MAMA: Meta-optimized Angular Margin Contrastive Framework for Video-Language Representation Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nguyen, Thong, Bin, Yi, Wu, Xiaobao, Dong, Xinshuai, Hu, Zhiyuan, Le, Khoi, Nguyen, Cong-Duy, Ng, See-Kiong, Tuan, Luu Anh
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916430694318080
author Nguyen, Thong
Bin, Yi
Wu, Xiaobao
Dong, Xinshuai
Hu, Zhiyuan
Le, Khoi
Nguyen, Cong-Duy
Ng, See-Kiong
Tuan, Luu Anh
author_facet Nguyen, Thong
Bin, Yi
Wu, Xiaobao
Dong, Xinshuai
Hu, Zhiyuan
Le, Khoi
Nguyen, Cong-Duy
Ng, See-Kiong
Tuan, Luu Anh
contents Data quality stands at the forefront of deciding the effectiveness of video-language representation learning. However, video-text pairs in previous data typically do not align perfectly with each other, which might lead to video-language representations that do not accurately reflect cross-modal semantics. Moreover, previous data also possess an uneven distribution of concepts, thereby hampering the downstream performance across unpopular subjects. To address these problems, we propose MAMA, a new approach to learning video-language representations by utilizing a contrastive objective with a subtractive angular margin to regularize cross-modal representations in their effort to reach perfect similarity. Furthermore, to adapt to the non-uniform concept distribution, MAMA utilizes a multi-layer perceptron (MLP)-parameterized weighting function that maps loss values to sample weights which enable dynamic adjustment of the model's focus throughout the training. With the training guided by a small amount of unbiased meta-data and augmented by video-text data generated by large vision-language model, MAMA improves video-language representations and achieve superior performances on commonly used video question answering and text-video retrieval datasets. The code, model, and data have been made available at https://nguyentthong.github.io/MAMA.
format Preprint
id arxiv_https___arxiv_org_abs_2407_03788
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle MAMA: Meta-optimized Angular Margin Contrastive Framework for Video-Language Representation Learning
Nguyen, Thong
Bin, Yi
Wu, Xiaobao
Dong, Xinshuai
Hu, Zhiyuan
Le, Khoi
Nguyen, Cong-Duy
Ng, See-Kiong
Tuan, Luu Anh
Computer Vision and Pattern Recognition
Computation and Language
Data quality stands at the forefront of deciding the effectiveness of video-language representation learning. However, video-text pairs in previous data typically do not align perfectly with each other, which might lead to video-language representations that do not accurately reflect cross-modal semantics. Moreover, previous data also possess an uneven distribution of concepts, thereby hampering the downstream performance across unpopular subjects. To address these problems, we propose MAMA, a new approach to learning video-language representations by utilizing a contrastive objective with a subtractive angular margin to regularize cross-modal representations in their effort to reach perfect similarity. Furthermore, to adapt to the non-uniform concept distribution, MAMA utilizes a multi-layer perceptron (MLP)-parameterized weighting function that maps loss values to sample weights which enable dynamic adjustment of the model's focus throughout the training. With the training guided by a small amount of unbiased meta-data and augmented by video-text data generated by large vision-language model, MAMA improves video-language representations and achieve superior performances on commonly used video question answering and text-video retrieval datasets. The code, model, and data have been made available at https://nguyentthong.github.io/MAMA.
title MAMA: Meta-optimized Angular Margin Contrastive Framework for Video-Language Representation Learning
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2407.03788