Embedding

- Overview

Embeddings are digital representations of real-world objects that are used by machine learning (ML) and artificial intelligence (AI) systems to understand complex domains of knowledge in the same way humans do. 

For example, the calculation algorithm considers the difference between 2 and 3 to be 1, which indicates that 2 and 3 are more closely related than 2 and 100. However, real-world data contains more complex relationships. 

For example, bird's nest and lion's den are similar words, while day and night are opposite words. 

Essentially, embeddings enable ML models to find similar objects. Given a photo or a document, a ML model that uses embeddings could find a similar photo or document. Since embeddings make it possible for computers to understand the relationships between words and other objects, they are foundational for AI.

Technically, embeddings are vectors created by ML models for the purpose of capturing meaningful data about each object.

- Embeddings in Machine Learning

In ML, embeddings are low-dimensional representations of objects, such as images, words, and videos, that are used by ML models and semantic search algorithms. 

Embeddings use numerical values to represent similarities between objects, and the process of creating them is a hidden layer in a neural network. Neural networks are deep learning models that mimic the human brain's architecture. 

Embeddings transform real-world objects into complex mathematical representations that capture the inherent properties and relationships between real-world data. The entire process is automated, with the AI ​​system creating embeddings on its own during training and using them as needed to complete new tasks.