Abstract
Weakly labeled data is hard to classify as it does not contain any time tag of the given data making it harder to identify the target to be trained. Previous works have used joint detection classification model using detector and classifier together to identify the presence of events in frames and classify them later. To increase the efficiency of the neural network, several models use very deep models with arbitrary depth but such model plateau very quickly and take long time to train. We propose skip layer connections in deep neural networks so rather than passing previous layer to the next layer, we pass identity of previous layer output and input. Due to this, the model can decide for itself how much of the previous layer output is necessary, thus pseudo-adapting their depth.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Guo G, Li SZ (2003) Content-based audio classification and retrieval by support vector machines. IEEE Trans Neural Netw 14(1):209–215
Stowell D, Giannoulis D, Benetos E, Lagrange M, Plumbley MD (2015) Detection and classification of acoustic scenes and events. IEEE Trans Multimedia 17(10):1733–1746
tp://tubularinsights.com/youtube-300-hours/
Kumar A, Raj B (2016) Audio event detection using weakly labeled data. In: Proceedings of the 24th ACM international conference on Multimedia. ACM
Mesaros A, Heittola T, Virtanen T (2016) Tut database for acoustic scene classification and sound event detection. In: 24th European signal processing conference, vol 2016
Kong Q, Xu Y, Wang W, Plumbley MD (2017, March) A joint detection-classification model for audio tagging of weakly labelled data. In: 2017 IEEE international conference on acoustics, speech and signal processing (ICASSP). IEEE, pp 641–645
Xu Y, Kong Q, Huang Q, Wang W, Plumbley MD (2017, May) Convolutional gated recurrent neural network incorporating spatial features for audio tagging. In: 2017 international joint conference on neural networks (IJCNN). IEEE, pp 3461–3466
Cakır E, Heittola T, Virtanen T (2016) Domestic audio tagging with convolutional neural networks. In: IEEE AASP challenge on detection and classification of acoustic scenes and events (DCASE 2016)
Phan H, Krawczyk-Becker M, Gerkmann T, Mertins A (2017) DNN and CNN with weighted and multi-task loss functions for audio event detection. arXiv preprint arXiv:1708.03211
Kong Q, Xu Y, Plumbley MD (2017, August) Joint detection and classification convolutional neural network on weakly labelled bird audio detection. In: 2017 25th European signal processing conference (EUSIPCO). IEEE, pp 1749–1753
Dennis J, Tran HD, Li H (2011) Spectrogram image feature for sound event classification in mismatched conditions. IEEE Sig Process Lett 18(2):130–133
Wyse L (2017) Audio spectrogram representations for processing with convolutional neural networks. arXiv preprint arXiv:1706.09559
He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 770–778
Foster P, Sigtia S, Krstulovic S, Barker J, Plumbley MD (2015) CHiME-Home: a dataset for sound source recognition in a domestic environment. In: Proceedings of the 11th workshop on applications of signal processing to audio and acoustics (WASPAA)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Sisodia, S.P.S., Sisodia, R.P.S., Bharath, K.P., Muthu, R.K. (2021). Tagging of Weakly Labeled Acoustic Data Using Skip Layer Connection Detection Classification Model. In: Komanapalli, V.L.N., Sivakumaran, N., Hampannavar, S. (eds) Advances in Automation, Signal Processing, Instrumentation, and Control. i-CASIC 2020. Lecture Notes in Electrical Engineering, vol 700. Springer, Singapore. https://doi.org/10.1007/978-981-15-8221-9_5
Download citation
DOI: https://doi.org/10.1007/978-981-15-8221-9_5
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-15-8220-2
Online ISBN: 978-981-15-8221-9
eBook Packages: EngineeringEngineering (R0)