Diffusion Models as Masked Audio-Video Learners

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nunez, Elvis, Jin, Yanzi, Rastegari, Mohammad, Mehta, Sachin, Horton, Maxwell
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911748752146432
author Nunez, Elvis
Jin, Yanzi
Rastegari, Mohammad
Mehta, Sachin
Horton, Maxwell
author_facet Nunez, Elvis
Jin, Yanzi
Rastegari, Mohammad
Mehta, Sachin
Horton, Maxwell
contents Over the past several years, the synchronization between audio and visual signals has been leveraged to learn richer audio-visual representations. Aided by the large availability of unlabeled videos, many unsupervised training frameworks have demonstrated impressive results in various downstream audio and video tasks. Recently, Masked Audio-Video Learners (MAViL) has emerged as a state-of-the-art audio-video pre-training framework. MAViL couples contrastive learning with masked autoencoding to jointly reconstruct audio spectrograms and video frames by fusing information from both modalities. In this paper, we study the potential synergy between diffusion models and MAViL, seeking to derive mutual benefits from these two frameworks. The incorporation of diffusion into MAViL, combined with various training efficiency methodologies that include the utilization of a masking ratio curriculum and adaptive batch sizing, results in a notable 32% reduction in pre-training Floating-Point Operations (FLOPS) and an 18% decrease in pre-training wall clock time. Crucially, this enhanced efficiency does not compromise the model's performance in downstream audio-classification tasks when compared to MAViL's performance.
format Preprint
id arxiv_https___arxiv_org_abs_2310_03937
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Diffusion Models as Masked Audio-Video Learners
Nunez, Elvis
Jin, Yanzi
Rastegari, Mohammad
Mehta, Sachin
Horton, Maxwell
Sound
Computer Vision and Pattern Recognition
Multimedia
Audio and Speech Processing
Over the past several years, the synchronization between audio and visual signals has been leveraged to learn richer audio-visual representations. Aided by the large availability of unlabeled videos, many unsupervised training frameworks have demonstrated impressive results in various downstream audio and video tasks. Recently, Masked Audio-Video Learners (MAViL) has emerged as a state-of-the-art audio-video pre-training framework. MAViL couples contrastive learning with masked autoencoding to jointly reconstruct audio spectrograms and video frames by fusing information from both modalities. In this paper, we study the potential synergy between diffusion models and MAViL, seeking to derive mutual benefits from these two frameworks. The incorporation of diffusion into MAViL, combined with various training efficiency methodologies that include the utilization of a masking ratio curriculum and adaptive batch sizing, results in a notable 32% reduction in pre-training Floating-Point Operations (FLOPS) and an 18% decrease in pre-training wall clock time. Crucially, this enhanced efficiency does not compromise the model's performance in downstream audio-classification tasks when compared to MAViL's performance.
title Diffusion Models as Masked Audio-Video Learners
topic Sound
Computer Vision and Pattern Recognition
Multimedia
Audio and Speech Processing
url https://arxiv.org/abs/2310.03937