Symposium Organic Computing

1
Symposium Organic Computing Towards structured
design of processes
Summary Section II - Bridges (Chair R.Douglas)
B.SendhoffHonda RD Europe GmbH Offenbach,
Germany
2
From where we started
We are here...
3
1st presentation - Prof. Meyer
Whats software engineering doing and where is it?
Spaghetticode the dark old ages
technology
programming - mathematics

• start from formal specifications
• make the proofing inherent at each step
• end of the process is the executable
• expensive process

programming methodology conditions - loops - etc.
object technology composition/decomposition modula
rity structure building / taxonomy
object technology degradation for the people

• components have to be trusted
• otherwise
• errors multiply and propagate

management
complete process management seamless
development the Escher waterfal spiral
structure/architecture
component based development re-usable elements
goal is better software quality
4
Discussion

• Strong bridge with regard to
• hierarchical system
• information hiding/masking
• abstracting the important fromthe unimportant

BiologicalInformation Processing
SoftwareEngineering
IBM eLiza project
Can logic solve all problems? Can it do it most
efficiently?
subject modularity
how can we enhance modularity? - Influence the
search distribution
is there a relation between the interchangability
of modules and strong causality?
are the interactions among modules sparse? are
they in nature?
subject proof of correctness
problem of proof is the richness of environment -
element of security
proof of correctness has a similarity to early
evolution, error threshold for replication
proof of correctness is only possible for
software, however we need a proof of system
correctness including hardware - impossible?!
proof is inherently impossible
instead of proofing we have to watch repair?
5
2nd presentation - Prof. Ritter
light space information or choosing the right
way of thinking
the brain and the computer
similarities process information and need memory
differences specialised structure (the brain)
vs. universal structure (the computer) wet vs.
hardware, production vs. growth
What can we learn from looking at simpler
systems? Evolution of light technology and the
spatial organisation of bricks in a bottle

• different scientific approaches might be needed
  and be beneficial to achieve the same
• do not put all of your money on
  one horse
• computation can arise effortlessly if the
  constraints are set right (i.e., the bottle, the
  bricks)
• self-organised solutions, which are robust, have
  natural dynamics and form structuresit occurs
  in very simple systems
• in the opposite in software design each brick
  has to be carefully programmed and the
  inter-brick interfaces have to be defined -
  nothing falls together!

aspect of purpose?
Where to go...

• find Newtons law for information objects
• in information systems, modules must naturally
  fit together, e.g. chemistry atoms - molecules
  - proteins

6
Discussion
subject information
(was about high time...)

• information and knowledge technical, semantic,
  pragmatic
• we have to take the change of the state of the
  receiver into account
• maybe we should not try to define it but live
  with it

subject evolution

• the structure of the system depends to a large
  degree on the path evolutiontook, it is not
  simply reproducable by restarting the process
• are there ways to controll the error limits of
  self-organisation?voting - commitee - averaging
• we need an active approach to robustness with
  prediction
• mechanism for building complex system is
  instantiation, thus we need a template
• Is there a macroscopic direction for the
  evolution of information processing,like energy
  or entropy in thermodynamics and mechanics?

7
Discussion
subject psychology

• does psychology provide the richest
  analogy?manipulation of symbols, what bit of
  brain makes introspection possible?

subject specialised vs. universal

• the brain is specialised but potentially
  universal, realised through learning
• the computer is universal but becomes very
  specialised with software
• computer science copying is cheap, in brains it
  is difficult, the same applies to natural
  evolution

8
3rd presentation - Prof. von Seelen
the brain and the genome
target organisation of structures by
information, both structures are the product
of evolution
thus look at the common production process
On adaptation and evolution

• both systems are inherently and principally
  flexible (evolvable and adaptation) construction
  phase is extended through adaptation
• complexity can be reduced by a problem-dependent
  variation of structure
• adapt the system to the structure of the problems
  is a definition for organic computing
• each task has a specialised structure

On modularity

• modularity as a common construction principle
• modularity in the vision system, microcolumns,
  etc.,
• modularity and neutrality, interconnecting path
  provide stable exploration?

9
Discussion
subject evolutionary biology

• central principle deal with the issue of
  exploitation
• for modularisation the degree of complexity
  should be equivalent approach is to subject the
  constuction process to attack during the process
• does the genetic machinery represent
  intelligence?
• the speed of evolution has been optimised during
  the process does it seem as if evolution
  optimises itself?

subject feedback

• how to exploit positive and negative feedback
  loops for construction, during dreams our
  system changes to a mode which it needs,
• a dilemma of feedback? being sensitive to the
  outside and remain stable
• feedback is often used in the brain, information
  processing at the stability borderedge of chaos
  hypotheses
• feedback in the eye to enhance the contrast in
  morphogenesis to tune the system spontaneously to
  a pattern

10
Discussion
subjecttraditional logical approach vs. organic
computing

• a pure logical approach cannot cope with
  real-world noise distributions, in particular in
  natural environments

what is the typical problem which we can solve
specifically with organic computing?
11
Summary Remarks
buzzword top ten from the discussion

• robustness
• specialised vs. universal
• modularity
• evolvability (same as 1?)
• hierarchy
• re-usable components
• flexibility

12
conclusion
Can we define some questions we should ask?
What can the organic aspect contribute to
computing and for which task?
Do we have to change our way of thinking about
the problem? There are some principal problems to
be solved
Can we grow a simple organic computer? get rid
of our current thinking...
13
the risk