We collected every email and every tweet from the 2020 Democratic primary. Here's what we found...

5 Clever Examples of How Machine Learning is Used Today

Devin Pickell
Devin Pickell  |  May 30, 2019

If you used Google, Spotify, or Uber in the past week, you’ve engaged with products that utilize machine learning.

Machine learning, a subset of artificial intelligence, uses algorithms and statistical models to train machines to perform tasks and find patterns without guidance. In regards to our examples above, these tasks are things like search recommendations, song suggestions, and estimated travel times.

But there are more ways machine learning is applied today, some of which we may not even be aware of.

In this guide, we’ll shed light on some clever real-world examples of machine learning. To source our examples, we reached out to five business leaders to explain the ways they use machine learning today.

5 machine learning examples

See an example that interests you? Feel free to skip ahead:

  1. Text analysis for content creation
  2. Analytics for real estate investing
  3. Call verification for fraud recovery
  4. Price tracking for air travel
  5. Deep learning for writing Shakespeare

1. Text analysis

Stephen Jeske, Content Strategist at MarketMuse

Every content marketer out there knows just how difficult topic mapping can be. It requires topic expertise, in-depth research, and collaborating with in-house teams to ensure content is both relevant and accurate. Stephen says MarketMuse applies machine learning to make the jobs of content marketers a bit easier.

“One application of machine learning (ML) concerns textual analysis; an important part of our ML implementation at MarketMuse. The ability to analyze text allows us to create a topic model of any given topic, and score content to help content marketers create a better experience. Machine learning for textual analysis enables us to establish semantic relevance between pages on a website. This is used to offer linking suggestions with appropriate anchor text for creating content clusters.

Taking it a step further, machine learning is a building block that allows us to discover topical clusters that exist on a website, which may not be readily apparent.”

Machine learning helps validate the content marketer’s assumptions of what users are searching on the web in regards to a topic. It also exposes them to new content ideas.

Give me the G2: See what real users are saying about MarketMuse and all its machine learning capabilities.

2. Real estate

Daniela Andreevska, Marketing Director at Mashvisor

Investing in real estate can be lucrative if done right. However, the process can be time-consuming and most don’t have the resources or expertise to consider investing. Daniela says Mashvisor utilizes machine learning and data analysis to shorten the process and make it simpler.

“Mashvisor's Property Finder is one of our prominent tools which uses machine learning to predict the most appropriate property for an investor based on the criteria which they enter as well as their background. Users can like and dislike properties which the Property Finder suggests to them based on their location of choice, budget, preferred property type, and other criteria. The more the user interacts with the tool, the more accurate its predictions become.”

Being able to provide highly personalized recommendations is an example of supervised learning, in which there are both input and output values for the machine learning algorithm. The more it’s used, the more accurate the model gets.

Related Content: Read more about the differences between supervised and unsupervised learning in our beginner’s guide.

3. Call verification

Tim Prugar, VP of Operations at Next Caller

About $190 billion is lost annually due to fraud – costing merchants, customers, and banks their peace of mind. Dealing with fraud is frustrating, and so is calling each merchant to keep up with the trail of fraud. Tim says Next Caller uses machine learning to speed up the recovery process from fraud.

“We leverage machine learning to combat the problem of phone fraud – specifically people committing account takeovers at banks, insurance companies, cable companies, airlines, and hotels. Our VeriCall product creates a positive customer experience through real-time call verification. We use ML to make sure you can be authenticated quickly, passively, and effortlessly – allowing businesses to help fix your problem without spending excruciating time figuring out who you are.”

Identify verification using machine learning is just another example of how automation is improving our daily lives. This is especially useful as fraud will only get more complex.

4. Price tracking

Valerie Layman, Chief Product and Services Officer at Yapta

So, you’re going on a work trip and your boss tasked you with finding the most cost-effective flight. What’s next? Valerie says Yapta applies machine learning for more intelligent price tracking on air travel.

“Yapta helps its corporate customers save on airfare and hotel costs using machine learning algorithms for the optimization of travel supplier negotiations and policy compliance. It utilizes a combination of airfare and hotel booking data, real-time pricing data, and machine learning to identify areas of focus where companies can create or improve opportunities to save.

At a glance, the technology delivers actionable insight into supplier utilization and performance, contract rate performance, and travel policy effectiveness. The technology also aggregates anonymous pricing data across the billions of travel itineraries tracked by Yapta, creating benchmarks by spend amount, geography, industry, and supplier.”

Price tracking, optimization, and prediction are some of the more pragmatic ways machine learning is applied today. As a matter of fact, Yapta has used tracking to save businesses more than $250 million on airfare already.

Give me the G2: See what else Yapta has to offer by checking out their G2 profile.

5. Writing Shakespeare

Rosaria Silipo, Ph.D., Principal Data Scientist at KNIME

This example may be less business-focused and more fun, but KNIME was able to show off the power of deep learning with long short-term memory (LSTM) units to generate original Shakespearean texts. Rosaria says:

“You know the problem of finding the most appealing, not-copyrighted name for your new product? The problem that requires a number of brainstorming meetings by the most creative minds in the company? Well, a deep learning neural network with a layer of LSTM units can be trained on a list of specific names – let’s say names with a common theme such as mountain names – and produce a list of mountain-sounding, not copyright-protected names to be used as candidates for the new product name.

A similar network can also be used to generate free texts, such as Shakespeare-like texts or rap songs. These drafts can be used as the basis for the final text or song.”

It’s worth noting that generating original texts with deep learning is quite difficult, and is even more difficult when going off of complex sentence structures and Shakespearean English. Regardless, this is still a fun example from KNIME.

The deep learning network was trained with Shakespeare’s plays “Othello,” “King Lear,” and “Much Ado About Nothing.” Click below to read the script:

Free Resource: Shakespeare Text Generated by Deep Learning Download Now →

The script seemed to start off hot but simmered near the end.

It’s worth noting that generating original texts with deep learning is quite difficult, and is even more difficult when going off of complex sentence structures and Shakespearean English. Regardless, this is still a fun example from KNIME.

Give me the G2: Read some real reviews of KNIME’s open-source data analytics platform, and how users are leveraging it today.

What’s next for machine learning?

From earlier detections of fraud to improving medical diagnoses, machine learning is behind many major technological breakthroughs today, but what’s next?

We asked five experts to give us their opinions of what the future of machine learning looks like. Read how quantum computing, search engines, and no-code environments will influence the future.

Devin Pickell
Author

Devin Pickell

Devin is a Content Marketing Specialist at G2 Crowd writing about data, analytics, and digital marketing. Prior to G2, he helped scale early-stage startups out of Chicago's booming tech scene. Outside of work, he enjoys watching his beloved Cubs, playing baseball, and gaming. (he/him/his)