During the last five years, Artificial Intelligence (AI) has exploded because of availability of infinite storage, big data and GPUs (Graphics processing unit). GPUs have enabled the parallel processing faster, cheaper and more powerful. These developments have lead to an Artificial Intelligence boom that has unleashed machine learning applications used by hundreds of millions of people every day.
Machine learning, is a subset of artificial intelligence, which provides systems the ability to learn by itself from experience without being explicitly programmed. Websites like Amazon, Facebook, Google, etc. collect huge amounts of data and they know what the user’s frequent search terms, preferences, likes, dislikes and other personalized information. Just for an example, when you search for a product in Amazon, you can see the advertisements of that same product in Gmail, Yahoo Mail, Google, Facebook and many other websites that you visit. This is all done by the processing of data, and this is the power of machine learning. The machine knows that if it shows the advertisement for a product that you have searched before, there is a high percentage of chance that you may click that advertisement and purchase the product. Machine learning is not a new concept. For instance, spam option in Gmail / Yahoo accounts was one of the early implementations of machine learning.
Why Machine Learning?
Machine learning (ML) can do far more complex and complicated tasks. ML solves basically four types of problems. They are:
- Classification: Classifying data into different classes. For example, if we give the image of an Apple or an Orange, the machine will tell whether it is an Apple or an Orange.
- Regression: Predicting future values based on previous data. For example, finding a student’s marks based on the number of hours studied.
- Clustering: Grouping the given data into different clusters or groups.
- Handling Associations: Finding the associations between different parameters in the available data.
Types of Machine Learning Algorithms
There are several types of machine learning algorithms used for training the machine based on the available data. The most commonly used algorithms are Linear Regression, K-Nearest Neighbors, Support Vector Machine, Decision Tree, Random Forest, etc. As per the need, an algorithm that is suitable is selected to train the machine. Basically, we can classify machine learning algorithms into three types.
- Supervised learning
- Unsupervised learning
- Reinforcement learning
In supervised learning, there is a dependent variable that needs to be calculated using independent variables. This is a repetitive training process, until the desired accuracy is achieved. Examples are applications of linear regression and K-nearest neighbors such as calculating house price index or HPI which is the numerical target. Age and loan are two numerical variables and are used as predictors to calculate HPI.
In unsupervised learning, we don’t have a target or estimation to be predicted. It is used for clustering segments into various groups. There is no definite output. Instead, the focus is on classification. E.g. applications of K-means algorithm in color quantization which involves diminishing the abundance of colours in an Image or picture to just 3 basic colors B,G, and R values while maintaining the overall appearance feature.
The reinforcement algorithm takes specific decisions based on situations. Usually, we expose the machine to an unknown environment, and it learns from there. This continuous approach to learning, which is known as trial and error, is just like the learning of humans. The machine learns from experience, hopefully, the best knowledge, and applies them in business logic. E.g. applications of Markov decision process such as automated trading strategy. It can be used to experiment with the buy or sell options.
Applications of Machine Learning
Machine learning is growing fast, and it is getting widely popular in almost all fields nowadays. Popular Applications of machine learning are:
- Suggestions from Amazon and other online shopping websites.
- Spam detection of Gmail/Yahoo Mail
- Quick reply suggestions for emails in Gmail
- Facebook detecting photos of your friends and automatically putting their corresponding names there.
- Google Maps suggesting which route you need to take, and based on that route how much time it will take.
- Customer Fraud Detection by Bank.
- Picture Quality Improvement.
- Voice recognition apps like Siri (from Apple), Cortana (from Microsoft) and Alexa (from Amazon).
Machine learning is the ability of a computer system to observe, learn and gain experience from lots of data, and use this experience to predict future results. After decades of hopes and disappointment, Artificial Intelligence is back and could be set to drive major changes in the global economy. Major tech giants like Amazon, Google, Microsoft and Apple are making huge investments on AI technology development and deploying it across their businesses. Now computers or machines can see and hear, and even speak. It is likely that there will be more impact from Vision/Voice based applications rather than data analytics/predictions. E.g. Self-driving cars, medical diagnosis, video analytics, autonomous machines/robotics and voiced based interactions.
Today’s business environment demands the leaders to be digital-savvy. Online MBA from IFHE integrates various machine tools and cloud environment. Students thus chance upon technology interface even as they learn the subject. To know more, check out @ https://online.ifheindia.org/
Suppose your cab aggregator gives you an option to choose between chauffeur driving and a self-driving (driverless) car for your travel. Which one would you choose? Why?
- ISHA SALIAN, August 2, 2018, Super Vize Me: What’s the Difference between Supervised, Unsupervised, Semi-Supervised and Reinforcement Learning?
- Iqbal H. Sarker, 22 March 2021, Machine Learning: Algorithms, Real-World Applications and Research Directions.
- Jason Brownlee August 12, 2019, A Tour of Machine Learning Algorithms.