Lazar Stojković, a machine learning enthusiast, has chosen NIRI as his guide and mentor in this exciting field of IT. As a Master student of information technology at the Faculty of Sciences and Mathematics in Niš, Lazar wanted to learn more about the mechanisms behind AI. He pointed out that he was interested in machine learning because it represents a synthesis of two of his loves – mathematics and information science, and because it fascinates him how much influence machine learning has on the state-of-the-art technologies.

Lazar, could you tell me something about your research here at NIRI? What are you currently working on?

Lazar: My focus is on IBM’s services used for machine learning and app development. IBM’s Bluemix is a service-based platform which is meant for the developers’ use, while the software itself, as the final product, is meant for the end user. As far as this platform is concerned, the developers are a part of an active environment and have access to a number of other additional services (API’s) which they can use to create their own apps.

What can IBM offer through Bluemix?

Lazar: IBM can provide the infrastructure, that is, the physical systems which have their own resources such as RAM, CPU time, etc. Their goal is to create a web-based, online environment for development with the help of different services. It’s great because it allows data scientists and machine learning enthusiasts to work together on projects in the most efficient way possible.

“The main feature of the Watson Data platform is going to be Cognitive Automatization of Machine Learning.”

In your presentation, you mention Data Science Experience and the Watson platform. Could you tell us something more about it?

Lazar: Well, Data Science Experience represents a sort of a project-working environment where data scientists can work and develop different software. Watson platform is in closed beta at the moment, while its main feature is going to be Cognitive Automatization of Machine Learning (CAML). IBM is the creator of both Data Science Experience and Watson data platform – all work on these platforms is done online, and all the developers’ accounts are connected to Bluemix which contains the necessary services.

What are the Watson services that IBM offers through Bluemix?

Lazar: The Watson services are very interesting. These services offer a number of different interactive API’s which developers can use. For example, the Alchemy API is a basic machine learning model which is used for a detailed text processing (finding different entities, keywords, names, etc). Personality Insight collects a number of different texts or tweets, and is able to determine the creators’ personality based on them. All these apps are made by developers and hosted by Bluemix on a Cloud, while the end users can access them with one click.

IMG_1016

What made you focus on IBM Watson Machine Learning service specifically?

Lazar: Well, the purpose of this service is to automatize the process of machine learning or app development which is based on machine learning. This is a long process that demands a lot of work. The first step is data ingestion and data cleansing, so the data which will be used to train a model must be chosen. IBM Watson along with SPSS can manipulate data and allow data analysts to operate them in any way they see fit. After the process of data ingestion is over, the developer can choose a machine learning model which will be trained based on these data. This allows the model to predict the output information for new data. After the training process, the model we get will be able to predict future results for new, unknown data. For example, we a table with a million people and the table shows the levels of Na and Cl in their blood. Our goal is to see which medication from, let’s say, three choices fits a certain patient based on these pieces of information. After we train the model, it will be able to process a brand new, unknown entry and find the right medicine for it. This is the essence of the Watson-SPSS model.

“The purpose of the IBM Watson Machine Learning service is to automatize the process of machine learning or app development which is based on machine learning.”

Wow, that’s quite impressive!

Lazar: Definitely! Right now, Cognitive Automatization of Machine Learning is being developed, and it makes the whole process of machine learning bigger and more complex. It processes data in more than one way and keeps doing it until it finds the model which is satisfactory. So, all of this will be web-based, and implemented within the Watson data platform.

Good luck on your research, it sounds like you’ve got a handful!

Lazar: [laughs] Thank you!

“I was particularly surprised at the number of women who work here! I’ve never seen that in any IT company.”

You’ve been in NIRI for more than a month. How are you liking the company and what are your general impressions?

Lazar: The office space is very impressive! NIRI has a huge positive feedback from people who used to work here and they have nothing but kind words about this company. The scientific aspect and the scientific work the company does is very important. The atmosphere is quite comfortable, and I was particularly surprised at the number of women who work here! I’ve never seen that in any IT company. I like the fact that there are students here also, the demography is very versatile. Some former students that I know from my university are working here also. Who knows, I might do the same in the future! This internship has showed me what it’s like to work for a company such as NIRI and I really enjoy it so far.