Let's Get Into Machine Learning

    

Hello and welcome back to the CosmoTechno - Dive Deep Into The Cosmos Of Technology! So get ready for diving deep into the cosmos of Machine Learning today. You might have been heard about this buzzword everywhere, but not really sure about what is it. Hold On! you don't have to worry about this as this blog will help you to get a valuable insight of Machine Learning.

Before starting with any discussion let's have a look at the technical term of Machine Learning,

Arthur Samuel says, "Machine Learning is field of study that gives computers the ability to learn without being explicitly programmed."

If I rephrase the above definition it will be something like, feeding lots of experience to the machine so that it will come up with a strategy to solve a particular problem. Here experience is in the form of data which can be a text, images, videos, etc. On the basis of this data we provide there are four types of Machine Learning:

                                                                                

1. Supervised Learning

                Supervised learning is a type of machine learning where data provided is labeled i.e. there is right answer provided for each record. Dataset of supervised learning consist of independent input and dependent output, for example, Housing Price Prediction project will have size of house, number of room, altitude, height, etc. as inputs which are independent and price will be the dependent output.

Supervised Learning further have two types of learning algorithms namely:

- Linear Regression Algorithm

- Classification Algorithm

2. Unsupervised Learning

                                                        

                Type of data provided to unsupervised learning is a unlabeled data, where correct answer is not given for any record. The dataset of unsupervised learning have input but no corresponding output. Have a look at one example of fruit classification, in this example we have given a data of all the fruits is fed to the unsupervised learning algorithm which will create different clusters according to the data, now if we feed one more record of pre-existing fruit it will automatically put that into a particular cluster.

Clustering algorithm is one of the most important unsupervised learning algorithm, the example mentioned above is also an example of Clustering algorithm.

3. Semi-Supervised Learning

                This type of machine learning is self introductory, it is combination of both supervised as well as unsupervised learning algorithm. Unsupervised learning serve as a major part of this algorithm whereas, supervised learning serve as a minor part. Google photo hosting service is one of the good example of semi-supervised learning.

4. Reinforcement Learning

                Reinforcement learning is another type of machine learning  where none of the data is fed to the learning algorithm. In this type of learning, agent is allowed to learn in environment by its own, where it gets award when it performs a desire work otherwise gets penalized. Reinforcement learning is very close to the learning methodology used by human being.

These were different types of Machine Learning, now let's move further towards the application of Machine Learning

Applications Of Machine Learning:

1. Recommender system of Netflix and Facebook

2. Breast Cancer Prediction

3. Automatic Tagging System of Facebook.

4. Gmail Spam Filtering.

5. Self Driving Cars.

............ and many more.

Along with applications it is also important to understand the challenges faced by any technology:

1. Underfitting:

                Underfitting is one of the challenge faced by Machine Learning, where algorithm is very simple as compared to the reality and hence it leads to the failing of algorithm.

2. Overfitting:

                Overfitting is another challenged faced by Machine Learning which computes a very critical algorithm which best fit to the data fed to the algorithm but fails to generalize.

Above two are the major challenges which is faced by machine learning algorithms but our technology specialist have invented a good solutions to avoid these kind of problems.

So, I hope you have been gotten a better intuition about the Machine Learning, now from next time whenever you listen this streaming buzzword, you won't be having any ambiguity but you'll think in a sophistication manner. If you haven't yet subscribed to our newsletter, please go ahead and do subscribe, to get daily updates about the technology!

Thank You!