Cognitive Self-Synchronisation


Published on

Published in: Technology, Education
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Cognitive Self-Synchronisation

  1. 1. Defining and MeasuringDefining and MeasuringDefining and MeasuringDefining and MeasuringCognitiveCognitiveCognitiveCognitive----Entropy andEntropy andEntropy andEntropy andCognitive SelfCognitive SelfCognitive SelfCognitive Self----SynchronizationSynchronizationSynchronizationSynchronizationCognitive SelfCognitive SelfCognitive SelfCognitive Self----SynchronizationSynchronizationSynchronizationSynchronizationMarco Manso (SAS-065 member, PT)Dr. James Moffat (DSTL, UK)
  2. 2. Agenda• Introduction– To concepts and base theory– To ELICIT, the experimentation platform used• Self-Synchronization in the CognitiveDomain– Definition of Cognitive-Entropy– Definition of Cognitive-Entropy– Definition of Cognitive Self-Synchronization• Preliminary lessons from Experiments– A simple model– Enablers and Inhibitors– Measurements and Results• Conclusions and Lessons Learned2Cognitive Self-Synchronization
  3. 3. Introduction• New military challenges – new C2approaches• NEC as an important step ?(SAS-065, 2010, pp. 27)3Cognitive Self-Synchronization
  4. 4. Introduction• Self - Synchronization:– NCW key-aspect– Describes the ability of a well-informed force toorganize and synchronize complex warfare activitiesfrom the bottom up(Cebrowski, Arthur K. and Garstka, 1998)Comprises 2 main aspects:Comprises 2 main aspects:1. Synchronization: as an output characteristic of theC2 processes that arrange and continually adapt therelationships of actions in time and space […]Synchronization takes place in the physical domain(Alberts et. al., 2001).2. Self: a result from the bottom up (in this context, asa result of developing shared awareness enabled bynetworking) without the need for guidance fromoutside the system (Atkinson and Moffat, 2005).Cognitive Self-Synchronization 4
  5. 5. Introduction• Self-Synchronization:– An important concept (in NEC and C2)– Should be applied to the cognitive-domain forassessment purposes (during and aftermissions)– Challenge taken herein:– Challenge taken herein:• Define and measure it (based on existingexperiments) !• Identify a set of enablers and inhibitors.– Concepts - first defined in (Manso and B.Manso 2010):• Cognitive Entropy• Cognitive Self-SynchronizationCognitive Self-Synchronization 5
  6. 6. Introduction – ELICIT experimental Platform• Research and experimentation platform• Developed to:– conduct research related with collaboration, information sharingand trust– test hypothesis related with edge and hierarchical (traditional)command and control practices.• Network-Enabled environment:– Played by 17 Subjects– Played by 17 Subjects– Must determine the who, what, where and when of a futureterrorist attack– Subjects receive pieces of information that they must share inorder to develop sufficient awareness to guess the solution.– Subjects may share information by posting it to websites (actionpost) and/or sending it directly to other subjects (action share).• The platform allows instantiating different C2 approaches(e.g., define roles and interactions allowed)• Data was available from experiments conducted in Portugal.Cognitive Self-Synchronization 6
  7. 7. Measuring Self-Synchronization• Two variables were created:– Cognitive Self-Synchronization (CSSync)– Cognitive Entropy (CE) (its counterpart)• First introduced in (Manso and B. Manso 2010)and based on Moffat’s work towards developinga knowledge metric (Moffat 2003) to measurea knowledge metric (Moffat 2003) to measurethe amount of uncertainty in a probabilitydistribution (Shannon’s Information Entropy)• Now based on the scientific field of Complexitytheory, namely, the Kolmogorov complexity - ameasure of the descriptive complexity of anobject (Cover and Thomas 1991)Cognitive Self-Synchronization 7
  8. 8. Measuring Self-Synchronization• Research Problem: how to measure(quantitatively) the degree of convergence of agroup towards the ELICIT problem?SubjectsProblem SpacesWhat is the group overallSubjects IDs @ time_tWhat is the group overallSelf-Synchronization(in the cognitive domain)?8Cognitive Self-Synchronization
  9. 9. Kolmogorov complexity• Expected description length of dataset D :log ( )P D− = Entropy of D= Kolmogorov Complexity of D• More generally:{ }{ }1 211 2( )log ( ) expected description length of the datasets , ,.....,information entropy of , ,.....,Ni i NiNp D p D D D DD D D=− ==∑9Cognitive Self-Synchronization
  10. 10. Defining and MeasuringCognitive-Entropy• Inputs:– 17 subjects playing the game (N=17)– 4 solution spaces: who, what, where andwhen (assumed independent for simplicity)– Subjects may ID over time (no ID=null case)• For each solution space i at time t, wethus define:( , , ) Number of IDs for solution space at time of typeS i t k i t k=10Cognitive Self-Synchronization
  11. 11. Defining and MeasuringCognitive-Entropy• Number of Positive IDs:– probability of each ID description:1, ( , , ) 0( , , )k Kk S i t kS i t k== ≠∑( , , )( , , )17S i t kp i t k =• Null case (no ID):– probability of this description:1, ( , , ) 017 ( , , )k Kk S i t kS i t k== ≠− ∑1( , , ) where denotes the null set.17p i t k = ∅ = ∅1711Cognitive Self-Synchronization
  12. 12. Defining and MeasuringCognitive-Entropy• Cognitive entropy CE– for solution space i– at time t1, ( , , ) 0 1, ( , , ) 01 1( , ) ( , , )log ( , , ) 17 ( , , ) log17 17k K k Kk S i t k k S i t kCE i t p i t k p i t k S i t k= == ≠ = ≠   = − + −      ∑ ∑Positive IDs Null case (no IDs)12Cognitive Self-SynchronizationUncertainty parcel
  13. 13. Defining and Measuring CSSync• Counterpart of Cognitive-Entropy• Where:PrPr( , )( , ) 1_oblemSpaceoblemSpaceCE i tCSSync i tMax Disorder= −– CE(i, t) is the Cognitive-Entropy of solution space i at time t.–– CSSync = 0 means system is fully disordered– CSSync = 1 means system is fully ordered11 1_ *log( ) log( )NProblemSpaceiMax Disorder NN N== − =∑13Cognitive Self-Synchronization
  14. 14. Defining and Measuring CSSync• For ELICIT, the overall CSSync is:( ) 0.25* ( , )i ProblemSpaceCSSync t CSSync i t== ∑• OBS:– used equal weights (25%) for each of the 4 solutionspaces.– Assumed each solution space to be independentfrom each other.14Cognitive Self-Synchronization
  15. 15. Defining and Measuring CSSync• Illustrative Example (1): fully-disordered system(Manso and B. Manso 2010)A????Z?15Cognitive Self-SynchronizationK??????J??B
  16. 16. Defining and Measuring CSSync• Illustrative Example (2): (about) half-orderedsystem (Manso and B. Manso 2010)AAAAAKKAKA?K??AKKAA16Cognitive Self-Synchronization
  17. 17. Defining and Measuring CSSync• Illustrative Example (3): fully-ordered system(Manso and B. Manso 2010)AAAAAAA AAAAAAAAAAAA17Cognitive Self-Synchronization
  18. 18. CSSync Enablers and Inhibitors:an exploratory viewUse existing experimentation data to explore thefollowing questions:• Q1: What aspects enable the emergence ofSelf-Synchronization?Self-Synchronization?• Q2: What aspects inhibit the emergence ofSelf-Synchronization?• Q3: What is the associated cost to Self-Synchronize?18Cognitive Self-Synchronization
  19. 19. CSSync Enablers and Inhibitors:an exploratory viewA Simple Model:Cognitive System(collective)Network access(members and websites)Organization goals,roles and structureAllocation ofdecision rightsC2ApproachCE and CSSyncEffort SpentExtent of Correct Awarenessdecision rightsProblem difficultyNumber ofsubjectsDistribution ofInformation (by server)Collaborative mechanisms(share/post/pull)IndependentVariablesOther relevantvariables(fixed)Subjects’ competence(assumed fixed)19Cognitive Self-Synchronization
  20. 20. CSSync Enablers and Inhibitors:an exploratory viewA Simple Model:Cognitive System(collective)Network access(members and websites)Organization goals,roles and structureC2ApproachCE and CSSyncEffort SpentConflicted C2Deconflicted C2(collective)Allocation ofdecision rightsProblem difficultyNumber ofsubjectsDistribution ofInformation (by server)Collaborative mechanisms(share/post/pull)C2ApproachIndependentVariablesOther relevantvariables(fixed)Extent of Correct AwarenessSubjects’ competence(assumed fixed)Coordinated C2Collaborative C2Edge C2CE and CSSyncBased on the N2C2M2C2 Approaches(SAS-065, 2010, pp. 27)20Cognitive Self-Synchronization
  21. 21. CSSync Enablers and Inhibitors:an exploratory viewAPPROACH Mean MIN MAXCONFLICTED 0.05 0.00 0.09DECONFLICTED 0.12 0.04 0.22COORDINATED 0.15 0.10 0.22COLLABORATIVE 0.34 0.22 0.42EDGE 0.41 0.40 0.42EDGE 0.41 0.40 0.4221Cognitive Self-Synchronization
  22. 22. CSSync Enablers and Inhibitors:an exploratory view• Conflicted C222Cognitive Self-Synchronization
  23. 23. CSSync Enablers and Inhibitors:an exploratory view• Deconflicted C223Cognitive Self-Synchronization
  24. 24. CSSync Enablers and Inhibitors:an exploratory view• Coordinated C224Cognitive Self-Synchronization
  25. 25. CSSync Enablers and Inhibitors:an exploratory view• Collaborative C225Cognitive Self-Synchronization
  26. 26. CSSync Enablers and Inhibitors:an exploratory view• Edge C226Cognitive Self-Synchronization
  27. 27. CSSync Enablers and Inhibitors:an exploratory viewCSSyncCategory CSSync Inhibitors CSSync EnablersQ1: What aspects enable the emergence of Self-Synchronization?Q2: What aspects inhibit the emergence of Self-Synchronization?Shared InformationResourcesNone or a few shared(mainly kept within ownentities)Shared across members.All information accessibleacross entities.Patterns ofInteractionsNon-existent or highlyconstrainedUnconstrained / broadand rich across entitiesand subjectsAllocation of DecisionRightsNone / fixed task-rolebasedDistributed (to allsubjects)27Cognitive Self-Synchronization
  28. 28. CSSync Enablers and Inhibitors:an exploratory viewQ3: What is the associated cost to Self-Synchronize?1 4001 6001 800Effort(cost)02004006008001 0001 200CONFLICTED DECONFLICTED COORDINATED COLLABORATIVE EDGEIds per HourPullsper HourPosts per HourSharesper Hour28Cognitive Self-Synchronization
  29. 29. CSSync Enablers and Inhibitors:an exploratory viewQ3: What is the associated cost to Self-Synchronize?0,8000,9001,000Relationbetween Effort and CSSync0,0000,1000,2000,3000,4000,5000,6000,7000 200 400 600 800 1000 1200 1400CSSyncEffortCSSyncLinear (CSSync)29Cognitive Self-Synchronization
  30. 30. Conclusions and Way Ahead• CE and CSSync concepts defined and measuredin experiments.• We raised first indicants for enablers andinhibitors for CSSync (as well as cost)• The ability to self-synchronize in the cognitivedomain shows a steady improvement with the C2domain shows a steady improvement with the C2Approach adopted in the game.• This steady improvement in cognitive self-synchronization with C2 Approach is also directlyrelated to the level of activity (the energy oractivity ‘cost’) required to sustain that C2Approach.• ELICIT has been shown to give important insightsfor the attack scenario used.Cognitive Self-Synchronization 30
  31. 31. Conclusions and Way Ahead• Increase the experimentation data set and observevalues for CSSync beyond 0.5• Measure CE and CSSync to C2-relatedexperiments using different experimentationplatforms, including DSTL’s WISE wargame. (Moffat2003).• Manipulate additional relevant input variables.• Manipulate additional relevant input variables.Cover multiple levels of complex networksincluding (i) Base level (network characteristics),(ii) Median Level (intelligent node interactions) and(iii) Top level (NEC Effects) (Moffat 2007).• Further extend the application of entropy tonetwork-entropy (Lin et. al. 2010) and information-entropy (Jin and Liu 2009) and identify relationsbetween them.Cognitive Self-Synchronization 31
  32. 32. Thank you for your attentionMarco MansoSAS-065 Member, Portugal(sponsored by theDr. James MoffatDefence Science andTechnology Laboratory, UK(sponsored by theCenter for Edge Power of theNaval Post Graduate School)me@marcomanso.comTechnology Laboratory, UK32Cognitive Self-Synchronization