BIG-MoE: Bypass Isolated Gating MoE for Generalized Multimodal Face Anti-Spoofing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ma, Yingjie, Yu, Zitong, Lin, Xun, Xie, Weicheng, Shen, Linlin
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910761315467264
author Ma, Yingjie
Yu, Zitong
Lin, Xun
Xie, Weicheng
Shen, Linlin
author_facet Ma, Yingjie
Yu, Zitong
Lin, Xun
Xie, Weicheng
Shen, Linlin
contents In the domain of facial recognition security, multimodal Face Anti-Spoofing (FAS) is essential for countering presentation attacks. However, existing technologies encounter challenges due to modality biases and imbalances, as well as domain shifts. Our research introduces a Mixture of Experts (MoE) model to address these issues effectively. We identified three limitations in traditional MoE approaches to multimodal FAS: (1) Coarse-grained experts' inability to capture nuanced spoofing indicators; (2) Gated networks' susceptibility to input noise affecting decision-making; (3) MoE's sensitivity to prompt tokens leading to overfitting with conventional learning methods. To mitigate these, we propose the Bypass Isolated Gating MoE (BIG-MoE) framework, featuring: (1) Fine-grained experts for enhanced detection of subtle spoofing cues; (2) An isolation gating mechanism to counteract input noise; (3) A novel differential convolutional prompt bypass enriching the gating network with critical local features, thereby improving perceptual capabilities. Extensive experiments on four benchmark datasets demonstrate significant generalization performance improvement in multimodal FAS task. The code is released at https://github.com/murInJ/BIG-MoE.
format Preprint
id arxiv_https___arxiv_org_abs_2412_18065
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle BIG-MoE: Bypass Isolated Gating MoE for Generalized Multimodal Face Anti-Spoofing
Ma, Yingjie
Yu, Zitong
Lin, Xun
Xie, Weicheng
Shen, Linlin
Computer Vision and Pattern Recognition
In the domain of facial recognition security, multimodal Face Anti-Spoofing (FAS) is essential for countering presentation attacks. However, existing technologies encounter challenges due to modality biases and imbalances, as well as domain shifts. Our research introduces a Mixture of Experts (MoE) model to address these issues effectively. We identified three limitations in traditional MoE approaches to multimodal FAS: (1) Coarse-grained experts' inability to capture nuanced spoofing indicators; (2) Gated networks' susceptibility to input noise affecting decision-making; (3) MoE's sensitivity to prompt tokens leading to overfitting with conventional learning methods. To mitigate these, we propose the Bypass Isolated Gating MoE (BIG-MoE) framework, featuring: (1) Fine-grained experts for enhanced detection of subtle spoofing cues; (2) An isolation gating mechanism to counteract input noise; (3) A novel differential convolutional prompt bypass enriching the gating network with critical local features, thereby improving perceptual capabilities. Extensive experiments on four benchmark datasets demonstrate significant generalization performance improvement in multimodal FAS task. The code is released at https://github.com/murInJ/BIG-MoE.
title BIG-MoE: Bypass Isolated Gating MoE for Generalized Multimodal Face Anti-Spoofing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.18065