Embed Size (px)
DESCRIPTION
In this presentation I discuss resilience and antifragility as behaviors resulting from the coupling of a system and its environment(s). Depending on the interactions between these two "ends" and on the quality of the individual behaviors that they may exercise, different strategies may be chosen: elasticity (change masking); entelechism (change tolerance); and antifragility (adapting to & learning from change). When the environment is very simple and only capable of so-called "random behavior", often the only effective strategy towards resilience is off-line dimensioning of redundancy as a result of a worst-case assessment of disturbances and/or threats. Much more complex and variegated is the case when both systems and environments are "intelligent" -- or at least able to exercise complex teleological and extrapolatory behaviors. In this case both system and ambient may choose among a variety of strategies in what could be regarded as a complex evolutionary game theory setting.
Citation preview
Vincenzo De Florio, INRIA, 14 Nov. 2014
A BehavioralInterpretation of Resilience
and Antifragility
Vincenzo De Florio / &
[email protected] This work is licensed under the Creative Commons Attribution-NoDerivatives 4.0 International
License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nd/4.0.
Vincenzo De Florio, INRIA, 14 Nov. 2014
ANTIFRAGILITY
RESILIENCE
Vincenzo De Florio, INRIA, 14 Nov. 2014
RESILIENCE
BEHAVIOR
Vincenzo De Florio, INRIA, 14 Nov. 2014
RESILIENCE
BEHAVIOR (SYSTEM, ENVIRONMENT)INTERPLAY b/w
IDENTITYPRESERVING
Vincenzo De Florio, INRIA, 14 Nov. 2014
Behavioristic approach• Rosenblueth, Wiener, & Bigelow, “Behavior,
purpose and teleology”, Phi.Sci. 10 (1943!):• “Given any system relatively abstracted from its
environment for study, the behavioristic approach consists in the examination of the output of the system and of the relations of this output to the input. - By output is meant any change produced in the environment by the system . - By input, conversely, is meant any event external to the system that modifies this system in any manner.”
Vincenzo De Florio, INRIA, 14 Nov. 2014
BehaviorPA
SSIV
E
AC
TIVE
RA
ND
OM
PUR
POSE
FUL
TELE
OLO
GIC
PREDICTIVE
System onlyreceives energy
from externalsource
System isthe source of the
output energy
behavior is not interpreted as
directed to a goal
A goal canbe identified
Distance fromthe goal steersthe behavior
Distance frompredicted position
of goal steersthe behavior
Vincenzo De Florio, INRIA, 14 Nov. 2014
Behavior classes
• From a systemic point of view:• Passive < active;
random < purposeful < teleological < predictive .
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilient behaviors
• Active behavior intended to retain the system identity• Aristotelian Entelechy!
"Being at work while staying the same"
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilient behaviors
• Active behavior, thus 1. only purposeful2. only teleological3. predictive
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilient, only-purposeful behavior
• ELASTICITY“The ability of a body that has been subjected to an external force to recover its size and shape, following deformation” (McGraw-Hill, 2003)
• No “advanced” behavior; system just makes use of its internal characteristics and resources so as to mask the action of external forces
→ Redundancy-based.
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilient, teleological & predictive behaviors
• ENTELECHISM• System is able to exert teleological or
predictive behaviors; it continuously adjusts its functions to compensate for changes
• so as to persist one's “identity”• IDENTITY: a system's
peculiar and distinctive functional and non-functional features
Vincenzo De Florio, INRIA, 14 Nov. 2014
Identity• Ability to "comply to one’s definition"• Related to fidelity: "compliance
between corresponding figures of interest in two separate but communicating domains"
• More in "Antifragility = Elasticity + Resilience + Machine Learning. Models and Algorithms for Open System Fidelity", http://goo.gl/VvtXRt
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilient behaviors (cont.d)
• Active behavior, thus 1. only purposeful2. only teleological3. predictive4. Auto-predictive
& evolving
ELASTICITY
ENTELECHISM
ANTIFRAGI-LITY
Vincenzo De Florio, INRIA, 14 Nov. 2014
Computational Antifragility• Behavior of a system that –predicts future sys-env fit – learns from the past & evolves
• "Being-at-work while improving-the-self"• System does not stay the same:
The feedback changes the "self"• Resilience + machine learning
(see cited paper)
Vincenzo De Florio, INRIA, 14 Nov. 2014
RESILIENCE
BEHAVIOR (SYSTEM, ENVIRONMENT)INTERPLAY b/w
IDENTITYPRESERVING
ANTIFRAGILITY
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilience: Interplay(S, E)
RESPONSE:PURPOSEFUL,NON-TELEO-LOGIC BEHA-
VIORS
INPUT:RANDOM
BEHAVIORS
SYSTEMENVIRONMENT
Better not to be too clever in the face of aturbulent environment!
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilience: Interplay(S, E)
RESPONSE:> PURPOSEFUL
BEHAVIORS
INPUT:PURPOSEFUL
NON-TELEOLOGICBEHAVIORS
SYSTEMENVIRONMENT
Vincenzo De Florio, INRIA, 14 Nov. 2014
Resilience: Interplay(S, E)
RESPONSE:> TELEOLOGIC/
PREDICTIVE BEHAVIORS
INPUT:TELEOLOGIC /PREDICTIVEBEHAVIORS
SYSTEMENVIRONMENT
Vincenzo De Florio, INRIA, 14 Nov. 2014
"Behavioral" Game Theory
Passive Random Purposeful Teleologic Predictive Antifragile
Passive ?,? -1,1 -1,1 -1,1 -1,1 -1,1
Random 1,-1 ?,? -1,1 -1,1 -1,1 -1,1 (*)
Purposeful 1,-1 1,-1 ?,? -1,1 -1,1 -1,1
Teleologic 1,-1 -1,1 1,-1 ?,? -1,1 -1,1
Predictive 1,-1 -1,1 1,-1 1,-1 ?,? -1,1
Antifragile 1,-1 1,-1 (*) 1,-1 1,-1 1,-1 ?,?
(*): if a player can learn that the other one is behaving randomly
ENVIRONMENT
SYST
EM
Vincenzo De Florio, INRIA, 14 Nov. 2014
Antifragility
• Conjecture: antifragility is (also) the ability to be "auto-resilient":–Monitor/Analyze/Learn the behaviors of the
"opponent"– Adjust one's behavior so as to maximize the
chances to "win"– Learn from mistakes and successes
Vincenzo De Florio, INRIA, 14 Nov. 2014
...Monitor/Analyze "the opponent"...
• Requirement: a model of the risk• The dynamic drifting of the risk as an
indication of the behaviorial class of E• Conjecture: the stability of the risk drifting
reveals aspects of E's behavior
Vincenzo De Florio, INRIA, 14 Nov. 2014
Conclusions
PASS
IVE
AC
TIVE
RA
ND
OM
PUR
POSE
FUL
TELE
OLO
GIC
PRED
ICTI
VE
System isthe source of the
output energy
A goal canbe identified
Distance frompredicted position
of goal steersthe behavior
ANTI-FRA-GILE
Social behaviorInterplay-aware
Feedback modifies"self"
Vincenzo De Florio, INRIA, 14 Nov. 2014
Conclusions
• Untrodden territory! Much yet to be done
• Urgent requirement!
Vincenzo De Florio, INRIA, 14 Nov. 2014
Conclusions
• ANTIFRAGILE 2015: http://goo.gl/4xJW69
• ERACLIOS: "Elasticity, Resilie-nce, Antifragility in CoLlective & Individual Objects and Systems" http://eraclios.blogspot.be/
• LinkedIn group "Computational Antifragility". http://goo.gl/1N0XB1
