships. An algorithm for thesauri used in automatic classification tasks is considered. The technology of conceptual hierarchical dynamic formation of structures is proposed. The proposed method of constructing a terminological network structure with hierarchical relationships for the collection of texts of a particular subject area is based on the sequential formation of lexicographical neighborhoods of “nuclear” terms, which are used as high-frequency nouns. To build up a conceptual structure a documental databases are used.

Classificação prosódica de marcadores discursivos

This work describes the discourse markers present in two corpora for European Portuguese, in different domains (university lectures and map-task dialogues). In this study, we also perform a multiclass automatic classification task based on prosodic features to verify in both corpora which words are discourse markers, which are disfluencies, and which are sentence like-units (SUs). Results show that the selection of discourse markers varies across domain and between speakers. As for the classification task, results show that the discourse markers are better classified in the lectures corpus (87%) than in the dialogue corpus (84%). However, cross-domain experiments evidenced that data trained with the dialogue corpus predicts better the events in the lecture corpus, since this domain displays more speakers and therefore complex patterns. In both corpora, markers are more easily classified as SUs than as disfluencies.

Reticular pattern detection in dermoscopy: an approach using Curvelet Transform

Introduction Dermoscopy is a non-invasive in vivo imaging technique, used in dermatology in feature identification, among pigmented melanocytic neoplasms, from suspicious skin lesions. Often, in the skin exam is possible to ascertain markers, whose identification and proper characterization is difficult, even when it is used a magnifying lens and a source of light. Dermoscopic images are thus a challenging source of a wide range of digital features, frequently with clinical correlation. Among these markers, one of particular interest to diagnosis in skin evaluation is the reticular pattern. Methods This paper presents a novel approach (avoiding pre-processing, e.g. segmentation and filtering) for reticular pattern detection in dermoscopic images, using texture spectral analysis. The proposed methodology involves a Curvelet Transform procedure to identify features. Results Feature extraction is applied to identify a set of discriminant characteristics in the reticular pattern, and it is also employed in the automatic classification task. The results obtained are encouraging, presenting Sensitivity and Specificity of 82.35% and 76.79%, respectively. Conclusions These results highlight the use of automatic classification, in the context of artificial intelligence, within a computer-aided diagnosis strategy, as a strong tool to help the human decision making task in clinical practice. Moreover, the results were obtained using images from three different sources, without previous lesion segmentation, achieving to a rapid, robust and low complexity methodology. These properties boost the presented approach to be easily used in clinical practice as an aid to the diagnostic process.

Classificação prosódica de marcadores discursivos

This work describes the discourse markers present in two corpora for European Portuguese, in different domains (university lectures and map-task dialogues). In this study, we also perform a multiclass automatic classification task based on prosodic features to verify in both corpora which words are discourse markers, which are disfluencies, and which are sentence like-units (SUs). Results show that the selection of discourse markers varies across domain and between speakers. As for the classification task, results show that the discourse markers are better classified in the lectures corpus (87%) than in the dialogue corpus (84%). However, cross‑domain experiments evidenced that data trained with the dialogue corpus predicts better the events in the lecture corpus, since this domain displays more speakers and therefore complex patterns. In both corpora, markers are more easily classified as SUs than as disfluencies.

Focal and diffused liver disease classification from ultrasound images based on isocontour segmentation

Preliminary diagnosis based on ultrasound scanning is the first step in the treatment of many abdominal diseases. The noisy nature of the ultrasound image coupled with minimal contrasting features complicates the task of automatic classification if not impossible. This study presents a segmentation-based approach to automatic classification of ten types of diffused and focal liver diseases from ultrasound images. A novel approach using Isocontour Segmentation based on Marching Squares, a computer graphics algorithm is presented. GLCM and fractal features are extracted from the segmented ultrasound images and classified using support vector machines and artificial neural networks (ANN) and the results are analysed. An overall classification accuracy of 92% is achieved using fractal features and ANN.

A fuzzy expert system for automatic seismic signal classification

Automatic classification of seismic events is of great importance due to the large amount of data received continuously. Seismic analysts classify events by visual inspection and calculation of event signal characteristics. This process is subjective and demands hard work as well as a significant amount of time and considerable experience. A reliable automatic classification task considerably reduces the effort required and makes classification faster and more objective. The aim of this study is to develop a fuzzy rule based expert classification system that is able to imitate human reasoning and incorporate the analyst’s knowledge of seismic event classification. The fundamental idea behind using this approach was motivated by the way in which human analysts classify seismic events based on a set of experiential rules. Additionally, this approach was chosen due to its interpretability and adjustability, as well as its ability to manage the complexity of real data. Relevant discriminant features are extracted from event signal. Using these features, the classification system was built based on the vote by multiple rule fuzzy reasoning method with three types of rules. Comparison of this method with the single winner classical fuzzy reasoning model was carried out. Classification results on real seismic data showed the robustness of the classifier and its capability to operate in on-line classification.

A Survey on Phrase Structure Learning Methods for Text Classification

Text classification is a task of automatic classification of text into one of the predefined categories. The problem of text classification has been widely studied in different communities like natural language processing, data mining and information retrieval. Text classification is an important constituent in many information management tasks like topic identification, spam filtering, email routing, language identification, genre classification, readability assessment etc. The performance of text classification improves notably when phrase patterns are used. The use of phrase patterns helps in capturing non-local behaviours and thus helps in the improvement of text classification task. Phrase structure extraction is the first step to continue with the phrase pattern identification. In this survey, detailed study of phrase structure learning methods have been carried out. This will enable future work in several NLP tasks, which uses syntactic information from phrase structure like grammar checkers, question answering, information extraction, machine translation, text classification. The paper also provides different levels of classification and detailed comparison of the phrase structure learning methods.

A Semantic Similarity Approach for Linking Tweet Messages to Library of Congress Subject Headings using Linked Resources: A Pilot Study

The objective of this study is to propose, implement, and test a framework of assigning relevant Library of Congress (LC) subject headings to tweet messages. In this study, the task of assigning LC headings is considered an automatic classification task that identifies relevant LC subject headings for given tweets. The classification task is conducted in two stages. In the first stage, tweets are clustered so that similar tweets are grouped together. In the second stage, the degree of similarity between a cluster of tweets and LC subject headings is measured by a popular similarity metric, Jaccard Coefficient (JC). In this pilot study, five selected tweet clusters and nine LC subject headings were carefully chosen and used. This pilot study demonstrates a positive result forthe proposed approach of identifying subject headings for tweets. In three cluster cases out of the five, JC selected the most relevant headings as the largest degrees of similarity. For the other two cases, JC was not successful in ranking the most relevant within the top three headings. In the next step, a more sophisticated clustering method will be explored and applied. Also, all possible LC subject headings will be employed to identify LC subjects for tweets in the next steps of this study.

Support Vector Machines and Kernel Functions for Text Processing

This work presents kernel functions that can be used in conjunction with the Support Vector Machine – SVM – learning algorithm to solve the automatic text classification task. Initially the Vector Space Model for text processing is presented. According to this model text is seen as a set of vectors in a high dimensional space; then extensions and alternative models are derived, and some preprocessing procedures are discussed. The SVM learning algorithm, largely employed for text classification, is outlined: its decision procedure is obtained as a solution of an optimization problem. The “kernel trick”, that allows the algorithm to be applied in non-linearly separable cases, is presented, as well as some kernel functions that are currently used in text applications. Finally some text classification experiments employing the SVM classifier are conducted, in order to illustrate some text preprocessing techniques and the presented kernel functions.

A new modified centroid classifier approach for automatic text classification

AbstractTo enhance the automatic text classification task, this paper proposes a novel approach for treating the problem of inductive bias incurred by the centroid classifier assumption. This approach is a trainable classifier, which takes into account tfidf as a text feature. The main goal of the proposed approach is to take advantage of the most similar training errors in the classification model for successively updating that model based on a certain threshold. The proposed approach is practical and flexible to implement. The complete performance of the proposed approach is measured at several threshold values on the Reuters‐21578 text categorization collection. Experimental results show that the proposed approach can improve the performance of the centroid classifier better than traditional approaches (traditional centroid classifier, support vector machines, decision trees, Bayes nets, and N Bayes) by 1, 1.2, 4.1, 7.5, and 11%, respectively. © 2013 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Gamma-Poisson Distribution Model for Text Categorization

We introduce a new model for describing word frequency distributions in documents for automatic text classification tasks. In the model, the gamma-Poisson probability distribution is used to achieve better text modeling. The framework of the modeling and its application to text categorization are demonstrated with practical techniques for parameter estimation and vector normalization. To investigate the efficiency of our model, text categorization experiments were performed on 20 Newsgroups, Reuters-21578, Industry Sector, and TechTC-100 datasets. The results show that the model allows performance comparable to that of the support vector machine and clearly exceeding that of the multinomial model and the Dirichlet-multinomial model. The time complexity of the proposed classifier and its advantage in practical applications are also discussed.

Discovering Time-Constrained Sequential Patterns for Music Genre Classification

A music piece can be considered as a sequence of sound events which represent both short-term and long-term temporal information. However, in the task of automatic music genre classification, most of text-categorization-based approaches could only capture temporal local dependencies (e.g., unigram and bigram-based occurrence statistics) to represent music contents. In this paper, we propose the use of time-constrained sequential patterns (TSPs) as effective features for music genre classification. First of all, an automatic language identification technique is performed to tokenize each music piece into a sequence of hidden Markov model indices. Then TSP mining is applied to discover genre-specific TSPs, followed by the computation of occurrence frequencies of TSPs in each music piece. Finally, support vector machine classifiers are employed based on these occurrence frequencies to perform the classification task. Experiments conducted on two widely used datasets for music genre classification, GTZAN and ISMIR2004Genre, show that the proposed method can discover more discriminative temporal structures and achieve a better recognition accuracy than the unigram and bigram-based statistical approach.