Machine Learning
in Practice

This lesson explains the concept of machine learning (ML). The process of training ML models is clarified. You will also become familiar with the concepts of parameters and hyperparameters. Hyperparameters used in the practical part of the course will be especially explained.

In this short introduction lesson, you will learn about the methods used in machine learning.

This lesson describes unsupervised learning used when data is neither classified nor labelled.

Supervised learning is the most often used method in machine learning. This lesson explains the fundamentals concerning this type of learning. Supervised learning will be used in the practical part of the course.

Reinforcement learning is the closest machine learning type to how humans learn. It utilizes a trial-and-error approach with a feedback-based process that allows the agent to learn from its experiences. See, how it works.

In the process of learning, three separate datasets are used: train, validation, and test. This lesson explains these datasets, and the role they play in the learning process.

This lesson explains some data pre-processing methods, including elimination of missing values, normalisation and one-hot encoding method.

In this lesson we will explain the concept of tensor. In the context of computer science and machine learning, a tensor refers to a data structure that stores and manipulates numerical data.

To work with the machine learning model you have to use an environment. In this lesson we explain Colaboratory, the Google tool for working with machine learning models.

In this short lesson we will explain the MNIST dataset.

In this lesson we will start from downloading the data. We will find out how numerous the train and test datasets are. Then we will display some digits and their labels.

To prepare data for training we will use reshape method, standardise data and use one-hot-encoding for labels.

In this lesson we will create and train the machine learning model for handwritten digit recognition. There are a considerable number of possibilities. We will create a deep neural network using the keras library.

In this lesson we will verify, if our model is well trained. To do this, we will create flow charts of the training and validation processes. We will also evaluate the model using the test data.

Now, we will create our own digit using the Paint application, to check if the model is able to recognise our digit.

In this lesson we will see if our model, trained using MNIST dataset, is able to recognise a digit drawn by us using the Paint application.