Prerequisite of mathematics in artificial intelligence: Math of AI

Introduction

Artificial intelligence is not something that you can master by just learning some mathematical concepts. You need to master basics pre-requisites of math called “Math of AI”. Mathematics and artificial intelligence are supposed to be the two branches of the same tree. Which are interlinked with each other? To master them you need a proper balance in both of them.

As of now most of the libraries and algorithm used in Artificial intelligence, totally or partially built on top of Math. So to Master AI you first need to master the “Mother of all Sciences called Mathematics“.

The basic fundamental concepts of “Math of AI” standby on 4 pillars of mathematics¬† ūüėé :

(1) Linear Algebra :

In ML/Deep learning/AI all the primary data build up is In the form of matrices & what does Linear algebra do -Operations on matrices.eg: Transpose, Multiplication.  Not only the matrix.

Linear algebra involves while you are creating graphs like 2D, 3D and creating Vectors, Creating Tensors.

When it comes down to understanding and creating the algorithms in machine learning like logistic regression, SVM, the basic ‚Äúcost‚ÄĚ functions for all the algorithms, all of it is done with linear algebra & ¬†If you want to know the math behind the algorithms, then at least basic linear algebra knowledge will be required..

(2)Probability:

In artificial intelligence, we are constantly dealing with uncertainty. It’s very little we can say for sure. Most of the time we have to settle for what is most likely and Probability theory is the Dominant framework for dealing with uncertainty.¬† In probability Generally, we are Dealing with 5 Outcome.

(i) Impossible Outcome

Impossible outcome

(ii) Unlikely outcome

Unlikely Outcome

(iii) Equally likely outcome

Equally likely Outcome

(iv) certain Outcome

Certain Outcome

(v)  Most Likely Outcome.

Most likely Outcome

Most of the machine learning algorithms used the concepts of Probability to create an analysis on the given data set and Using the probabilistic analysis they create a prediction on the test data set to predict the target value.

(3) Statistics:

Artificial intelligence(AI) is intrinsically data driven. The basic need to create an AI model is Data & the initial data we acquire from different sources is in the form of junk. To search the right data for our model we need to perform the Data Analysis, for process statistics turn to be pretty handy.

For that process we use two main statistical methods:

(1) Descriptive statistics, which summarize data from a sample using indexes such as the mean or standard deviation

(2)  Inferential statistics, which draw conclusions from data that are subject to random variation (e.g., observational errors, sampling variation)

The data we get after all that statistical process is the set of the features which are most relevant according to our task.

(4) Calculus:

We use calculus a lot in AI. As artificial intelligence algorithms are nothing but simple mathematical functions.

Whether it is –

  • A step function like ReLu, tanh.
  • An optimization function like Gradient Descent.
  • Cost function.

All these functions built on top of the conceptual application of differential and Integral Equations. And all these calculus concepts act as electricity that powers AI.

Conclusion:

Machine learning | AI | DL are built on mathematical principles like Calculus, Linear Algebra, Probability, Statistics, and Optimization, So before getting directly jumping off make sure you at know the basics otherwise you find this very daunting.

About the author

Vikram singh

Founder of Ai Venture,
An artificial intelligence specialist who teaches developers how to get results with modern AI methods via hands-on tutorials.
GANs are my favorite one.

View all posts