Skip to content

FOAA: Flattened Outer Arithmetic Attention For Multimodal Tumor Classification, ISBI (Oral) - 2024

Notifications You must be signed in to change notification settings

omniaalwazzan/FOAA

Repository files navigation

FOAA: Flattened Outer Arithmetic Attention For Multimodal Tumor Classification

Fusion models implemented in our paper 'FOAA: Flattened Outer Arithmetic Attention For Multimodal Tumor Classification' Omnia Alwazzan, Ioannis Patras, Gregory Slabaugh, ISBI 2024

The paper has been accepted by The IEEE International Symposium on Biomedical Imaging (ISBI) and will be available soon.

This paper serves as an extension to our previous methodology, MOAB. Here, we introduce a novel approach where the MOAB block is integrated into the attention mechanism, replacing the standard dot product operation between Query and Key in a cross-attention manner for multimodality fusion. For more comprehensive insights, please take a look at the paper's arXiv link. This repository provides all functional fusion methods that can be applied to any domain with any CNN.

  • Available fusion models

    • The file FOAA_one_branch can be adjusted by a user to incorporate
      • Cross Attention Outer Addition (OA)
      • Cross Attention Outer Product (OP)
      • Cross Attention Outer Division (OD)
      • Cross Attention Outer Subtraction (OS)
    • Cross OA+OP
    • Cross OA+OP+OS
    • CNN FOAA SA
    • FOAA

About

FOAA: Flattened Outer Arithmetic Attention For Multimodal Tumor Classification, ISBI (Oral) - 2024

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages