Machine Learning

Machine learning is not a future promise, rather it is a current and developing technology. Whether it is the spam filter in your personal mail account or the online shopping site that suggest products of your likes, all these are examples of machine learning. In this blog we will draw an outline of different types of machine learning.

What is Machine Learning?

“[Machine Learning is the] field of study that gives computers the ability to learn without being explicitly programmed.”

—Arthur Samuel, 1959

Machine Learning is the science that programs a computer to learn from a set of sorted or unsorted set of data.

Types of Machine Learning

Based on different criteria, machine learning can be classified as follows:

1. Based on the need of Human Intervention (i.e., whether they require human supervision):
2. Supervised Learning: The type of machine learning in which the training set you feed to the algorithm includes labelled data. Some important supervised learning algorithms are K – Nearest neighbor’s, Linear Regression, Logistic Regression, Support Vector Machine (SVM), Decision Trees and Random Forests and Neural Networks.
3. Unsupervised Learning: The type of machine learning in which the training set you feed to the algorithm includes unlabeled data. Some important unsupervised learning are Clustering (K-Means, DBSCAN, Hierarchical Cluster Analysis), Anomaly detection and novelty detection (One-class SVM, Isolation Forest), Visualization and dimensionality reduction (Principal Component Analysis or PCA, Kernel PCA, Locally Linear Embedding or LLE, t – Distributed Stochastic Neighbor Embedding or t-SNE)
4. Semi – supervised Learning: This type of machine learning will often involve more unlabeled instances and few labelled instances. It is costly and time -consuming.
5. Reinforcement learning: In this type of machine learning the machine learns by itself by figuring out the best possible strategy. The learning system or the agent observes the environment and performs an action, and gets reward in return or penalties in the form of negative remarks.

2.  Based on the learning pattern: (i.e., whether or not the system can learn incrementally from a stream of incoming data)

1. Batch Learning: Also known as offline learning. It is time consuming and require lots of computing resources. In this type, first the system is trained and then it is launched for production. It runs without updating itself anymore and just applies what it has learned.
2. Online Learning: It is just the opposite of Batch learning. It trains the system incrementally by accepting the incoming data sequentially, either individually or in groups.

3. Based on working pattern (i.e., work by simply comparing new data points to the known data points or instead figure out new patterns and make predictions)

1. Instance – Based Learning:  In this method the system learns the examples by heart and then generalizes new cases by comparing it to the learned ones.
2. Model – Based Learning: In this type the system learns from the data given and derives some parameters. Based on this parameter it creates a model. And using this model it predicts the output.