Adaptive Systems Lecture 1: Introduction to Concepts

1
Adaptive SystemsLecture 1 Introduction to
Concepts

• Dr Giovanna Di Marzo Serugendo
• Department of Computer Science
• and Information Systems
• Birkbeck College, University of London
• Email dimarzo_at_dcs.bbk.ac.uk
• Web Page http//www.dcs.bbk.ac.uk/dimarzo

2
Organisation

• Lectures 6.00pm-7.20pmRoom
• Practical Sessions 7.30pm-9.00pmRoom
• Simulations
• Presentations

3
Reading Material

• Lectures, practical sessions, information
  available from
• http//www.dcs.bbk.ac.uk/dimarzo/courses/as.html/
  
• Assessment
• Exam 70
• Presentation 10
• In-Lab-Test 20

4
Reading Material

• E. Bonabeau, M. Dorigo, and G. Théraulaz. Swarm
  IntelligenceFrom Natural to Artificial Systems
  Santa Fe Institute Studies on the Sciences of
  Complexity. Oxford University Press, UK, 1999.
• S. Camazine, J.-L. Deneubourg, Nigel R. F., J.
  Sneyd, G. Téraulaz, and E.Bonabeau.
  Self-Organisation in Biological Systems.
  Princeton Studies in Complexity. Princeton
  University Press, 2001.
• J.H. Holland, Emergence from Chaos to Order.
  Oxford University Press, 1998
• M. Wooldridge An Introduction to Multi-Agent
  Systems. Wiley, 2003
• L. M. de Castro Fundamentals of Natural
  Computing Basic Concepts, Algorithms, and
  Applications. Chapman Hall/CRC. 2006.

5
Syllabus

• Introduction to concepts
• Natural Adaptive Systems
• Adaptation Mechanisms
• Analysis / Simulation Methods
• Artificial Adaptive Systems
• Engineering Adaptive Systems

6
Motivation (1)

• Why a course on adaptive systems?

7
Motivation (2)

• Traditional engineered systems
• Client/Server
• Characteristics
• Distributed
• Centralised

8
Motivation (3)

• Current engineered systems
• P2P systems
• Grid systems
• Agent-based systems
• Ad-hoc networks
• Characteristics
• Decentralised
• Autonomous components
• Heterogeneity
• Pervasiveness
• Hybrid control

9
Motivation (4)

• Future engineered systems
• Self-managing systems
• Self-configuring, self-healing, self-protecting,
  self-optimising
• Self-organising networking systems
• Overlay networks
• Intrusion Detection Systems
• Sentient Computing
• Smart spaces
• Intelligent transport services
• Characteristics
• Large scale (number/worldwide)
• Decentralised Control
• Self-organisation
• Adaptive Systems

10
Engineered Adaptive Systems

• Inspiration
• From natural systems
• Social insects ants / wasps / termites
• Biological systems immune system
• Interactions mechanisms leading to
• Decentralised control
• Self-organisation
• Emergent Behaviour
• Adaptation and Robustness
• Artificial / ad-hoc techniques

11
Aims and Scope

• Study of engineered adaptive systems
• Engineering techniques
• Inspired from natural systems OR
• Developed specifically for artificial systems
• Engineering Issues
• Verification
• Control

12
Concepts

• Self-Organisation
• Emergent Phenomena
• Decentralised Control
• Adaptation
• Dynamic Change
• Complexity

13
Self-Organisation

• Natural systems
• Non-living systems
• Physical Process of pattern formation
• Living systems
• Biological Process of pattern formation
• Animals (stripes, coat patterns)
• Plants
• Collective behaviour
• Micro-organisms (cells)
• Social behaviour Swarms
• Human behaviour
• Engineered systems

14
Self-Organisation

• Natural systems
• Non-living systems
• Sand dune ripples
• http//www.desertusa.com/magjan98/dunes/jan_dune1.
  html

15
Self-Organisation

• Natural systems
• Non-living systems
• Mud cracks
• http//www.jameskay.com/gallery6/67_MD_25.html

16
Self-Organisation

• Natural systems
• Non-living systems
• Mud cracks
• http//scienceviews.com/photo/library/SIA0323.html

17
Self-Organisation

• Natural systems
• Living systems
• Zebra stripes
• http//grace.evergreen.edu/artofcomp/examples/zebr
  a/Zebra.html

18
Self-Organisation

• Natural systems
• Living systems
• Plants
• http//nov55.com/mr/

19
Self-Organisation

• Natural systems
• Living systems
• Ants Foraging
• Ants Foraging simulation http//website.lineone
  .net/john.montgomery/demos/ants.html
• Termites on trail

20
Self-Organisation

• Natural systems
• Living systems
• Schools of fish
• http//www.acclaimimages.com/_gallery/_pages/0018-
  0402-2904-5646.html

21
Self-Organisation

• Natural systems
• Living systems
• Human Behaviour
• http//www.terragalleria.com/vietnam/picture.viet8
  219.html

22
Self-Organisation

• Engineered Systems
• Distributed problem-solving
• Travelling salesman problem
• Routing in networks
• Graph partitioning
• But also
• Overlay networks
• Distributed operating systems
• P2P systems

23
Self-Organisation

• Some definitions
• Living systems
• Self-organisation is a process in which pattern
  at the global level of a system emerges solely
  from numerous interactions among the lower-level
  components of the system.
• Moreover, the rules specifying interactions among
  the systems components are executed using only
  local information, without reference to the
  global pattern.
• Camazine et al.

24
Self-Organisation

• Engineered Systems
• Strong self-organising systems are those systems
  where there is re-organisation with no explicit
  central control, either internal or external.
• Dimarzo et al. (2005)
• Weak self-organising systems are those systems
  where, from an internal point of view, there is
  re-organisation under an internal central control
  or planning.
• Dimarzo et al. (2005)
• Self-organisation is the process enabling a
  system to change its organisation in case of
  environmental changes without explicit external
  command .

25
Self-Organisation

• Interaction Modes
• Positive and Negative Feedback
• Negative Feedback
• Perturbation applied to system triggers response
  that counteracts the perturbation
• Towards stabilisation
• Avoids fluctuations
• Ex regulation of blood sugar level
• Negative feedback Increase of blood sugar level
  
• Response Insulin is released
• Result Blood sugar level regulated (decrease of
  blood glucose)

26
Self-Organisation

• Positive Feedback
• Initial change in a system is reinforced in the
  same direction as initial change
• Towards amplification
• Implies changes in the system
• Ex Fish nesting
• Initial change some fishes nest close to each
  other
• Positive feedback rule I nest where other
  similar individuals nest
• Increased aggregation of fishes at the same place
• )

27
Self-Organisation

• Positive Feedback coupled with Negative Feedback
• Positive feedback pushes system towards its
  limits
• Negative feedback or physical constrains provide
  inhibition and maintain system under control
• Ex Fish nesting
• Rule I nest where other similar individuals
  nest unless there are too much fishes
• - Camazine et al (1999)

28
Self-Organisation

• Interaction Mechanisms
• Direct and Indirect (through environment)
  communication
• Direct Communication
• Schools of fishes
• Information acquired from neighbour
• Indirect communication
• Information acquired from shared local
  environment and work-in-progress Stigmergy
• Social Insects
• State of termites mound provides information to
  builders
• Ant pheromone trails
• pheromone deposited into environment guides ants

29
Self-Organisation

• Characteristics
• Large number of components
• Large number of (local) interactions
• Decentralised control (No control imposed from
  outside)
• Local information
• Individuals components have simple behaviour
• No global pattern / recipe / rule to refer to
• No one knows how to build to whole
• Simple interactions lead to complex /
  sophisticated structures
• Pattern is an emergent property
• Pattern is not a property imposed on the system
  by external ordering influence

30
Emergent Phenomena

• Structure-formation
• Structure pattern / function / property
• Examples
• Sand dune ripples
• Zebra stripes
• Emergence of consciousness in human
• Emerges from neurons interactions
• Shortest path to food found by foraging ants
• Engineered systems
• TSP shortest path to visit all cities

31
Emergent Phenomena

• Some definitions
• The whole is more than the sum of the parts.
• Holland (1998)
• A structure (pattern, property or function), not
  explicitly represented at the level of the
  individual components (lower level), and which
  appears at the level of the system (higher
  level).

32
Emergent Phenomena

• External Observer
• Emergent phenomena have a meaning for an observer
  external to the system but not for the system
  itself.
• Emergent Phenomena
• Observed patterns or functions which have no
  causal effect on the system itself.
• Stones ordered by sea ordering of the stones has
  no effect at all on the whole system made of the
  stones and the sea (Castelfranchi, 2001)
• Observed functions which have a causal effect on
  the system.
• Desired or not
• Have an effect on the system behaviour.
• Cause the individual parts to modify their own
  behaviour.

33
Emergent Phenomena

• Engineered systems
• Large number of individual components, i.e., of
  autonomous computation entities.
• Large number of interactions occur, among these
  components, whose ordering, content and purpose
  are not necessarily imposed.
• Difficult to predict the exact behaviour of the
  system taken as a whole because of the large
  number of possible non-deterministic ways the
  system can behave.
• Use of simulations to predict / tune the result

34
Emergent Phenomena

• Engineered systems
• But since we have built the system, the
  individual behaviours components and the local
  rules governing the system are known, it becomes
  then in principle possible to determine the
  (emergent) systems behaviour.
• In practice, current techniques or calculations
  (essentially simulations) are not sufficient and
  make it almost impossible to determine the
  result. That is why the result, functions or
  properties, is said to be emergent.

35
Decentralised Control

• Coordination of work without central decisions
• Self-organisation mechanisms are (mostly) based
  on decentralised architectures of information
  flow
• No instruction issued from leaders
• Individuals gather information (directly or not)
  and decides what to do

36
Decentralised Control

• Two kinds of engineered systems with
  decentralised control
• Case 1 Large set of autonomous components,
  pertaining to the same system and providing as a
  whole expected properties, or functions.
• Engineering with emergent functionality in mind
• Case 2 Large set of autonomous components,
  spontaneously interacting with each other, for
  possibly independent or competing reasons.
• No expected emergent global function or
  properties
• In both cases, autonomous components may be
• heterogeneous
• dynamically joining and leaving the system.

37
Decentralised Control

• Decentralised control (Case 1) allows
• Distribution of computation and decisions among
  the different components
• no need for a central powerful computer
• Increased robustness
• the system does not rely on a single node that
  may fail and crash the whole system
• Better use of network and CPU resources
• communication does not occur among a dedicated
  central node and a large number of components,
  but locally among the whole set of components
• Flexible schema for communication in highly
  dynamic systems
• Communication with a neighbour instead of with
  the central entity

38
When Self-organisation meets Emergence

• Self-organisation without emergent phenomenon
• When there is internal central control
• System finds a new organisation fully deducible
  from the central entity
• Ex Termites under central control of queen
• Weak self-organisation

39
When Self-organisation meets Emergence

• Emergent phenomenon without Self-organisation
• No causal effect on the system
• No re-organisation
• Ex stones ordered by sea
• Physical systems with negative feedback only

40
When Self-organisation meets Emergence

• Self-organisation together with emergent
  phenomenon
• Dynamic individual components
• Decentralised control
• Local interactions among components
• Self-organisation is a structure-formation
  process (Camazine def.)
• Strong self-organisation

41
Adaptation

• In living systems
• Individual components behave according to genetic
  programs (rules) tuned by natural selection
• Natural selection tunes interactions rules
• Natural selection shapes the emergent structures
  (patterns or function)
• Adaptation to long-term environmental changes

42
Adaptation

• In living systems
• Individual components behave autonomously
• Local information (up-to-date)
• Immediate response of system
• Adaptation to immediate environmental changes
• Adaptation expected for engineered systems

43
Dynamic Change

• Continual interactions among components
• Components join and leave system at any time
• Dynamic systems
• Ants foraging
• Skype system of users

44
Complexity

• Individuals organisms use relatively simple
  behavioural rules
• Generated structure and patterns are more complex
  than the components from which they emerge
• Complexity comes from
• Sensitivity to initial conditions (butterfly
  effect)
• Non-linear interactions among components
  involving amplification and cooperation
• Living systems are more complex than non-living
  ones
• Non-living ones subject to physical laws only
• Living ones subject to physical laws
  behavioural interactions influenced by
  genetically controlled properties
• - Camazine et al (1999)

45
References

• Camazine et al (2001) -- S. Camazine, J.-L.
  Deneubourg, Nigel R. F., J. Sneyd, G. Théraulaz,
  and E. Bonabeau. Self-Organisation in Biological
  Systems. Princeton Studies in Complexity.
  Princeton University Press, 2001.
• Holland (1998) -- J.H. Holland, Emergence from
  Chaos to Order. Oxford University Press, 1998.
• G. Di Marzo Serugendo, M.-P. Gleizes, A.
  Karageorgos Self-organisation and emergence in
  MAS an overview, Informatica, Ljubljana,
  Slovenia. In press, 2005.