Machine learning, probably one of the hottest words in the current decade. No matter what the product is, everyone wants to have machine learning capabilities in their product, because we know it’s a real head-turner, so what really is Machine Learning in layman terms?
Imagine a classroom where teacher has given students a few maths problems to solve and teacher knows answers of them all, teacher is clearly supervising how each student is solving, whether it’s a correct solution or not, that’s what a supervised learning is, where the teacher is a programmer, the number of problems is our dataset and the students, our machine learning algorithm. The programmer works on the labeled dataset and trains the machine learning algorithm until expected accuracy is met. No rocket science.
Secondly, we have unsupervised learning, the difference here is the dataset we get is not labeled or in reference with the previous example, teacher doesn’t know the answers, now what to do? Imagine another scenario where we have a dataset of animal photos and we have to train our model to find out whether the photo has a bird or a dog, except we don’t know how these animals or birds look like. In such scenarios, grouping of data is performed and comparison is made by the model to predict the output. If the information such as if it has feathers, it’s a bird and if it’s fluffy and has a curly tail, it’s a dog is used, model can distinguish animals successfully. There you go.
Lastly, there’s reinforcement learning, in this method the machine learning algorithm learns to react on its own to the environment. In such case, the environment has a start state and end state for an agent. Imagine a scenario where a child is learning to walk, the child tries to make a balance on its feet and move towards mother where she is luring the kid with its favorite toy, the closer the child gets to the mother, chances are child gets to play with the favorite toy. Now from machine learning perspective, the child is an agent where it is manipulating the environment by taking steps, the agent or the child starts from start state and gets rewarded only when moved towards the end state where mother is standing, it’s not rewarded otherwise. So, this is how reward based reinforcement learning works.
We have merely scratched the gigantic world of machine learning and its applications in real world, hope you understood and enjoyed this blog. Toodles.
Get started on your intelligent automation journey with our experts!