Chapter 1 V.1
Why Join Vision and Language
Vision and language are joined because the supervision each provides is exactly what the other lacks.
Each modality supplies the supervision the other lacks.
How this chapter is built
M0Narrative
No mathematical load. Notation may appear; nothing is derived.
basics2/4what the words mean
concept2/2what to picture
theory—why it works, and when it does not
mathematics—derive it, then compute it
practice0/3build it, break it, read the papers
Five strands, not one. Mathematics is the spine; the other four are the body. A chapter cannot pay its way out of teaching with problems, nor out of problems with teaching.
Before you start
The problem
Images without captions have no semantics a model can name; text without images has no referent it can point at. Pairing them makes each a label for the other, at a scale neither could be annotated to reach.
What this chapter covers
- Separate modalities
- Representation spaces
- Multimodality
- Grounding
Propositions
Not yet written. The topics above are the plan for this chapter; each will become a proposition with its own figure.