Types of machine learning | ML – Unsupervised machine learning


What are the types of unsupervised learning?

unsupervised learning diagram

Unsupervised Learning Algorithm :

After dealing with the supervised learning now lets discuss about the unsupervised learning.

It is the training of Artificial intelligence model/algorithm where the information are neither classified nor labeled and allowing the algorithm to act on that model without any guidance.
In this type of learning we don’t give target to our model while training. The training model has only input values and itself has to find which way it can learn. In the below data-set in Figure A we can see the mall data that contains information of its clients that subscribe to them. Once the users subscribed they are provided a membership card and so the mall owner can collect complete information about customer and also every purchase.

Using this data and unsupervised learning techniques, It can be very easy for the mall group clients data based on the parameters we are feeding in.

Training data we are feeding can be–

  • Unstructured data

    : May contain noisy(meaningless) error data, missing values or unknown data
  • Unlabeled data

    : Unlabeled data is pieces of data that have not been tagged with labels. Data only contains value for input values, there is no output values.

 

 

types of unsupervised learning

Types of Unsupervised Learning :-

  • Clustering Machine Learning:

Grouping of similar object in the set which is known cluster. Here the objects in one cluster when compared with the objects grouped under another cluster is different from each others.

It can be applied to group data based on different patterns, our machine model finds.

In the above case we are not given output parameter value, so this method can be used to group clients based on the input parameters provided by our data.

 

  • Association:

Rule-based machine learning techniques for discovering relations between variables in large databases

It finds out some very useful relations between parameters of a large data set. For e.g. Computer hardware stores use algorithms based on this technique to find out relationship between sale of one product w.r.t to others sale based on customer behaviours.

If such model get trained then it can be used to increase their sales by planning different offers based on the customer behaviours.

Some algorithms:

  • K-Means Clustering
  • Density-Based Spatial Clustering of Applications with noise
  • Balanced Iterative Reducing and Clustering using Hierarchies
  • Hierarchical Clustering

 
Semi-supervised Learning:

Semi-supervised learning is a class of machine learning tasks that makes the use of a small amount of labeled data with a large amount of unlabeled data for training purpose.
The name itself suggest, its working lies between Supervised and Unsupervised techniques and we use these techniques when we are working with a data whose small data is a labelled and rest large portion of data is unlabeled.

semi-supervised learning

For example, imagine you are developing a model for a large bank to detect fraud. The fraud can be of many type and some of them are known by you while other instances of fraud slipped by without your knowledge. You can label those dataset with the fraud instances you’re aware of, but the remaining of your data will remain unlabelled:

This technique is mostly applicable in image processing data-sets where usually all images are not labelled.


Reinforcement Learning:

Reinforcement learning is one of three basic machine learning paradigms, alongside supervised learning and unsupervised learning where the model keeps on increasing its performance using reward feedback to learn the behaviour.

These algorithms are precise to a particular problem. For example : Google Self Driving car, AlphaGo where there is competition between the human being and the bots and even itself to getting better and better performer. Each time we feed the data in, so that the model can increase the knowledge. So, if the model learn more then better it get trained and hence experienced.

  • Agents detect input.
  • Agent achieves an action by making some decisions.
  • After its performance, agent receives reward and accordingly reinforce and the model stores in state-action pair of information in the system.

Some algorithms:

  • Temporal Difference (TD)
  • Deep Adversarial Networks
  • Q-Learning
Sending
User Review
0 (0 votes)

Leave a Reply