Types of Machine Learning
Machine learning can be classified into many categories based on different criterias. We will see the How it can be classified on the basis of Human supervision.
Human supervision
Wheather or not it is trained with human supervision or not
- Supervised learning
- Unsupervised learning
- SemiSipervised learning
- Reinforcement learning
Supervised learning
When we have data that includes both the inputs(features) and the desired output (label) and we train a machine learning model on that then it is called Supervised Learning.
Example : we have data regarding the spam mails that include the label wheater they are spam or not
like above if we have lot of spams and hams data we train a model and ask for a new one whether it is spam or not
Unsupervised learning
Let's us suppose you surveyed a group of people and collected data. Now, you want to know which groups are sharing common connection.
There can a group who like coffee and shop online atleast twice a week.
or
the other group who didn't like coffie and they shop online very rarely.
I know that doesn't make sense , but let's assume they mug up all the coffee in a week that's why they shop frequently.
anyway, that's the task for Unsupervised learning
Some of the task for Unsupervised learning are
- Clustering into groups
- Anomaly detection (detecting outliers)
- Dimensional Analysis (reducing dimensions)
- Association rule ( detection if user like this then he/she will also like that. Something like Amazon recommendations)
|
Clustering |
SemiSupervised learning
Something we have unlabeled data or few data are labeled. Some of the algorithm are there, which dealt with semi labeled data
Most of them are the combination of a Supervised and an Unsupervised algorithms.
For Example:
Google photos have algorithm, that automatically recognise the faces and group them together. It forms different clusters (unsupervised) on the basis of the faces appeared in the image . Now if we give a name to any image then it runs a supervised algorithm to find similar images and label all of them.
Reinforcement learning
Reinforcement learning is very different from the above techniques. Here the learner is called an agent . It observes the environment select and perform actions and get reward or panelties in return. It learns by it's own .
For example:
Robots often implement Reinforcement learning to learn different tasks.
On May 2017, a reinforcement learning algorithms defeated the world champion in the game of
Go. It learns the game by playing it millions of time.
That's all for now. See you soon
Comments
Post a Comment