Efficient Training Of Audio Transformers With Patchout

Abstract

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. In transformers, the compute and memory complexity is known to grow quadratically with the input length. Therefore, there has been extensive work on optimizing transformers, but often at the cost of degrading predictive performance. In this work, we propose a novel method to optimize and regularize transformers on audio spectrograms. Our 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 t

Stats

• citations181
• S2 citations—
• github stars379
• HF likes0
• heat score22.11
• arxiv keykoutini2021efficient

Related papers

• Dynamic Convolutional Neural Networks As Efficient Pre-trained Audio Models (2023)0.00
• Efficient Large-scale Audio Tagging Via Transformer-to-cnn Knowledge Distillation (2022)17.68
• Audiomer: A Convolutional Transformer For Keyword Spotting (2021)0.00
• Improved Zero-shot Audio Tagging & Classification With Patchout Spectrogram Transformers (2022)0.00
• SSAST: Self-supervised Audio Spectrogram Transformer (2021)17.61
• Asit: Local-global Audio Spectrogram Vision Transformer For Event Classification (2022)8.35
• Integrated Parameter-efficient Tuning For General-purpose Audio Models (2022)0.00
• Input-independent Attention Weights Are Expressive Enough: A Study Of Attention In Self-supervised Audio Transformers (2020)0.00