5 Examples How You Are Using Machine Learning Now

5 Examples How You Are Using Machine Learning Now
Share this:

1. Using Google Maps 

Using Google Maps

Although most people use Google Maps, not many question how the tool gets the job done.

How does Google come up with maps that show every alley, avenue, and street without compromising on accuracy?

The answer to this question is artificial intelligence (AI), and most importantly machine learning.

Google Maps is powered by accurate real-time details drawn from thousands of images of locations and street signs that are analyzed by cutting-edge Optical Character Recognition algorithms.

Since the algorithms have to learn how to ignore irrelevant information, the task involved is challenging.

However, Google trained its algorithms using the sophisticated French Street Name Signs database.

After the machine learning algorithms were able to recognize such complex images correctly, they became more accurate with time.

READ MORE: 30 Artificial Intelligence Examples in 10 Industries

2. Posting Content on Facebook

Using Google Maps

Facebook utilizes machine learning in ensuring that you can see the most important news posted by your friends.

Most of us have numerous Facebook friends, but not all are evenly close.

Sociologists have determined that it is impossible to have over 150 close friends, also known as the Dunbar number.

Machine learning aids Facebook in determining which one of your Facebook connections is actually a close acquaintance whose life you need to hear about among other things.

Facebook sifts through all the possible posts you could view, and assigns every one of them a given score.

This particular score is determined by various factors such as your previous “comment” and “like” activity; whether the post has experienced a tremendous time-consuming activity or has many likes or comments from many people;

whether the Facebook post comes from a friend of a friend or among your friends; whether the post entails an interaction between two individuals or a corporate page and a person.

The scores drawn from all your possible posts are ranked, and such ranking ends up becoming your Newsfeed.

3. Ordering an Uber or Lyft

Ordering an Uber or Lyft

Whether you prefer Uber or Lyft, both services leverage artificial intelligence (AI) in powering their decisions.

Calculation for your estimated time of arrival (ETA), the driver’s ETA and driver assignments is done by algorithms that are regularly refined and tested in real-time through machine learning as well as large volumes of data collected from customers and drivers.

Rideshare companies are currently leveraging the power of machine learning (ML) to combat the feared ‘surge price’.

Time-limited price increments or surge-pricing now compensates for the moments where there are insufficient cars operating to serve all passengers who request for rides.

4. Using Priority Tags and Spam Filters to Keep You Organized

Using Priority Tags and Spam Filters to Keep You Organized

Your inbox is one of the most unlikely places you would expect machine learning to operate.

However, it is vital to note that AI technologies are behind one of the most vital email tools, spam filter.

Basic rules-based filters are leveraged for the spam filter.

Consider phrases and words such as “Nigerian prince,” “pharmacy” or “you’ve won the lottery.”

Even though you may be good friends with a Nigerian prince, if a given message appears suspicious, or comes from an unrecognized sender, then it will most likely be flagged and forwarded to spam.

The filter is trained through the content found in the emails, recognizing signals by collecting inferences from word relationships.

In turn, the filter counter-maneuvers spammers who may be trying to outwit it using updated messages.

Aside from this general function, the filter also used machine learning to train itself what you individually consider spam.

It does this by feeding itself with data, particularly on what you mark as spam mail or delete.

This data works together with the data drawn from the whole user base.

With such approaches operating together, various reports place the spam filters from Gmail at a success rate of 99.9%.

What’s more, researchers tested the efficacy of Priority Inbox, specifically on Google employees and discovered that workers with Priority Inbox “spent 6% less time reading email overall, and 13% less time reading unimportant email.”

The same technique is being utilized in tagging emails for pre-determined promotion, social and primary inboxes.

5. Keeping Credit Fraud at Bay

Keeping Credit Fraud at BayA database made up of consumer complaints drawn from the Federal Trade Commission reported that 1.3 million or 42% of all the 3 million complaints made in 2016 were related to fraud, with about $744 million worth of losses.

Out of all the cases, 55% of them were identity theft or fraud related – that exceeds one in every 200 Americans each year.

Note that 7% of American households don’t have bank accounts, and only 77 percent of all American citizens are aged beyond 18 years.

Since this is a huge problem, how do financial entities determine if a given transaction is fraudulent?

Bank of America has about 58 million clients, and like many banking institutions, its everyday transaction volumes are incredibly high for manual review.

To go through this massive volume of data and separate illegal purchases from normal ones, banking institutions utilize machine learning.

For instance, FICO, a renowned credit ratings producer, utilizes neural networks to forecast fraudulent transactions.

Intended for simulating the neural networks in human brains, such systems analyze labeled data examples, learning how to spot unlabeled inputs and creating their own characteristic markers.

Factors considered consist of the client’s recent frequency of transaction size, transactions and the type of retailer involved. (It is ideal for the banking institution too — researchers from the Massachusetts Institute of Technology discovered in 2010 that machine learning can be applied to customer transactions, could minimize bank losses, especially from delinquency existing between 6% and 25%).

These examples form the tip of the iceberg.

In fact, artificial intelligence (AI) is being employed in numerous fields.

The technology is used in the fashion industry, where firms such as Stitch Fix leverage stylist expertise, customer feedback and machine learning in delivering clothes to your house through subscription, and garner more than $750 million in yearly sales.

Artificial intelligence is employed in sports coaching, especially to thousands of historical plays and strategies in a bid to make recommendations and predictions based on this user inputs and training data.

Share this:

Leave a Reply

avatar
  Subscribe  
Notify of