Core contribution:
- JEAP-enabled (self-supervised learning) cross-modal semantics (or knowledge) sharing/ transition facilitates two main benefits:
- We realized the ‘best’ traffic compression which means you may needn’t transmit anything, as what you want could be reconstructed from something related that you’ve already have (e.g., camera1 → camera2, or camera1 + wifi csi→ Camera2)
- Although Intuitive sensing has not been achieved so far, it must be another essential goal of this JEAP-enabled DMSN. let’s talk about it later.
Example (Yusuke)
Version 1
Version 2
Version 3
Version 4
Version 5
Version 6