Vector and Betting - Vector Search and Vector Data Bases

A good vector representation has to be both compact as well as it has to capture the semantic similarity. The main reason to create these bector embedings is that nearness in the embedding space tells you something about nearness in your imagesor or in the things that you care about. This sort of idea extends to any sort of product oretail recommendation. On the other end of the spectrum, you can also do things like detecting near duplicates and images,.