Today we welcome guest contributor Tanmay Bakshi, an inspiring 13-year-old app developer, teacher and YouTuber. We met Tanmay while he was giving a keynote speech to our tech associates (yes, he gives keynote speeches around the world). Please enjoy Tanmay’s article on 5 machine learning services powered by Watson.
Computers have never been good at understanding non-mathematical language, but a branch of artificial intelligence aims to change that. Machine learning (ML), a subset of artificial intelligence, aims to enable computers to understand natural language and allow them to reason for themselves. Leading this initiative are services like IBM’s Watson, an open platform accessible to developers around the world.
In today’s world, everyone is using machine learning technologies on a daily basis, whether they realize it or not. From the suggestions list on Netflix, to self-driving cars, artificial intelligence is all around us. Despite their ubiquity, machine learning algorithms are hard to implement, causing many developers to refrain from using them in their applications. This is where Watson comes in handy, by making cognitive computing algorithms simpler to implement.
Watson exists as a set of services (APIs) that developers can bake into their applications in a simple to use, yet powerful way. Below, I’ve highlighted five Watson services I find particularly interesting and useful.
As the name suggests, Visual Recognition is a service geared towards image recognition. Let’s say you want to create an application that can recognize a picture of a dog and identify the breed. You would input around 300 images of each breed, and around 100 images each of cats, lions, and tigers. Watson will then train itself to understand the differences between dog breeds, as well as animals that are not dogs. Once it finds those patterns, Watson will be able to tell, within a percentage of confidence, whether a picture is of a dog, and if it is, which breed it is.
Visual Recognition can be useful in a number of areas, ranging from brand recognition to security. I’ve used it to build an application that uses facial recognition to prevent false identification.
I’ve also put together a tutorial on how to build the dog breed classifier mentioned above.
Natural Language Classifier
The Natural Language Classifier (NLC) is one of my favorite services, mainly because it’s simple to train, handles extremely large datasets smoothly, and provides solutions for a wide variety of use cases. The NLC’s goal is to classify Natural Language into classes. That classified data can then be used by other applications. For example, my NLQA (Natural Language Question Answering) application AskTanmay uses NLC to determine what type of question a user has asked.
For example, if you ask, “Who is the CEO of Walmart?” you’re looking for a person. If you ask, “Where is the Taj Mahal?” you’re looking for a location. If you ask, “Which worldwide retail chain makes the most profit?” you’re looking for an organization.
The NLC is the service that obtains this metadata. Once it completes its work, Watson can then use it to train itself to better understand the patterns between question and question type.
Retrieve and Rank
This service is quite different from the ones I’ve already covered. It’s quite difficult to train, in the sense that it needs a lot of data before it can find meaningful patterns. The training process is also extremely manual. But the end result makes it all worthwhile.
Retrieve and Rank (R&R) allows you to retrieve data using a query against a SOLR cluster. Watson will then rank the results in the order it believes to be most relevant to your query. This allows you to input a corpus of knowledge, which can be in PDF, MS Word, or HTML form. The R&R service will then split up the data into “Answer Units,” which are essentially chunks of text. You can then input sample questions and link the questions and Answer Units. Once that’s done, you can ask the R&R service a brand new question and it will output relevant answer units, along with a measure of how relevant Watson thinks they are.
This was the very first Watson service I played around with. You can learn more about my work with R&R here.
What makes Personality Insights different from Watson’s other modules is that it’s one of the few services that requires no training. It can take text (100 words minimum) and output “attributes” of the personality of the person who wrote it. It does this under three main categories, “Personality”, “Consumer Needs” and “Values”. There are a lot of these attributes. Here’s a screenshot of a few:
Personality Insights can be used for a variety of use cases, such as detecting and charting an individual or group’s behavioral changes over time.
Tradeoff Analytics is a service that allows you to input data along with parameters for how it should be weighed to get a relevant output. For example, you could input a vehicle’s MPG, average rating, number of seats, cylinders and price, and then specify which parameters matter most. Watson will take all that into account to find the ideal balance. Finally, Tradeoff Analytics will suggest which vehicle it believes you should buy.
The author is a guest contributor and is not affiliated with Walmart Technology. The views and opinions expressed in this article are solely those of the author and do not reflect any official policy or position of Walmart Technology. Walmart Technology also does not endorse or recommend any commercial products, processes, or services described in this article. Walmart Technology shall not accept any liability regarding the views and opinions expressed by the author.