Sound classification plays a crucial role in enhancing the interpretation, analysis, and use of acoustic data, leading to a wide range of practical applications, of which environmental sound analysis is one of the most important. In this paper, we explore the representation of audio data as graphs in the context of sound classification. We propose a methodology that leverages pre-trained audio models to extract deep features from audio files, which are then employed as node information to build graphs. Subsequently, we train various graph neural networks (GNNs), specifically graph convolutional networks (GCNs), GraphSAGE, and graph attention networks (GATs), to solve multi-class audio classification problems. Our findings underscore the effectiveness of employing graphs to represent audio data. Moreover, they highlight the competitive performance of GNNs in sound classification endeavors, with the GAT model emerging as the top performer, achieving a mean accuracy of 83% in classifying environmental sounds and 91% in identifying the land cover of a site based on its audio recording. In conclusion, this study provides novel insights into the potential of graph representation learning techniques for analyzing audio data.
Brain-computer Interfaces (BCIs) are control and communication systems based on acquisition and processing of brain signals to control a computer or an external device. Usually, BCI is focused in recognizing acquired events by different neuroimage methods, but the most used is the electroencephalography (EEG). Feature extraction over EEG signals for BCI systems is crucial to the classification performance. In this paper a comparison between various methods for feature extraction of EEG signals in BCI systems is presented. Different methodologies were taken into account, both at extracting frequency information from each electrode as for extracting shared information between electrodes.
Brain computer systems interface (BCI) is a continuing and growing field in evolution. During the last decades, many laboratories have begun to explore technologies BCI as a new communication option for people with neuromuscular disabilities that prevent them from using conventional augmentative communication methods. In this work is presented a methodology for the classification of motor imagination, using features extracted from power spectral density (PSD) as feature extraction technique. Two approaches are evaluated, first computing PSD features from the raw data and the second, performing a decomposition by means of the discrete wavelet transform (DWT). Obtained classification results show that features obtained through PSD with DWT achieves an accuracy of 85%, a sensitivity of 90% and a specificity of 80% while the results obtained with PSD over the raw data has a precision of 70%, one sensitivity 60% and specificity of 80%. Demonstrating that the performance of the classifier with the proposed preprocessing stage was improved.
En el año 2016, el Departamento Administrativo de Ciencia, Tecnología e Innovación (Colombiano) – Colciencias elaboró y publicó la Política para mejorar la calidad de las publicaciones científicas, en la que uno de los objetivos principales es «aumentar la presencia de las revistas científicas nacionales en los índices citacionales y bases de datos que las comunidades científicas de las diferentes disciplinas reconocen como espacios para la divulgación de resultados de investigación con alto impacto científico». En pro de este objetivo, Colciencias rediseñó el modelo de clasificación de las revistas científicas nacionales, que determina la indexación en el Índice Bibliográfico Nacional IBN-Publindex [1].
El desarrollo de Internet y las comunicaciones digitales generó uno de los mas grandes revolcones en la forma cómo se desarrolla la mayoría de las actividades en qué interviene el ser humano: y la divulgación científica no ha sido la excepción. La publicación en línea no sólo ha obligado a modificar los procesos editoriales de las revistas científicas, también ha obligado a replantear la función principal de estas. Mientras que hace algunos años las revistas científicas eran determinantes para decidir los contenidos que estarían disponibles para consulta por la comunidad científica, en la actualidad, el autor tiene la posibilidad de publicar sus documentos de forma autónoma y ponerlo a discusión en redes sociales académicas o blogs personales. En este escenario, las revistas científicas tienen la función principal de ser garantes de la calidad técnica, científica y del contenido de los trabajos que en ella se publican, y ofrecer a los autores la confianza de que su trabajo será evaluado de formar rigurosa y transparente.