InfoMAE: Pair-Efficient Cross-Modal Alignment for Multimodal Time-Series Sensing Signals

Authors:
  • TKTomoyoshi Kimura
    Tomoyoshi Kimura
  • XLXinlin Li
    Xinlin Li
  • OHOsama Hanna
    Osama Hanna
  • YCYatong Chen
    Yatong Chen
  • YCYizhuo Chen
    Yizhuo Chen
  • DKDenizhan Kara
    Denizhan Kara
  • TWTianshi Wang
    Tianshi Wang
  • JLJinyang Li
    Jinyang Li
  • XOXiaomin Ouyang
    Xiaomin Ouyang
  • SLShengzhong Liu
    Shengzhong Liu
  • MSMani Srivastava
    Mani Srivastava
  • SDSuhas Diggavi
    Suhas Diggavi
Preprint
Publisher website
Google Scholar
Abstract
Standard multimodal self-supervised learning (SSL) algorithms regard cross-modal synchronization as implicit supervisory labels during pretraining, thus posing high requirements on the scale and quality of multimodal samples. These constraints significantly limit the performance of sensing intelligence in IoT applications, as the heterogeneity and the non-interpretability of time-series signals result in abundant unimodal data but scarce high-quality multimodal pairs. This paper proposes InfoMAE, a cross-modal alignment framework that tackles the challenge of multimodal pair efficiency under the SSL setting by facilitating efficient cross-modal alignment of pretrained unimodal representations. InfoMAE achieves \textit{efficient cross-modal alignment} with \textit{limited data pairs} through a novel information theory-inspired formulation that simultaneously addresses distribution-level and instance-level alignment. Extensive experiments on two real-world IoT applications are performed to evaluate InfoMAE's pairing efficiency to bridge pretrained unimodal models into a cohesive joint multimodal model. InfoMAE enhances downstream multimodal tasks by over 60% with significantly improved multimodal pairing efficiency. It also improves unimodal task accuracy by an average of 22%.
All Related Versions

This publication has 20 references indexed in Scilit: