In machine learning, the problem of unsupervised learning is that of trying to find hidden structure in unlabeled data. Since the examples given to the learner are unlabeled, there is no error or reward signal to evaluate a potential solution. This distinguishes unsupervised learning from supervised learning and reinforcement learning. 

When learning a model in supervised learning, you have examples in the data set that include the target concept (or as Ka-Chun Wong describes, "y"). The algorithms build the model based using the target concept. For example, Decision Trees. In unsupervised learning the learned concept is not part of the data set. For example, regression models, or K-Means clustering.

In machine learning, the problem of unsupervised learning is that of trying to find hidden structure in unlabeled data. Since the examples given to the learner are unlabeled, there is no error or reward signal to evaluate a potential solution. This distinguishes unsupervised learning from supervised learning and reinforcement learning. 

hello

In supervised ML models, the learning of the algorithms is under the control of the researcher, in the sense that, the learning of the method is under finite lebels.

But in the other, things are not under your control, in the sense that you can not predict what the system is going to learn.

Mathematically, The supervised learning is based on a known [finite] sets of datasets and the function may be known or may not be known.

Lets say, you want to use a algorithm to differentiate 'good' and "bad " mangoes from a slot.

But in unsupervised, the data may not be available at hand. The mathematical function is not known most of the time. For example, the learning of a robot arm to catch a ball coming from a distance.

Hope the above answer will convince you to some extent.

Thank you.

Best,

Indrajit Mandal

Thank you everyone for making the explanation so easy for me. Im novice to Machine Learning thing and trying to grasp things in a way I my brain indices can hold.