Multiclass voice commands classification with multiple binary convolution neural networks

• Autorzy: Szkoła Jarosław

Identyfikatory

DOI

10.31648/ts.8098

• Języki publikacji: EN

Abstrakty

EN

In machine learning, in order to obtain good models, it is necessary to train the network on a large data set. It is very often a long process, and any changes to the input dataset require re-training the entire network. If the model is extended with new decision classes, the entire learning process for all samples must be repeated. To improve this process, a new neural network architecture was proposed that uses a combination of multiple smaller independent convolutional neural networks (O’Shea, NaSh 2015, ZeghidOuret al. 2019) with two outputs, and a voting mechanism (COrNeliO et al. 2021, dONiNi et al. 2018) that ultimately determines the response of the network decision, rather than one large single network. The main purpose of using such an architecture is the need to solve the problem that occur in the case of most multiclass neural networks. For a typical neural network, extending with new decision classes requires changing the network architecture and re-learning the model for all data. In the proposed architecture, adding a new decision class requires only adding a small independent neural network, and the learning process applies to new cases with small subset of original dataset. This architecture is proposed for large datasets with many decision classes.

Słowa kluczowe

EN

multiclass convolution neural networks; voting decision mechanism; voice commands classification; multiclass classifier; sound wave processing; sound wave classification

• Wydawca: Wydawnictwo Uniwersytetu Warmińsko-Mazurskiego w Olsztynie
• Czasopismo: Technical Sciences / University of Warmia and Mazury in Olsztyn
• Rocznik: 2022
• Tom: nr 25(1)
• Strony: 149--170
• Opis fizyczny: bibliogr. 8 poz., rys., tab., wykr.

Twórcy

autor

Szkoła Jarosław

• Instytut Informatyki, Zakład Metod Przybliżonych, Uniwersytet Rzeszowski, ul. Pigonia 1, 35-310 Rzeszów

• https://orcid.org/0000-0002-6043-3313

Bibliografia

• Cornelio C., Donini M., Loreggia A., Pini M.S., Rossi F. 2021. Voting with random classifiers (VORACE): theoretical and experimental analysis. Autonomous Agents and Multi-Agent Systems, 35(22). https://doi.org/10.1007/s10458-021-09504-y.
• Donini M., Loreggia A., Pini M.S., Rossi F. 2018. Voting with Random Neural Networks: a Democratic Ensemble Classifier. RiCeRcA 2018. arXiv:1909.08996. https://doi.org/10.48550/arXiv.1909.08996.
• Hoffmann J., Borgeaud S., Mensch A., Buchatskaya E., Cai T., Rutherford E., de Las Casas D., Hendricks L.A., Welbl J., Clark A., Hennigan T., Noland E., Millican K., van den Driessche G., Damoc B., Guy A., Osindero S., Simonyan K., Elsen E., Rae J.W., Vinyals O., Sifre L. 2022.Training Compute-Optimal Large Language Models. https://arxiv org/abs/2203.15556. https://doi.org/10.48550/arXiv.2203.15556.
• O’Shea K., Nash R. 2015. An Introduction to Convolutional Neural Networks. arXiv:1511.08458. https://doi.org/10.48550/arXiv.1511.08458.
• Shafahi A., Saadatpanah P., Zhu Ch., Ghiasi A. , Studer C., Jacobs D., Goldstein T. 2020. Adversarially Robust Transfer Learning. ICLR 2020 Conference Blind Submission. https://openreview.net/pdf?id=ryebG04YvB.
• Warden P. 2017. Speech Commands: A public dataset for single-word speech recognition. http://download.tensorflow.org/data/speech_commands_v0.01.tar.gz.
• Warden P. 2018. Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition. arXiv:1804.03209. https://doi.org/10.48550/arXiv.1804.03209.
• Zeghidour N., Xu Q., Liptchinsky V., Usunier N., Synnaeve G., Collobert R. 2019. Fully Convolutional Speech Recognition. arXiv:1812.06864. https://doi.org/10.48550/arXiv.1812.06864.

Uwagi

Opracowanie rekordu ze środków MEiN, umowa nr SONP/SP/546092/2022 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2022-2023)