Recognition of acoustic events using deep neural networks. Gencoglu, O., Virtanen, T., & Huttunen, H. In 2014 22nd European Signal Processing Conference (EUSIPCO), pages 506-510, Sep., 2014. Paper abstract bibtex This paper proposes the use of a deep neural network for the recognition of isolated acoustic events such as footsteps, baby crying, motorcycle, rain etc. For an acoustic event classification task containing 61 distinct classes, classification accuracy of the neural network classifier (60.3%) excels that of the conventional Gaussian mixture model based hidden Markov model classifier (54.8%). In addition, an unsupervised layer-wise pretraining followed by standard backpropagation training of a deep network (known as a deep belief network) results in further increase of 2-4% in classification accuracy. Effects of implementation parameters such as types of features and number of adjacent frames as additional features are found to be significant on classification accuracy.
@InProceedings{6952140,
author = {O. Gencoglu and T. Virtanen and H. Huttunen},
booktitle = {2014 22nd European Signal Processing Conference (EUSIPCO)},
title = {Recognition of acoustic events using deep neural networks},
year = {2014},
pages = {506-510},
abstract = {This paper proposes the use of a deep neural network for the recognition of isolated acoustic events such as footsteps, baby crying, motorcycle, rain etc. For an acoustic event classification task containing 61 distinct classes, classification accuracy of the neural network classifier (60.3%) excels that of the conventional Gaussian mixture model based hidden Markov model classifier (54.8%). In addition, an unsupervised layer-wise pretraining followed by standard backpropagation training of a deep network (known as a deep belief network) results in further increase of 2-4% in classification accuracy. Effects of implementation parameters such as types of features and number of adjacent frames as additional features are found to be significant on classification accuracy.},
keywords = {acoustic signal processing;backpropagation;Gaussian processes;hidden Markov models;mixture models;neural nets;signal classification;unsupervised learning;acoustic event recognition;deep neural networks;acoustic event classification task;neural network classifier;Gaussian mixture model;hidden Markov model classifier;unsupervised layer-wise pretraining;standard backpropagation training;deep belief network;adjacent frames;Acoustics;Accuracy;Training;Artificial neural networks;Hidden Markov models;Feature extraction;acoustic event classification;artificial neural networks;deep belief networks;deep neural networks;pattern classification},
issn = {2076-1465},
month = {Sep.},
url = {https://www.eurasip.org/proceedings/eusipco/eusipco2014/html/papers/1569924623.pdf},
}
Downloads: 0
{"_id":"Yxh7HfhQ4GLxwNXfW","bibbaseid":"gencoglu-virtanen-huttunen-recognitionofacousticeventsusingdeepneuralnetworks-2014","authorIDs":[],"author_short":["Gencoglu, O.","Virtanen, T.","Huttunen, H."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"firstnames":["O."],"propositions":[],"lastnames":["Gencoglu"],"suffixes":[]},{"firstnames":["T."],"propositions":[],"lastnames":["Virtanen"],"suffixes":[]},{"firstnames":["H."],"propositions":[],"lastnames":["Huttunen"],"suffixes":[]}],"booktitle":"2014 22nd European Signal Processing Conference (EUSIPCO)","title":"Recognition of acoustic events using deep neural networks","year":"2014","pages":"506-510","abstract":"This paper proposes the use of a deep neural network for the recognition of isolated acoustic events such as footsteps, baby crying, motorcycle, rain etc. For an acoustic event classification task containing 61 distinct classes, classification accuracy of the neural network classifier (60.3%) excels that of the conventional Gaussian mixture model based hidden Markov model classifier (54.8%). In addition, an unsupervised layer-wise pretraining followed by standard backpropagation training of a deep network (known as a deep belief network) results in further increase of 2-4% in classification accuracy. Effects of implementation parameters such as types of features and number of adjacent frames as additional features are found to be significant on classification accuracy.","keywords":"acoustic signal processing;backpropagation;Gaussian processes;hidden Markov models;mixture models;neural nets;signal classification;unsupervised learning;acoustic event recognition;deep neural networks;acoustic event classification task;neural network classifier;Gaussian mixture model;hidden Markov model classifier;unsupervised layer-wise pretraining;standard backpropagation training;deep belief network;adjacent frames;Acoustics;Accuracy;Training;Artificial neural networks;Hidden Markov models;Feature extraction;acoustic event classification;artificial neural networks;deep belief networks;deep neural networks;pattern classification","issn":"2076-1465","month":"Sep.","url":"https://www.eurasip.org/proceedings/eusipco/eusipco2014/html/papers/1569924623.pdf","bibtex":"@InProceedings{6952140,\n author = {O. Gencoglu and T. Virtanen and H. Huttunen},\n booktitle = {2014 22nd European Signal Processing Conference (EUSIPCO)},\n title = {Recognition of acoustic events using deep neural networks},\n year = {2014},\n pages = {506-510},\n abstract = {This paper proposes the use of a deep neural network for the recognition of isolated acoustic events such as footsteps, baby crying, motorcycle, rain etc. For an acoustic event classification task containing 61 distinct classes, classification accuracy of the neural network classifier (60.3%) excels that of the conventional Gaussian mixture model based hidden Markov model classifier (54.8%). In addition, an unsupervised layer-wise pretraining followed by standard backpropagation training of a deep network (known as a deep belief network) results in further increase of 2-4% in classification accuracy. Effects of implementation parameters such as types of features and number of adjacent frames as additional features are found to be significant on classification accuracy.},\n keywords = {acoustic signal processing;backpropagation;Gaussian processes;hidden Markov models;mixture models;neural nets;signal classification;unsupervised learning;acoustic event recognition;deep neural networks;acoustic event classification task;neural network classifier;Gaussian mixture model;hidden Markov model classifier;unsupervised layer-wise pretraining;standard backpropagation training;deep belief network;adjacent frames;Acoustics;Accuracy;Training;Artificial neural networks;Hidden Markov models;Feature extraction;acoustic event classification;artificial neural networks;deep belief networks;deep neural networks;pattern classification},\n issn = {2076-1465},\n month = {Sep.},\n url = {https://www.eurasip.org/proceedings/eusipco/eusipco2014/html/papers/1569924623.pdf},\n}\n\n","author_short":["Gencoglu, O.","Virtanen, T.","Huttunen, H."],"key":"6952140","id":"6952140","bibbaseid":"gencoglu-virtanen-huttunen-recognitionofacousticeventsusingdeepneuralnetworks-2014","role":"author","urls":{"Paper":"https://www.eurasip.org/proceedings/eusipco/eusipco2014/html/papers/1569924623.pdf"},"keyword":["acoustic signal processing;backpropagation;Gaussian processes;hidden Markov models;mixture models;neural nets;signal classification;unsupervised learning;acoustic event recognition;deep neural networks;acoustic event classification task;neural network classifier;Gaussian mixture model;hidden Markov model classifier;unsupervised layer-wise pretraining;standard backpropagation training;deep belief network;adjacent frames;Acoustics;Accuracy;Training;Artificial neural networks;Hidden Markov models;Feature extraction;acoustic event classification;artificial neural networks;deep belief networks;deep neural networks;pattern classification"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://raw.githubusercontent.com/Roznn/EUSIPCO/main/eusipco2014url.bib","creationDate":"2021-02-13T17:43:41.616Z","downloads":0,"keywords":["acoustic signal processing;backpropagation;gaussian processes;hidden markov models;mixture models;neural nets;signal classification;unsupervised learning;acoustic event recognition;deep neural networks;acoustic event classification task;neural network classifier;gaussian mixture model;hidden markov model classifier;unsupervised layer-wise pretraining;standard backpropagation training;deep belief network;adjacent frames;acoustics;accuracy;training;artificial neural networks;hidden markov models;feature extraction;acoustic event classification;artificial neural networks;deep belief networks;deep neural networks;pattern classification"],"search_terms":["recognition","acoustic","events","using","deep","neural","networks","gencoglu","virtanen","huttunen"],"title":"Recognition of acoustic events using deep neural networks","year":2014,"dataSources":["A2ezyFL6GG6na7bbs","oZFG3eQZPXnykPgnE"]}