Skip to main content

Machine Learning Introduction

 

Do you want to know what Machine Learning is all about in the AI field? Then let’s get started with the basic introduction in understanding ML models and datasets.

What is Machine Learning (ML)?

In normal terms for us humans learning means acquiring knowledge through studies, experience, or a lesson. Here it is a machine that is going to learn by itself without any human interference. Machine Learning is part of AI ( Artificial Intelligence).

So let’s see the actual definition of machine learning, the study of computer algorithms that can improve automatically through experience and use of data.

In Machine learning, the given datasets are divided into two halves. One is for training and another is for testing.

Datasets Division in Machine Learning:

  • Train dataset
  • Test dataset

The training dataset is always taken for building the ML models. The training dataset is also known as sample data. The accuracy of the model results is predicted or decided by giving the testing dataset.

Usually across the industries, 75% of the whole datasets is considered for the train dataset and the rest 30% is considered for the test dataset. And this is not the fixed rule of thumb, the ratio might vary based on the applications we build and based on the requirement.

How does Machine learn?

The machine learns by identifying the hidden patterns in the datasets.

Examples:

Data(X)    Result (Y)
0    0
1        1
2    8
7    343
5    125
10    ?

In the above-given datasets, the machine initially understands Y=X for the first 2 cases. i.e; 0=0 and 1=1. And when it comes to the third data point. The pattern identified is not right as Y<> X so here it starts to learn by looking into the pattern and identifies Y=X^3. So when 10 is given it gives 1000 (10^3). 

2) The group of red and green balls is mixed in the given datasets. Say now the machine needs to identify these red and green balls to different categories. A machine can identify the patterns( based on color, size, shape), and it groups them into one specific category.

Applications of Machine Learning

  • Email spam filtering
  • Speech Recognition
  • Image Recognition
  • Recommendation systems
  • Banking system
  • Speech Analysis
  • Anomaly detection
  • Fraud detection
  • Sentiment Analysis
  • Computer vision

What is a model in Machine Learning?

Machine Learning is statistical learning as most of the ML algorithms are built upon the mathematical equation. So when we say we build a model, we actually find the value of these mathematical equations.

For example, the underlying equation used in the case of Linear Regression is :

y = m x + c

where we find the value for slope (m) and intercept(c).

What are the different types of Machine Learning?

There are 4 major Machine Learning approaches. They are:

  • Supervised Learning: The feature and target columns are given to the machine to predict the desired output for a given input. Under the supervision of a teacher (target column acts as a teacher in ML models) the machine identifies the patterns. Algorithms: Linear Regression, Polynomial Regression, Logistic Regression, Random Forest.
  • Unsupervised Learning: Only the feature columns are given to the machine to predict the desired output for a given input. Here there is no supervision of a teacher (no target column). It can be a goal in itself. The machine identifies the hidden patterns by looking into the feature columns. Unsupervised can also be termed feature learning. Algorithms: K-means algorithm etc.
  • Semi-supervised Learning: As the name indicates semi, the learning falls between supervised learning and unsupervised learning. Here when a machine is given a combination of labeled (with target column) data and an unlabelled datasets. The machine can learn the hidden patterns and provide a better accuracy model. Algorithms: Customer sentiment analysis (GAN) etc.
  • Reinforcement Learning: As the name indicates reinforcement is the process of encouraging or establishing a belief or pattern of behavior. Here software agents are taught by taking actions in an environment. Based on the action taken, the machine can be rewarded or punished. Self Driving cars, playing games against a human opponent are popular examples of reinforcement learning. Algorithms: Deep Deterministic Policy Gradients (DDPG), Soft Actor-Critic (SAC), etc.

I hope you got a lot out of this Machine Learning guide. What are the other different applications (areas) where machine learning is widely used? Share them with me in the comments below!

Like this guide? Don’t forget to share it.

Together sharing is caring : )

Comments

Post a Comment

Popular posts from this blog

Types of Datasets in Machine Learning

  Do you want to build a machine learning model? Don’t know what Dataset is? Confused about which type of dataset to be used while building the model? (New to ML? Read our Machine Learning Introduction . ) Then let’s get started with the quick guide to machine learning datasets. What is a Machine Learning(ML) Dataset? Dataset as the name says its a set of data. Dataset is a  collection of data that is treated as a single unit for doing analytics and predictions . The dataset used in Machine learning problems can be a population or sample dataset. Most of the time the dataset used in machine learning is a sample dataset. Based on the patterns identified from this dataset the model makes predictions. Once the model is trained it is tested for accuracy and we look for the model working with the test dataset. Example: Let us consider the test scores dataset of a student. Subject Marked obtained      Performance Level En...
Post a Comment
Read more

What is Dimension in Machine Learning(ML)?

  This is the complete guide to understanding dimensions in machine learning. (New to ML ? )   Dimension in Machine Learning The number of input variables (or feature columns) in the given dataset i s termed as dimensions in machine learning. Example: Salary of employees based on designation and year of experience.   Emp_num Designation Years_of_experience Salary in 1000$ 51 Software Engineer 2 15 108 Software Developer 5 45 67 Software Tester 4 28 89 Data Analyst 5 50 Here, there are 3 feature or input variables. And hence dimension is 3 in this case. In the above example Enum, designation, and year_of exp are the feature columns and Salary in 1000$ is the label or output column.   What happens if you have high dimensions in the given dataset? This would be the same usual problem for the machine learning model as well to identify the patterns or relationships. Example: Salary of employees based on designation   Emp_num Designati...
Post a Comment
Read more