Explaining Vector Embedding to all the moms.

In today’s modern AI time vector embedding is very import thing used in all the AI models but very less people know about this mostly non-tech people like our moms. So, Today I m going to try explaining vector embedding to our mums.

Now, let me start with explaining the meaning of vector embedding. In AI world vector is nothing much just the list of several number. Length can be any for every word the length will be different. e.g.: [1, 2.22, 334, ..]. In simple terms embedding just means to convert any word or sentence in vector form. We need this because AI is just a machine and machines do not understand our language like Hindi, English or any language you speak. They only understand numbers so to make them work we first embed our sentences/words into vectors.

So, suppose we say to our machines two words like “King” and “Queen” now they have to tell how close they are or better to say how they relate. So what our machines will do is first they convert both the words into vectors and if vectors of both the word are very close then it will define as they relate to each other. In this case vectors will be very close and it will give the result as they relate or close to each other. Let’s take another example of “shoe” and ”lion” when we convert these into vectors those will be very very far not even close to each other then on that basis our AI models will tell us they are in any way do not relate to each other.

Now with this I want to conclude that vector embedding is nothing just a model or technique we use in our ai models where they map each word in graph by first converting into vectors then they will map them on the graph point and the words which relate to each other you will see that those are very close to each other. And these graph is not our normal graph it has many dimensions each AI model has their own vector embedding of different dimensions which vary.