Text Embeddings Reveal Almost As Much As Text - Iterative Embedding Inversion
Offered By: Yannic Kilcher via YouTube
Course Description
Overview
Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore a video lecture examining how text embeddings can potentially be used to reconstruct original embedded text under certain conditions. Learn about the Vec2Text method for iterative embedding inversion, its training process, and experimental results. Discover potential prevention techniques and considerations regarding sequence lengths. Gain insights into the privacy implications of text embeddings and their ability to reveal information about the original text.
Syllabus
- Intro
- Vec2Text: Iterative Embedding Inversion
- How to train this?
- Experimental results
- How can we prevent this?
- Some thoughts on sequence lengths
Taught by
Yannic Kilcher
Related Courses
Create Text Embeddings for a Vector Store using LangChainGoogle Cloud via Coursera Introduction to Vertex AI Embeddings: Text and Multimodal
Google Cloud via Coursera Product Recommender System: OpenAI Text Embedding
Coursera Project Network via Coursera Vector Search and Embeddings - Bahasa Indonesia
Google Cloud via Coursera Vector Search and Embeddings - Deutsch
Google Cloud via Coursera