Organizations that use their data well, have competitive advantages. So good design of data infrastructure is becoming increasingly important. “But data must be meaningful. Where does that data come from? From which systems does this data come and which connections can be made between datasets? Also important, how is the data governance set up? “Remus Lazar, Watson Data Platform Director, Data Science and Machine Learning at IBM is working on these questions every day. He takes the visitors of his presentation at the Big Data Expo in Mechelen on a journey along the challenges of Big Data. The Watson Data Platform forms the basis.
Big data vs. quantum computers
Recently, more news was known about the development of a supercomputer that works according to the quantum principle: Delft and Eindhoven physicists wrote in the renowned scientific journal Nature about new evidence for the existence of the Majorana particle. That particle would be a good building block to base a quantum computer on. This means that a stable quantum computer has come a step closer. The physicist Leo Kouwenhoven from Delft already announced in 2012 at a conference in Boston that he was near the discovery of this particle, which eighty years in vain was sought after.
The Italian theoretician Ettore Majorana (1906-1938) predicted in 1937 Majorana particles must exist. Since then scientists have searched extensively but never found them. The Delft researchers did get strong indications that they could generate Majorana-like behavior in superconducting chips.
IBM tackles it differently
IBM and Google use different methods and have already started using the first test computers that may eventually be artificial support intelligence and answer human questions. There is a race with other technology giants, such as Microsoft. That has established the hope for Majorana particles for the construction of reliable quantum computers, because in theory they are very stable in a memory chip. This is due to their topological properties. For mathematical reasons Majorana particles can not change state.
Interpretation of big data
All supercomputers revolve around the interpretation of large amounts of data. Logically the central theme of the Big Data Expo. A special feature of IBM’s Watson data platform is that it has already proven itself in practice. In February 2011 Watson participated in the TV program Jeopardy !, a TV quiz where candidates can win money if they answer questions correctly. This American quiz has a Flemish version under the name Waagstuk, a game program on the Belgian channel VTM in the 90s.
Watson wins Jeopardy!
The first version of Watson, which was developed in 2007, only knew the to answer half of the questions correctly. To beat the best human participants, he had to have 90 percent of the answers right. Watson, for example, had to recognize irony and riddles, something that computers have more trouble with than people. In February 2011 Watson again participated in Jeopardy! and he took on the two biggest players in the history of the program: Brad Rutter and Ken Jennings. Rutter played the same round against Watson in the first round but IBM’s supercomputer then convincingly won all other rounds. The final ranking of the three-day competition was: Watson: $ 77,147, Jennings: $ 24,000 and Rutter: $ 21,600. Two weeks after his successful TV appearance, Watson took it in a Jeopardy trial! against five members of the US Congress. Watson also won this competition ($ 40,300- $ 30,000) but lost to Congressman Rush D. Holt, Jr., a former participant.
Watson consists of a self-learning system that has a spoken language can interpret. It uses a collection of encyclopaedias, books, magazines, scientific articles and downloaded websites to give a good answer to questions within seconds. Watson uses tailor-made ‘deep-question-and-answer software’ that goes beyond conventional artificial intelligence. Watson also calculates the reliability of his answers, which explains the good scores on the Jeopardy!
Big data challenges
Remus Lazar explains how IBM meets Watson’s challenges with big data: “Watson started as an internal project. Since September 2016 we have also made this platform available for external use. Engineers, application developers and business analysts can use these to develop their own big data applications. We respond to their needs by optimally controlling access control. Naturally, the governance, supervision and management of our datasets and analysis models satisfies the European Union’s requirements for this by means of the [ddbpr legislation. In our high-security data centers in London and Frankfurt, IBM can store data reliably and redundantly and make it available to our customers for analysis and research for pipeline computing .
According to Lazar, developers, engineers and business analysts are faced with the combination of big data and machine learning formerly unprecedented tools to lift artificial intelligence to a higher level. “But it is important to first think carefully about the implications that technology will have for our lives. IBM has a long history in this area and can make data accessible in many ways, taking into account the risks. Computers will increasingly be able to communicate with people in ordinary language. In my presentation I will explain how IBM helps organizations to simplify the management of the necessary data to keep a grip on them. “