Just think of all the ways people use machine learning applications every day. Whether we realize their full extent or not, machine learning examples are everywhere and have become mainstream — even to the point we take the technology for granted. And the applications of machine learning listed here are only a handful of the ways ML and similar applications of artificial intelligence have made their way into modern culture.
• Virtual personal assistants, like Siri, Alexa, and Google, which incorporate speech recognition, speech-to-text conversion, and natural language processing supported by machine learning applications.
• Self-driving cars and vehicles, which are still being “trained” to learn the rules of the road in different cultures globally using various applications of artificial intelligence.
• Online recommendation systems, such as those on Amazon and Netflix.
• Personalized playlists in Pandora and Spotify, both of which are popular machine learning examples.
• Google personalized searches and ads, as well as Google Adwords. Google News also relies on the company’s various applications of machine learning for news clustering, related stories, questionable content identification, handwriting recognition, automatic closed captioning, and machine translation.
• Product recommendations, which conveniently remind us of a product we’ve previously researched or shopped for online.
• Online customer support, seemingly on every other website these days, with chatbots as a concierge for questions, queries and answers. IBM's Watson and Kindle Store speech recognition systems, which are supported with machine learning applications, apply here.
• Facebook’s automatic face recognition and friend recommendations. Pinterest also uses machine learning-based computer vision to extract
information from images and videos, which is how the Pinterest app identifies the objects (or pins) in images and recommends similar corresponding pins.
• Uber, which uses machine learning applications to predict customer demand for drivers and their pre-location status.
• The Bing search engine. Although Bing isn’t used to the same extent as Google and Yahoo, its applications of machine learning models are what drives it.
Beyond this list, however, there are certain other applications of machine learning we should appreciate more than we do. The guess is that many people simply don’t realize the role machine learning plays in these solutions. Or maybe we just take them for granted, too, although we shouldn’t.
Machine learning examples in healthcare
Consultants at Daffodil Software pointed to these examples a couple years ago as the applications of artificial intelligence and machine learning picked up momentum for healthcare solutions. Thousands of similar solutions have been introduced since then, but these are great indicators of the machine-based influence on healthcare best practices.
Identifying treatment plans: IBM’s Watson for Oncology is a cognitive technology that recommends evidence-based cancer treatment options to physicians. As the Watson for Oncology web page says, this solution combines oncologists’ expertise in cancer care with the speed of IBM Watson… “fueled by information from relevant guidelines, best practices, and medical journals and textbooks.” With AI’s help, the solution assesses all information from a patient’s medical record and arrives at potential treatment options ranked by level of confidence based on training and supporting evidence.
Mining medical records: For patient diagnosis and treatment, collecting patient records, storing them, and tracing their lineage can be a recipe for life-threatening inaccuracies. Google’s AI research branch launched its DeepMind Health project as a solution to mine a patient’s data more accurately. DeepMind Health is now an integral part of Google Health and its innovative applications of artificial intelligence for medical record keeping.
Assisting repetitive tasks: Johnson and Johnson’s Sedasys system delivers anesthesia for standard procedures like colonoscopy and was one solution involving machine learning applications that received approval from FDA early on. More importantly, Sedasys and other systems like it have paved the path to things like virtual nursing and drug development and testing that leverage machine intelligence.
While Facebook friend recommendations and playlists in Pandora are one thing, the intersection of artificial intelligence, machine learning and healthcare to improve patient care and records management is something we should all be more aware of — and thankful for.
Applications of machine learning for fraud detection
$32 billion in 2020. That’s what experts predict online credit card fraud to soar to in the coming year alone as more payment channel options — credit/debit cards, smartphones, wallets, UPI and others — drive more transactions. Worse, more criminals have become even more adept at finding loopholes for fraudulent practices on consumers by the millions. Because most consumers aren’t aware of how fraud works and what to do to prevent it, this makes fraud detection one of the most critical and necessary applications of machine learning.
(Quick educational moment: Whenever a customer carries out a transaction, a machine learning model x-rays their profile extensively to search for suspicious patterns, typically framing problems like fraud detection as classification problems. For companies themselves, Paypal is just one financial services organization using machine learning applications to distinguish between legitimate or illegitimate transactions, primarily for protection against money laundering.)
On Black Friday, Cyber Monday, and throughout the holiday shopping season and the rest of the year, be thankful machine learning has your back each time you use a credit card online.
Email spam and malware filtering
Email clients use several different spam filtering approaches, although rules-based spam filtering fails to trace many of the spammer tricks that show up on a constant basis. Machine learning applications determine that spam filters are continuously updated with techniques such as Multi-layer Perceptron and C 4.5 Decision Tree Induction. As for malwares, more than 325,000 are detected every day, with each piece of code mirroring its previous versions at a 90% to 98% similarity. Fortunately, system security programs that incorporate a machine learning model understand the coding pattern and routinely detect new malware with a 2% to 10% variation to provide protection against them.
The next time you securely send or receive an email and see an inbox minus an influx of spammed messages, don’t take these safeguards for granted.
Google Maps, one of the more recognized machine learning examples
Yes, Google Maps gets you from point A to point B and is still the most used app for directions and traffic. Directions are easy, but detecting real-time traffic flows to avoid things like construction zones and congested highways is an added bonus. When Google Maps shows a red line on the route it outlines for you, it’s machine learning at work. Or more specifically, it’s data coming from drivers currently using the service — providing their location, average speed, the direction they’re traveling and other details — combined with historic data of the route collected over time. Google gathers these massive amounts of data, analyzes it with the help of a machine learning model, predicts the upcoming traffic pattern, and adjusts your route accordingly.
As you travel for the holidays and use Google Maps as your guide, be thankful when a planned route is marked in red. Minor detour, perhaps, but it could save you a lot of headaches.
Dynamic pricing, the hidden jewel of machine learning
In economic theory, and especially at a time when most people are just trying to make ends meet, setting a fair market price for a product or service is an age-old problem. Pricing strategies can affect the foods we buy, gas prices, drug prescriptions, a new car, a plane ticket, cab fares — virtually everything is dynamically priced. To a significant degree, applications of artificial intelligence have enabled pricing solutions to track buying trends and determine lower, more competitive costs for a product or service.
While dynamic pricing doesn’t always benefit consumers (Uber’s Geosurge machine learning model actually drives up pricing based on rider demand and urgency), it’s at least a step in the right direction toward fairer price points between shoppers and businesses. So be sure to thank any business that adopts a dynamic pricing model in the name of their customers and fair-mindedness.
In fact, be thankful the applications of machine learning and AI have made the inroads they have. They continue to make our lives better… every day.
ClearObject knows all about machine learning applications and AI solutions as both an IBM Gold Business Partner and certified Google Cloud Partner. In addition to IoT Engineering, Analytics and Connected Product Development, we’re experts at developing and implementing targeted data analytics strategies to get the most value from your data — including deploying and managing serverless cloud solutions for AI tools and end-to-end machine learning models.