Machine Learning
“The question of whether a computer can think is no more interesting than the question of whether a submarine can swim.”
Seems interesting, right? A computer thinking like a human.
Starting with what is machine learning….
Machine Learning is the science of getting computers to act without being absolutely programmed. People generally got confused about the differences in AI and ML, so let’s clear that first. ML is a technique of Artificial Intelligence(AI) which consists of various algorithms that make computers, or computer-controlled robots think intelligently, just like an intellectual human being thinks. Various applications of machine learning include self-driving cars, practical speech recognition, effective web search, and image recognition and classification. Machine learning is so prevalent today that you probably use it dozens of times a day without knowing it.

Getting into the technicalities, Machine Learning can be broadly classified into two categories :- Supervised Learning and Unsupervised Learning. A brief explanation of some of the machine learning algorithms follows:
“The goal is to turn data into information, and information into insight.”
Regression:
Regression is a mathematical procedure used to find a curve that closely fits a given series of data. The goal of the analysis is to minimize the difference between the data point and the value predicted by the function. There are various different techniques, the most common being by the method of least squares.
For instance, say you want to find the price of a house given its size. You already have some data points i.e. the price of a house for various sizes. Then the price can be predicted by finding a curve which fits the available data and plotting the required data point on that curve.

Logistic Regression:
Logistic regression is a regression model where the dependent variable is categorical. A binary dependent variable can take only two values, “0” and “1”, which represent outcomes such as pass/fail, win/lose, alive/dead or healthy/sick whereas multinomial logistic regression can also be used to classify objects according to more than two categories.
Suppose, you want to find out whether an e-mail is spam or not then you can use the concept of binary dependent variable. And if you want to classify given flowers into say, three categories then multinomial regression will be used.
Support Vector Machine:
It is a discriminative classifier formally defined by a separating hyperplane. In other words, given labeled training data (supervised learning), the algorithm outputs an optimal hyperplane which categorizes new examples. Right hyperplane is chosen by using the large-margin classifier.
Feeling like rotating around technical terms? Take a look at the whole process here.
Neural Networks:
It is a beautiful biologically-inspired programming criterion which enables a computer to learn from observational data. Neural networks are a subset of algorithms built around a model of artificial neurons spread across three or more layers. Nowadays, problems like image recognition, speech recognition, and natural language processing are solved using neural networks and deep learning.
Clustering algorithms:
Clustering algorithms are used in search engines that try to group similar objects in one cluster and the dissimilar objects far from each other. The goal is to cluster unlabeled data(unsupervised learning) into number of clusters and then providing the result for the searched data according to the nearest similar object which are grouped around the data to be searched. The most commonly used type of clustering algorithm is K-means.
As an illustration, take a collection of 1,000,000 different genes, and find a way to automatically group these genes into groups that are somehow similar or related by different variables, such as lifespan, location, roles, and so on.
Google uses clustering algorithms to classify different contents as News by parsing though the matter and examining the keywords.
Recommendation engines:
The algorithm represents each customer as a vector of all items on sale. Each entry in the vector is positive if the customer bought or rated the item, negative if the customer disliked the item, or empty if the customer has not made his or her opinion known. Most of the entries are empty for most of the customers. The algorithm then creates its recommendations by calculating a similarity value between the current customer and everyone else. Amazon/Flipkart/Netflix use collaborative filtering for recommendation.

Naïve Bayesian:
The best known use of Naïve Bayesian classification is spam filtering. It is a probabilistic classifier based on Bayes’ theorem.
For example, Emails use Bayes’ formula for calculating the probability of an email to be classified as a spam, given already existing spams. This can be done by calculating probabilities associated with each word of the text to be classified as a spam.
A brillliant application of Machine Learning algorithms by Microsoft is in the following video:-
For more details of the topic, you can refer to these lectures.
Looking forward to see you at ArIES — Sakshi.
