Watson in a more practical view

ByAlex Lattaro in

In a previous article, we did a short introduction about neuroscience, artificial intelligence, cognitive computing, and IBM Watson. Today, we’ll cover IBM Watson in a slightly more practical way. A software that many people knows and use on their cell phones, as in Windows Phone, iOS or Android, are the virtual secretaries. These secretaries work much like IBM Watson: the more you talk to them, the more they learn about you; And gradually learn your tastes, your way of living and your way of interacting.

IBM Watson works similarly, but it’s a little bit grander. He is able to deal with words in a much more comprehensive way because he understands the context. His ability to understand what is offered in the form of words, goes far beyond the understanding of a simple virtual secretary. Secretaries need you to use some key words so she understands what you really want.
Watson is different. He can understand independently of words because he is able to deal with unstructured data. That is, the Watson allows a greater interaction between human and machine, causing them to interact in a more fluid way through natural language. Watson can learn and identify patterns and insights more closely the way a human would be able to.

Watson’s capabilities are so advanced that he is able to help doctors in the quest for the best treatments. In addition, it can also help you reshape businesses like:

  • Retail;
  • Tourism;
  • Banking services.

Let’s follow one of the examples cited above to make this article more didactic. Imagine a system that is able to keep up with the financial market. In addition, it is able to read all the reports related to the subject and all the news connected to the topic. After getting all this information, which I believe are the consumer dream of any shareholder, he has the ability to understand the scenario (here is the more complex part) and then assist executives in making more assertive decisions.

Another example, based on the previous ones, is medicine. Imagine a system capable of listing all the research related to a particular disease. Then, after this collection, he would cross all the information, understand the characteristics of the patients in Brazil, for example, and indicate to the doctor, what is the best treatment for that patient’s illness.

For such exploits to be possible, it’s necessary training this system, as with any neural artificial network (NAN). For those who do not know, NANs are computational techniques based on mathematical models based on the operation of organic neural networks, which acquire knowledge through experience. The big difference between an NAN and an organic system, currently, is in capacity. An NAN could have thousands of processing units, artificial neurons, while an organic network, the brain of a mammal, may have billions of neurons.

To understand some of the complexity behind Watson processing, you need to understand a bit what makes NAN intelligent. The intelligence of the neural network is possible thanks to the communication channels that are associated with different weights. Thus, each processing unit performs the necessary operations on top of its local data. These data are sent to the next artificial neurons, which remake the process and connect with others. Thus, in a very simple way, after a series of processing and reprocessing, it provides us with a valid solution.

However the beginning is not very reliable. As we have seen above, there is a need for training. And this training should be very hard to make your network smarter. And so does Watson. The larger the database he has, and the more training available to him, the better and more appropriate his answers will be.

On the IBM website you will find some tutorials available to create and use a web server, where you can send questions to Watson using the REST APIs and receive the answers through JADE. This is one of the ways you find to train your network.

Are you interested in the subject? Would you like to know what Watson is capable of doing in your project? Make no mistake, it can help in both small and large applications. For you to get the most out of Watson, we suggest you try to find out more about it on the IBM website.

Now, if you really want to delve into this world, we strongly advise you to study IA, RNA, and advanced algorithms.
If you would like to know more about cognitive computing, follow the IBM link.

Leave a comment! 0

read more