Search

Today's modern world is filled with uncertainties and contradictions. As artificial intelligence (AI) advances, machines are frequently expected to mimic the human brain and, consequently, face the conflicts associated with this task. To overcome them, feature engineering has emerged as the field of science responsible for turning raw data into relevant input information, setting up classifiers in the fused digital signal processing (DSP) and pattern recognition (PR) domain. Despite the ongoing efforts to improve feature learning, handcrafted extraction still plays a very important role. In this context, a careful choice of features is extremely relevant for creating an accurate classification. This article sheds light on the problem of feature quality by using a nonclassical logical system capable of handling conflictive situations. It is known as paraconsistent logic (PL). [ABSTRACT FROM AUTHOR]

Entropy ( H ) is the main subject of this article, concisely written to serve as a tutorial introducing two feature extraction (FE) methods for usage in digital signal processing (DSP) and pattern recognition (PR). The theory, carefully exposed, is supplemented with numerical cases, augmented with C/C++ source-codes and enriched with example applications on restricted-vocabulary speech recognition and image synthesis. Complementarily and as innovatively shown, the ordinary calculation of H corresponds to the outcome of a partially pre-tuned deep neural network architecture which fuses important information, bringing a cutting-edge point-of-view for both DSP and PR communities. [ABSTRACT FROM AUTHOR]

This tutorial, dedicated both to young professionals and students working with digital signal processing and pattern recognition, introduces three feature extraction approaches based on signal energy, characterising alternative and innovative ways for its use. The proposed theory, smoothly presented, is complemented with numerical examples, source-codes in C/C++ programming language, and applications in a diversity of computational problems, namely, neurophysiological signal processing, speech processing, and image processing. The lack of novelty in current energy-based approaches and the feasibility of a balance among creativity , simplicity , and accuracy constitutes the motivation for this text, which reveals how relevant the concept of signal energy may be, if properly employed. [ABSTRACT FROM AUTHOR]

Abstract: In this letter, we present different approaches for music genre classification. The proposed techniques, which are composed of a feature extraction stage followed by a classification procedure, explore both the variations of parameters used as input and the classifier architecture. Tests were carried out with three styles of music, namely blues, classical, and lounge, which are considered informally by some musicians as being “big dividers” among music genres, showing the efficacy of the proposed algorithms and establishing a relationship between the relevance of each set of parameters for each music style and each classifier. In contrast to other works, entropies and fractal dimensions are the features adopted for the classifications. [Copyright &y& Elsevier]

• Innovatively, speech pathology detection is carried out based on paraconsistency. • Discriminative Paraconsistent Machine (DPM) is the essence of the proposed approach. • Feature extractions is based on energy, zero-crossing rate and entropy, providing relevant information from raw speech data. • DPM allows for a wider inspection in contrast to binary results from usual classifiers. Voice disorders are related to both modest and severe health problems, including discomfort, pain, difficulty speaking, dysphagia and also cancer. Widely adopted worldwide, the combined invasive and subjective diagnosis of voice disorders is troublesome and error-prone. Contrarily, acoustic-based digital assessment allows for a non-intrusive and objective examination, stimulating the applications of computer-based tools. Consequently, this work describes a novel algorithm to investigate speech pathologies from the sounds of sustained vowels, particularly exploring a potential gap: the classification of co-existent issues for which the major phonic symptom is the same, implying in similar inter-class features. By using the concepts of signal energy (SE), zero-crossing rates (ZCRs) and signal entropy (SH), which provide a joint time-frequency-information map, the proposed approach classifies voice signals based on the discriminative paraconsistent machine (DPM), allowing for the application of paraconsistency to treat indefinitions and contradictions. An accuracy level of 95% was obtained under a subset of voices from the Saarbrucken voice database (SVD), with just a modest training. In complement, the proposed approach offers wider possibilities in contrast to current state-of-the-art systems, allowing for the inputs to be mapped into the paraconsistent plane in such a way that intermediary states can be found. Different from current algorithms, our technique focuses on a particular problem in the field of speech pathology detection (SPD), not yet explored in detail, proposing a way to successfully solve it. Furthermore, the results we obtained stimulate broaden studies involving speech data inconsistencies whilst providing a valid contribution. [ABSTRACT FROM AUTHOR]

[Display omitted] • Acoustic detection of cicadas of the species Quesada gigas is investigated. • A hybrid algorithm based on EMD-PFE-SVM methods is proposed. • Influence of EMD stopping criteria on classification is investigated. • Algorithm performance is validated by more than 18 million tests combining parameters. • Obtained performance is over 90.79% for all stopping criteria and test series. The sounds emitted by various insect species are highly specific and, thus, can be used as a way to acoustically characterize them. Consequently, acoustic insect detection has been widely studied by the scientific community in the field of pattern recognition. In Brazil, the cicada species Quesada gigas is considered a pest in coffee plantations, because the insects feed on the sap of the plants and can cause losses to farmers in mass attacks. Based on the fact that the most striking feature of cicadas is the emission of sounds for breeding purposes, this paper presents an alternative algorithm for acoustic detection of cicadas. The algorithm combines sound feature extraction with feature analysis based on Empirical Mode Decomposition (EMD) and Paraconsistent Feature Engineering (PFE), respectively, followed by a classification step based on a Support Vector Machine (SVM). Specifically, a study on the influence of eight EMD stopping criteria on the classification of sounds is presented. The results show that the proposed methodology can obtain accuracy values above 98 % considering the Energy Difference Tracking (EDT) stopping criterion, vectors with 18 features and at least 46 % of the vectors for SVM training. In the computational cost aspect, the stopping criterion Standard Deviation (SD) stands out, providing accuracy values above 96.67 % for vectors with only two features. These results show that this study is feasible for Internet of Things applications, favoring the development of detection devices for field use with long-lasting autonomy. Technologies like these can enable the implementation of more and more daring projects involving Smart Farms and e-waste, aiming to reduce impacts to the environment. Suggestions for future work based on the PFE are also presented. [ABSTRACT FROM AUTHOR]