Decoding the World of Machine Learning: The Fundamentals You Need to Know

Decoding the World of Machine Learning: The Fundamentals You Need to Know

In this article, we will be looking at the fundamentals of machine learning and how mathematics is used in machine learning. But, at first look at Machine Learning in general

Machine Learning is the study of computer algorithms that improves themselves through data

In normal programming problems, we provide the data and some set of rules on which the data is processed and we expect the answer as output. In ML problems, we provide the answers and data to the program and the program finds the rules which connect the data to the answers. These relations are generated during the training process

Fundamental Ideas

  • All problems are data and all solutions are relations/functions: In ML problems, we need our answers and data to give to our algorithm as input. The work of the algorithm is to find out the relation between the answers and the data we provided

  • We convert qualitative inputs into numbers which become the input data. We call these numbers INPUT VECTORS

  • Similarly, the answers or the output we need after training the ML algorithm has to be also converted into numbers which are called OUTPUT VECTORS

  • As written earlier, the goal of that algorithm is to find out the relation between the INPUT and OUTPUT VECTORS

Steps in Machine Learning

  1. Gathering Data: We need to gather the data that is needed to train the machine

  2. Preparing Data: We need to prepare the data to be used in that algorithm

  3. Choosing a Model/ Algorithm that meets the needs

  4. Training the model on the data

  5. Testing Model

  6. Hyperparameter Tuning: Tuning the model using parameters whose values are used to control the learning process

The above 6 steps are steps of Training/ Testing the model

  1. Prediction: After everything is done, you will deploy your model and then it can be used for the prediction of your needs

Mathematics for Machine Learning

Linear Algebra

ML is about the mapping of INPUT VECTORS to their corresponding OUTPUT VECTORS. To map these vectors, we need to use matrix-based transformations

So, we need to know about Matrices and Vectors in Linear Algebra


There can be uncertainty in data, e.g if you want to make an algorithm, that predicts the prices of stocks, there can be uncertainty

So, we need to know about PROBABILITY


It is the science of collecting, analyzing, presenting and interpreting data. So, we need to use it in the process of Preparing data


Optimization of data is done using concepts of calculus