Database Reference
In-Depth Information
Chapter 1
Brave New Realtime World: Introduction
Abstract The chapter offers a general introduction to methods of realtime
analytics and sets out their advantages and disadvantages as compared with
conventional analytics methods, which learn only from historical data. In particular,
we stress the difficulties in the development of theoretically sound realtime analyt-
ics methods. We emphasize that such online learning does not conflict with
conventional offline learning but, on the opposite, both complement each other.
Finally, we give some methodical remarks.
1.1 Historical Perspective
“Study cybernetics!” the Soviet author Viktor Pekelis urged his young readers, of whom
I was one, in 1977 [Pek77]. But I didn't, not least because by the time I could have done so,
cybernetics was no longer available as a study option. By the end of the 1970s, after more
than 25 years, the wave of enthusiasm for cybernetics had finally ebbed [Pia04]. So what
had gone wrong?
Cybernetics was established in the late 1940s by the American mathematician Norbert
Wiener as a scientific field of study exploring the open- and closed-loop control of
machines, living organisms and even entire social organizations [Wien48]. Cybernetics
was also defined as the “art of control,” and feedback in particular played a central role
here. Its purpose was to ensure that systems do not get out of hand but instead adapt
successfully to their environment. The thermostat is a classic example of a cybernetic
control.
In fact the scientific benefits were immense: cybernetics brought together such diverse
disciplines as control theory, neurology and information theory, and leading scientists such
as John von Neumann, Warren McCulloch and Claude Shannon were involved in its
development. It caused a sensation in the media. The possibilities offered by this new
discipline seemed infinite: robots would take on day-to-day chores, factories would manage
themselves, and computers would write poetry and compose music. More ambitiously still,
from 1971 onwards the Cybersyn project in Chile headed up by the Englishman Stafford
Beer sought to establish a centralized system of cybernetic economic control [Beer59]. And
in the Soviet Union the OGAS project [GV81] led by pioneering cyberneticists Viktor
Glushkov and Anatoly Kitov aimed to bring the entire Soviet planned economy under
automated control.
