Machine Learning - A High level overview
Machine Learning is automated learning through algorithms by machines or by computer systems. It'd data analytics technique and is a subset of Artificial Intelligence. It is a methodology that makes computer system or machine to learn from experience like how human beings do. In CRISP - DM the Cross-Industry Standard Process framework for Data Mining the fourth stage is the Modelling stage. In the Modelling stage Machine Learning Models are prepared.
And where the Machine Learning models are being used?
Machine learning is applied where there is need for computer system to learn and to interpret without pre-coded programs. Machine Learning Models are used in Predictive analytics.
In day today life we are asking Apple’s Siri or Google Assistant in the mobile phone “Get me the route to Central Library” and we have the route map displayed on the screen from the current location. The mobile phone has recognized the voice and it responded. The process of recognizing voice is called as Speech Recognition and the underlying technology is Machine Learning Model which predicts the best route or routes. You are adding photos in your Facebook page and FB is asking whether you want to tag a friend. This is possible as FB is running Face Recognition application which in turn supported by Machine Learning algorithm that identifies friends in the photo.
The Machine Learning models based on the task they perform and the nature of the output are classified as :
Supervised Machine Learning
Supervised machine learning algorithms uncover insights, patterns, and relationships from a labelled training dataset. Here the target or output is known. Supervised Machine Learning is further categorized into :
Ø Regression : and it’s application
Regression predicts a numerical value based on previous observed data.
Application : Based on performances in previous matches sports analyst predict the number of runs a batsman would take in next match. Simple Linear Regression is the handy tool used by the analyst here.
Ø Classification : and it’s application
Classification predicts the category the data belongs to.
Application : Our email has Spam folder which has the set of mails that are classified as SPAM and most of the time the classification is right. In case Spam folder has non-spam mail wrongly classified spam, we are marking the mail is non-spam and the Machine Learning Model in e-mail server learns from our action. Next time when one such mail is received, right classification will be done by the mail server.
Unsupervised Learning Method
In unsupervised learning, an AI system is presented with unlabelled, uncategorized data and the system’s algorithms act on the data without prior training.
Ø Clustering : and it’s application
Clustering is a Machine Learning technique that involves the grouping of data points.
Application : In marketing there is a need for finding groups of customers with similar behaviour given a large database of customer data containing their properties and past buying records.
Ø Anomaly detection : and it’s application
Anomaly detection is the identification of rare items, events, mismatch in logic, finding outliers in given dataset.
Application : In cyber security engineer has to identify the intrusion attempts and exploits and anomaly detection algorithms helps the engineer in billions of connection attempts.
Ø Neural Networks : and it’s application
Neural networks NN are designed inspired by the structure of the brain. NN is designed to recognize patterns. Real world data like text, image, sound are translated to numerical vectors and NN helps us to cluster and classify.
Application : In customer relationship management, to identify whether the customer is angry or happy with the service, NN Models are being used.
Ø Association Rule Mining : and it’s application
Association rule mining used in finding interesting associations and relationships among large sets of data items.
Application : Market-basket analysis attempt to analyze customer buying patterns by uncovering associations between items that customers put into their baskets.
Ø Latent variable models : and it’s application
Latent variables are hidden variables that is never observed. Latent variable model (LVM) is a probability distribution over two sets of variables x,z. x variables are observed at learning time in a dataset D and z are never observed.
Application : In the field of economics study of quality of life, business confidence, morale and happiness these are all variables which cannot be measured directly. But linking these latent variables to other observable variables like wealth, employment, environment, physical and mental health, education etc., helps in inferring the values of the latent variables.
Note : What is what
Machine Learning :- Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.
Machine Learning Algorithms :- Machine Learning Algorithms are step-by-step computational procedures for solving a problem.
Machine Learning Models :- Algorithmic models that tell you which outcome is likely to hold true for your target variable based on your training data.
