Categories of Machine Learning System
Let’s look at more generalized ML systems
We all are pretty familiar with the names of supervised, unsupervised, reinforcement learning etc. There are more generalized terms that divide learning systems in a more basic way.Let’ take a look at those.
1.Supervised/Unsupervised Learning:
These learning systems come into play according to the amount and the type of supervision they get during training. Subcategories are supervised, unsupervised, semi-supervised, reinforcement learning.
Supervised learning:
Supervised learning involves training data as well as their desired output/solutions called labels. Classification and regression comes under this category.
Unsupervised learning:
Unsupervised , as you might guess involves unlabeled training set. In general the system tries to learn without a teacher. Clustering, anomaly/novelty detection, dimensionality/visualization reduction, association rule learning are a part of unsupervised learning.
Semi-supervised learning:
Semi-supervised learning comes when the data is partially labeled and mostly unlabeled. Most semi-supervised learning algorithms are combinations of supervised and unsupervised algorithms. Example: deep belief networks(DBNs).
Reinforcement learning:
Reinforcement learning the most advanced learning technique which can observe the environment,select and perform actions and get rewards in return that can be positive or negative.
2. Batch or Online learning:
This category checks whether or not the system can learn incrementally from a stream of incoming data.
Batch learning:
This learning system is incapable of learning incrementally, it must be trained using all the available data, which is also called offline learning.
Online learning:
In online learning, the system is trained incrementally by feeding it data instances sequentially, either individually or by small groups called mini-batches. The data flow is continuous and hence the learning system need to adapt to changes rapidly or autonomously.
3. Instance-based and model-based learning:
This category is all about generalization means how the model generalize.
Instance-based learning:
Instance-based learning is when the system learns the examples by heart and then generalizes to new cases by comparing them to the new examples, using a similarity measure. When I say learn by heart, it means for example, in case of spamming emails, the model that learned to spam emails that are identical to known spam emails, also flags emails that are very similar to known spam emails.
Model-based learning:
Another way to generalize from a set of examples is to build a model of these examples, then use that model to make predictions. This is called model-based learning.
Happy learning!
