Machine learning (ML) is a subset of artificial intelligence (AI) that provides software the ability to automatically learn and improve from experience without being explicitly programmed.
There are many types of machine learning approaches that can be used independently to augment AI development. The difference between AI and ML is AI is designed to simulate a human behaviour like making the decisions and actions based on input data. We can think of AI as a general idea, whereas machine learning is a subset of AI and further down we can see deep learning which in itself is a subset of Machine learning.
Machine Learning is basically defined as algorithms that process data, learn from that data, and then make new decisions based on the learning process.
Machine learning is extensively used in recommendation systems for example, in the case of an on-demand video streaming service the algorithm has to make a decision about which new videos or movies to recommend to a user. Machine learning algorithms associate the listener’s preferences with other listeners who have a similar musical taste. This kind of algorithm is known as collaborative filtering. Sometimes content-based filtering is also used which makes a decision on the user’s past usage, though for this to work properly lots of user data is needed, so usually a mix of algorithms are used.
Similarly, Machine learning fuels all sorts of automated tasks from predicting stock prices to hunting down malware. There are wide applications of Machine learning algorithms.
Suppose there is an input X and output Y. When we write the code for machine learning, we do not hard code it. Lot of variables and functions are written in such a way that they can always change according to the loss function (for supervised learning).
At the core of every machine learning algorithm, all the algorithm is trying to do is find a relationship from the input X to the output Y. How exactly it finds this function varies by the algorithm.
But almost all algorithms require a lot of computations and a lot of revisions and usually a good amount of data.
So, when the algorithm is provided with lot of inputs and corresponding outputs, it tries to put together a function which best approximates the relationship between the inputs and the given outputs. When given a new input, it uses the function which it learnt, to calculate the output from the input.
A fantastic use case for machine learning is to track and predict weather. Satellite data can be used to track, monitor and predict the severity of hurricanes and other natural disasters
12 Hay Hill, London, W1J
26 Broadway, New York, NY
Bangalore, Mumbai & Trivandrum