AdaFlow: Non-blocking Inference with Heterogeneous Multi-modal Mobile Sensor Data

Fengmin Wu, Sicong Liu, Bin Guo, Xiaocheng Li, Yuan Gao, Zhiwen Yu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Multi-modal deep learning offer conceptual advantages by integrating information from different vantage points. However, data flow blocking and corruption due to modalities asymmetric which make exist methods can hardly balance latency and accuracy still pose many challenges. To address these challenges, this paper proposes AdaFlow to establish a multi-modal mapping mechanism using an affinity matrix, achieving non-blocking data flow in multi-modal systems.

Original languageEnglish
Title of host publicationProceedings - 2024 IEEE Coupling of Sensing and Computing in AIoT Systems, CSCAIoT 2024
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages8-9
Number of pages2
ISBN (Electronic)9798350363388
DOIs
StatePublished - 2024
Event2024 IEEE Coupling of Sensing and Computing in AIoT Systems, CSCAIoT 2024 - Hong Kong, China
Duration: 13 May 2024 → …

Publication series

NameProceedings - 2024 IEEE Coupling of Sensing and Computing in AIoT Systems, CSCAIoT 2024

Conference

Conference2024 IEEE Coupling of Sensing and Computing in AIoT Systems, CSCAIoT 2024
Country/TerritoryChina
CityHong Kong
Period13/05/24 → …

Keywords

  • Affinity Matrix
  • Multi-modal
  • Non-blocking Data Flow

Fingerprint

Dive into the research topics of 'AdaFlow: Non-blocking Inference with Heterogeneous Multi-modal Mobile Sensor Data'. Together they form a unique fingerprint.

Cite this