Machine Learning 101 – Check the examples for yourself

AI

  • Machine Learning and Human Learning

Machine Learning is a concept which defines how a machine learns about something. For a human-being to learn something, humans have senses which produce perception. Human beings as they grow and learn is almost a good inspiration for machine learning. Using senses, humans collect data, store data in brain. Using some awesome ingenious algorithms inside the brain, humans analyze data, classify the data, make self-recommendations of the data, make predictions about the data. The best thing about human beings is entire life is an input data, all experiences, all voices, all images, videos, any experience produced by sensory perception is an input data which the human brain learns upon. Taking machine learning as our focus here, given a particular scenario and data, the machine can do wonders.

  • Machine Learning Life Cycle

Let’s take a look at the following machine learning lifecycle, as illustrated in Machine Learning for java:

  1. Pick a problem
  2. Collect Data
  3. Model
  4. Train
  5. Recommend/Classify/Predict
  6. Re-train

For example, take the case of e-commerce platform.

Consider the problem-scenario of customers browsing an ecommerce website, say like amazon.com. Our objective of our machine learning solution is to recommend products to customers. (This is one way of keeping the customers within the website, part of gamification, encouraging them to buy more products as well).

How shall we do it? We have defined our problem. Now we need to collect data. Based on the user-base, collect data regarding how users uses a website, thats the user-engagement data at the website. This becomes the data for our problem. For example, we will need to answer questions like: “Suppose a customer is searching for a book on Job Interview, what are the things which interests him, which he tries to click or view?” Observations regarding how user engages the website can be collected and over period of time, create a model of user engagement in an e-commerce website. The system would tune the model by training based on various user input at the website.

Once the training is in process, based on collected data and model, the system can go to the next step of classifying/recommending/predicting. For example, for a user interested in beauty products, the website could recommend more products related to beauty to that particular user, since that parameter is very relevant to that user, as the user is interested in it. The system could make some predictions about related items, such as : based on location of the user, determine a beauty salon, predict that user might be interested in that OR predict that user might be interested in related beauty items and display them. As a part of this, classification has been performed by the system, as to which category of interest the user is looking into.

  • See for yourself – machine learning live in action in different scenarios

See for yourself in any e-commerce website:-

Once you search for a particular item, the system would start recommending related items. And based on history of other customer usage, it also recommends a combination of items to be purchased.

The next step is with ads- related ads are fetched and shown to the user, recommending user to buy related product.

See for yourself in a social network website, for example: Facebook:-

Once you send a friend request to some person, the system based on recommendation system, suggests some more friends, since they are mutual. If you join a group, the system predicts that you might become a friend to a person in that group, it shows suggestions. If you publish a post, the words of the post is analyzed and based on machine learning, make a prediction and recommend by showing a suggested post or suggested ad in the vicinity of browsing area.

See for yourself in twitter:-

Once you follow a particular user, it will start popping small tabs of related person/topic to follow. You might start wondering at how the system predicted what you want.

See for yourself in wordpress:-

Suppose you start following blogs related to technology, wordpress system will start recommending other blogs of similar interest.

See for yourself in Quora:-

Quora is explicit in its suggestions, like it shows up a topic and says, because you follow that related topic. It suggests a friend and says, two of your friends are already friends to this new person.

See for yourself in Google:-

Once we have searched for some phrase, Google tries to predict what we wanted to search for and what related things are there to search for and shows up all these predictions for the user to select.

So, we can see machine learning in action than any time before, in many fields. Since most of the websites engaging users collect data, train the system, and retain the users as much as possible by intelligently learning, recommending, classifying, predicting. Machine Learning aids gamification of a website which is answering a question as to how a user can be engaged with the website. It can also be put to use for many other areas.

A final thought: Human learning is beautiful and machine learning is aspiring to reach that level. But nowadays, humans are depending on these same machines for their learning and with all the recommendations, predictions and classification made by the system, humans are learning based on these. In a way, learning of humans is being controlled by these machines, based on the exposure to the machines.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s