Efficient Training of Audio Transformers with Patchout

The great success of transformer-based models in natural language processing (NLP) has led to various attempts at adapting these architectures to other domains such as vision and audio. Recent work has shown that transformers can outperform Convolutional Neural Networks (CNNs) on vision and audio tasks. However, one of the main shortcomings of transformer models, compared to the well-established CNNs, is the computational complexity. Compute and memory complexity grow quadratically with the input length. Therefore, there has been extensive work on optimizing transformers, but often at the cost of lower predictive performance. In this work, we propose a novel method to optimize and regularize transformers on audio spectrograms. The proposed models achieve a new state-of-the-art performance on Audioset and can be trained on a single consumer-grade GPU. Furthermore, we propose a transformer model that outperforms CNNs in terms of both performance and training speed.1

[1]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[2]  Gerhard Widmer,et al.  Receptive Field Regularization Techniques for Audio Classification and Tagging With Deep Convolutional Neural Networks , 2021, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[3]  Quoc V. Le,et al.  SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition , 2019, INTERSPEECH.

[4]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[5]  Andrew Zisserman,et al.  Perceiver: General Perception with Iterative Attention , 2021, ICML.

[6]  Li Yang,et al.  Big Bird: Transformers for Longer Sequences , 2020, NeurIPS.

[7]  Lukasz Kaiser,et al.  Reformer: The Efficient Transformer , 2020, ICLR.

[8]  Nitish Srivastava,et al.  Dropout: a simple way to prevent neural networks from overfitting , 2014, J. Mach. Learn. Res..

[9]  Frank Hutter,et al.  Decoupled Weight Decay Regularization , 2017, ICLR.

[10]  Taejin Lee,et al.  Designing Acoustic Scene Classification Models with CNN Variants Technical Report , 2020 .

[11]  Georg Heigold,et al.  An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale , 2021, ICLR.

[12]  Karol J. Piczak.  ESC: Dataset for Environmental Sound Classification , 2015, ACM Multimedia.

[13]  Brian McFee,et al.  OpenMIC-2018: An Open Data-set for Multiple Instrument Recognition , 2018, ISMIR.

[14]  Matthieu Cord,et al.  Training data-efficient image transformers & distillation through attention , 2020, ICML.

[15]  Aren Jansen,et al.  Audio Set: An ontology and human-labeled dataset for audio events , 2017, 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[16]  Ramesh Nallapati,et al.  Multi-passage BERT: A Globally Normalized BERT Model for Open-domain Question Answering , 2019, EMNLP.

[17]  Annamaria Mesaros,et al.  Acoustic Scene Classification in DCASE 2020 Challenge: Generalization Across Devices and Low Complexity Solutions , 2020, DCASE.

[18]  Gerhard Widmer,et al.  The Receptive Field as a Regularizer in Deep Convolutional Neural Networks for Acoustic Scene Classification , 2019, 2019 27th European Signal Processing Conference (EUSIPCO).

[19]  James Glass,et al.  AST: Audio Spectrogram Transformer , 2021, Interspeech 2021.

[20]  Ming-Wei Chang,et al.  BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding , 2019, NAACL.

[21]  James Glass,et al.  PSLA: Improving Audio Tagging With Pretraining, Sampling, Labeling, and Aggregation , 2021, IEEE/ACM Transactions on Audio, Speech, and Language Processing.

[22]  Mark D. Plumbley,et al.  PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition , 2019, IEEE/ACM Transactions on Audio, Speech, and Language Processing.