Learning by machine is a concept which is really difficult to believe. However it is true. Machines learn from the input data. Create models using algorithms and do forecasting. This is one of the latest and most popular technology of deriving intelligence from raw input data. Machine learning is one of the very important aspect of data science covered as part of distance MBA in AI and ML.
Types of Machine learning
Machine learning is a method in which computer itself learns using the past data as input. Using the input data a model is created. With this model, on the currently sought new data, predictions can be done. Machine learning is of four different types
- Supervised learning
- Un-supervised learning
- Semi-supervised learning
- Active Learning
Supervised learning
Let us understanding about machine learning with an example. Imagine we are in an ice cream business. We believe that daily temperature variations affect ice cream sales. We have a lot of past data about temperatures during the day. We want to use this data to predict what would be the daily ice cream sales.
We use supervised learning here. Supervised learning is the method where your have two variable, input and output. We employ a predictive algorithm to find out a mapping function from the input variable to the output variable.
Y = f(X)
Here X is called independent variable and Y is called dependent variable. The input data or values of X which is used to determine function f(X) is called as training data set. This training data set is fed to special algorithms known as predictive algorithms. These algorithms create a mapping function f which is called the predictive model. The very objective of this mapping function or model is for forecasting. If you’ve got new input information (X), based on the it you will be able to forecast the output (Y). Here this method has a peculiarity. From the past data as outcomes are known it is called supervised learning. This resembles an instructor observing and helping the learning process under his supervision. In this supervised learning methodology, the predictive algorithm makes iterations while doing predictions. Learning is paused when desired level of accuracy and precision is achieved.
The algorithm discussed here is called linear regression and is a type of supervised learning algorithm. Regression, random forest for classification, support vector machines are examples of popular supervised learning algorithms. Supervised learning is a key component of distance MBA in AI and ML.
Un-supervised learning
Unsupervised learning are different from supervised learning. Here there is not specific output variable to be predicted. Clustering is kind of synonym for un-supervised learning.
Clustering is used to create groups or clusters of similar or equivalent records based on various parameters. The idea is to characterize these groups so that these groups are really useful for business purposes. One of the popular usage of this clustering technique is to create market segmentation. Customers are always categorized based on various parameter like age, income, sex, transaction history and so on. Using clustering, such distinct segments can be identified. This segmentation approach has benefits. We can create a distinct marketing strategy for each segment.
Semi-supervised Learning
Semi-supervised learning techniques use both examples of data where outcome is known and where outcome is not known. It is kind of combination of supervised and un-supervised learning methods. This is useful when you have large amount of input data where outcome is known for some cases and is not known for some cases. With such mixed data, semi-supervised learning is cost effective.
Distance MBA in AI and ML would have case studies to understand machine learning. Through appropriate case studies one can know whether to apply supervised or learning or unsupervised learning based on the problem at hand. Studying these machine learning algorithms is one the key knowledge area for distance MBA in artificial intelligence and machine learning.
Active Learning
All the method discussed above can be completely automated with no manual intervention in the learning process. The algorithm, once chosen by a data scientist, would read the training data and create the model. This model would be used to predict the outcomes for new values of input variables. Active learning is a machine learning technique which lets the user play an active role in the learning process. Here the learning method can ask use or domain expert to label or fix the output for certain input values. This manual intervention is done to actively acquire knowledge from a human expert. This help optimize the quality of model and hence the prediction.
Distance MBA in machine Learning
All distance MBA programs in machine learning would cover many of these methods. Supervised learning methods are more popular. Before selecting the distance MBA program in machine learning please check if it is comprehensive enough. Does it covers popular methods like regression, classification, clustering, neural networks etc.
