ML, DL, AI — Are they the same?
Artificial intelligence (AI) has been the talk of the town for quite some time now. But it has certain subsets like ML (Machine learning) and DL (Deep learning) that aren’t known to the layman. In case you have been living under a rock, AI technology aims at making machines think like humans. Machine learning, on the other hand, is a subset of AI that is extensively data-driven and dependent on learning algorithms. Deep learning(DL) is a further kind of machine learning which comprises techniques that rely on learning algorithms, mapping, networking etc. and functions like a human brain in trying to distinguish objects, data etc. fed to it.
AI is also further classified in terms of how ‘intelligent’ it is. Artificial Narrow Intelligence is the least intelligent one and functions on strict instruction based datasets (eg. Amazon’s Alexa, Tesla autopilot, etc.). Artificial General Intelligence is the next in line which is at par with human intelligence and still a work in progress (eg. Sophia the robot, Honda’s ASIMO, etc.). Artificial superintelligence is hypothetical but it refers to AI that is smarter than humans like the ones seen in Sci-Fi movies.
Similarly, ML has its own classification based on complexity and its algorithms thrive mainly on smaller datasets. Supervised machine learning is when there are predefined datasets where certain inputs result in certain outputs only. Unsupervised machine learning is when the algorithms are left to figure out the output on their own. Reinforcement learning is a different approach that relies heavily on a hit or miss mechanism.
A very simple example of deep learning could be personality diagnosis tests where you tick certain options to given questions and eventually end up with one specified persona based on your answers. DL algorithms unlike ML algorithms work great for larger datasets. Depending on task requirements, dataset sizes etc. we can choose whether to use ML, DL or AI. Deep learning has great potential in the future as we come across many new techniques and networks that can be applied in it. But it all ties into AI and we are definitely making strides in that direction in terms of making artificial general intelligence much more realistic and enhanced by the day.
About the Author: Arryan Singh is a second year Computer Engineering student at RAIT.