CS 420/527 Contact Information Biologically-Inspired .

3y ago
35 Views
3 Downloads
4.77 MB
11 Pages
Last View : 1m ago
Last Download : 2m ago
Upload by : Abram Andresen
Transcription

1/11/12Contact InformationCS 420/527 Instructor: Bruce MacLennanBiologically-Inspired Computationmaclennan@eecs.utk.eduMin Kao 425Office Hours: 3:30–5:00 WF (or make appt.) Teaching Assistant:Bruce MacLennanweb.eecs.utk.edu/ mclennan/Classes/420Chuanjun He (che3@utk.edu)1/11/12CS 420 vs. CS 527Grading CS 420: Undergraduate credit (but graduatestudents can count one 400-level course) CS 527: Graduate credit, additional work (CS 527 is approved for the InterdisciplinaryGraduate Minor in Computational Science)1/11/123 You will conduct a series of computerexperiments, which you will write up Some of these will be run on off-the-shelfsimulators Others will be run on simulators that you willprogram Graduate students will do additional experimentsand mathematical exercises No exams1/11/12Prerequisites4Textbook CS 420 & 527: None per se, but you will berequired to write some simulations (in Java,C , NetLogo, or whatever) CS 527: Basic calculus through differentialequations, linear algebra, basic probabilityand statistics1/11/1225Flake, Gary William. The ComputationalBeauty of Nature. MIT Press, 19981/11/1261

1/11/12Contents of Flake CBN1/11/12What We Will Cover71/11/128Course Web SiteReading for Next Week web.eecs.utk.edu/ mclennan/Classes/420 Syllabus Link to Flake CBN site (with errata,software, etc.) Links to other interesting sites Handouts: Flake: Ch. 1 (Introduction) Flake: Ch. 15 (Cellular Automata)– assignments– slides in pdf format (revised after class) Models (simulation programs)1/11/1291/11/1210What is Biologically-InspiredComputation? Computer systems, devices, and algorithmsbased, more or less closely, on biologicalsystems Biomimicry applied to computing Approximately synonymous with: bioinspired computation, organic computingB. 2

1/11/12Two Kinds of ComputationMotivated by BiologyNatural Computation “Computation occurring in nature orinspired by that occurring in nature” Information processing occurs in naturalsystems from the DNA-level up through thebrain to the social level We can learn from these processes andapply them in CS (bio-inspired computing) In practice, can’t do one without the other Computation applied to biology– bioinformatics– computational biology– modeling DNA, cells, organs, populations, etc. Biology applied to computation––––biologically-inspired computationneural networksartificial lifeetc.1/11/1213Biological Computation1/11/12Why Do Bio-Inspired Computation? Biological systems are: Refers to the use of biological materials forcomputation––––––– e.g. DNA, proteins, viruses, bacteria Sometimes called “biocomputing” Goal: Biocomputers Bio-inspired computing need not be done rotectingself-*etc.1/11/12Some of the Natural Systems We Will Study adaptive pathminimization by ants wasp and termite nestbuilding army ant raiding fish schooling and birdflocking pattern formation inanimal coats coordinated cooperation inslime molds1416Some of the Artificial Systems We Will Study synchronized fireflyflashing soft constraint satisfactionin spin glasses evolution by naturalselection game theory and theevolution of cooperation computation at the edge ofchaos information processing inthe brain17 artificial neural networkssimulated annealingcellular automataant colony optimizationartificial immune systemsparticle swarm optimizationgenetic algorithmsother evolutionary computation systems1/11/12183

1/11/12C. AntsWhy Ants? Ants are successful:– 30% of Amazon biomass is ants and termites– Dry weight of social insects is four times that ofother land animals in Amazon– Perhaps 10% of Earth’s total biomass– Comparable to biomass of humansThink about the value of havingcomputers, networks, and robotsthat could do these things. Good source: Deborah Gordon: Ants atWork (1999)1/11/1219Intelligent Behavior of Harvester Ants1/11/12Army Ants Find shortest path to food Prioritize food sources based on distance & easeof access Adjust number involved in foraging based on:––––– No permanent nest Create temporary“bivouacs” frombodies of workers Raiding parties of upto 200 000 Act like unified entitycolony sizeamount of food storedamount of food in areapresence of other coloniesetc.1/11/122021Army Ant Raiding Patterns1/11/1222Coordination in Army Ant Colonies Timing:– nomadic phase (15 days)– stationary phase (20 days) Navigation in stationary phase5 mEciton hamatum1/11/12– 14 raids– 123 apartEciton burchellifrom Solé & Goodwin, Signs of Life231/11/12244

1/11/12Collective NavigationWeaver Ants Ants may use polarized sunlight todetermine direction But army ants have single-facet eyes – most insects have multiple facet eyes Theory: the two facets of individual ants ingroup function collectively as a multiplefacet eye1/11/1225Workers Bridging Gap1/11/12Form chains of bodies to bridge gapsOthers may cross these bridgesUse chains to pull leaf edges togetherConnect edges with silk from larvae held byworkers1/11/1226Adults Using Larvae as “Glue Guns”271/11/12(fig. from Self-Org. Biol.Sys.)28Fungus Cultivator Ants “Cultivate” fungi undergroundConstruct “gardens”Plant sporesWeed out competing fungiFertilize with compost from chewed leaves1/11/1229FungusCultivatorNest1/11/12(fig. from AntColony.org)305

1/11/12Transport of Cut LeavesLeaf Cutting Cut leaves are transported from source to nestalong trails Some temporarily held in caches near the tree Leaves being cut by workers1/11/12(fig. from AntColony.org)311/11/12Transporting Cut Leaves to Nest(fig. from AntColony.org)32Protection by Minims Small workers (minims) ride piggy-back Protect large workers from parasitic fly trying tolay eggs on head1/11/12(vid. from www.blueboard.com/leafcutters)331/11/12(fig. from AntColony.org)34Nest ConstructionA Large Nest Two mounds, 50 cm indiameter Part of a single nest Foraging trail visible Several tons of earth may be removed by large colony1/11/12(fig. from AntColony.org)351/11/12(vid. from www.blueboard.com/leafcutters)366

1/11/12Leaf Brought to Fungus Garden Leaf being brought to fungus garden in nest Leaf mulch is fed to fungus1/11/12(fig. from AntColony.org)The Fungus Garden Fungus grows special nutritional structures Ant larvae and adults can eat these371/11/12 Queen stays in fungusgarden Lays eggs Hatched larvae eatfungus Larvae cared for bynurse workers(fig. from AntColony.org)38Dump ChambersQueen in Fungus Garden1/11/12(fig. from AntColony.org)39 Dump chamber in lab In nature, may be 2munderground Contain:– waste leaf material– dead fungus– dead ants(fig. from AntColony.org)1/11/12Maeterlinck on “White Ants” (Termites)40Emergent Aspects Colony size 8 106“What governs here? What is it that issuesorders, foresees the future, elaborates plans,and preserves equilibrium, administers, andcondemns to death?”but no one is “in charge”! Colony lifetime 15 years Colonies have a “life cycle”– older behave differently from younger But ants live no longer than one year– Males live one day!1/11/12411/11/12427

1/11/12How Do They Do It?How do they do it? Semiochemically: deposit pheromones– 10-20 signs, many signal tasks– ants detect pheromone gradients and frequencyof encounter Communicationin Red HarvesterAnts Good source:DeborahGordon: Ants atWork (1999)1/11/12 Follow trails imperfectly exploration Feedback reinforces successful trails biased randomness(video from Stanford Report, April 2003)431/11/1244Circular CausalityDemonstration: Simulation of Ant ForagingRun NetLogo Ant-Foraging Global pattern emergent from total system Individuals respond to local field1/11/12451/11/12fig. from Solé & Goodwin461/11/12Fig, from EVALife48Stigmergy From stigmo/j pricking e1rgon work The project (work) in the environment is aninstigation Agent interactions may be:– direct– indirect (time-delayed through environment) Mediates individual and colony levels1/11/12478

1/11/121/11/12Fig, from EVALife49Fig. from EVALife50EmergenceAdvantages of Stigmergy The appearance of macroscopic patterns,properties, or behaviors that are not simply the “sum” of themicroscopic properties or behaviors of thecomponents Permits simpler agents Decreases direct communication between agents Incremental improvement Flexible, since when environment changes, agents respond appropriately1/11/121/11/12– non-linear but not chaotic Macroscopic order often described by fewer& different variables than microscopic order– e.g. ant trails vs. individual ants– order parameters511/11/1252D. Self-Organization Order may be imposed from outside asystem– to understand, look at the external source oforganization In self-organization, the order emerges fromthe system itself– must look at interactions within system In biological systems, the emergent orderoften has some adaptive purpose– e.g., efficient operation of ant colony1/11/12Part II531/11/12Fig. from NECSI549

1/11/12Why Are Complex Systems & Self-Organization Important for CS? Fundamental to theory & implementation of massivelyparallel, distributed computation systems How can millions of independent computational (orrobotic) agents cooperate to process information & achievegoals, in a way that ust in the face of damage or attackSome Principles UnderlyingEmergent Systems “More is different”“Ignorance is useful”“Encourage random encounters”“Look for patterns in signals”“Pay attention to your neighbor” (“Local information leads to global wisdom”)— Johnson, Emergence, pp. 77-9.1/11/12551/11/1256Comparison of Ant Colonies and Neural NetworksSimilar Principles of SO Ant coloniesNo. of unitsRobustnessConnectivityMemoryConnect. stabilityGlobal patternsComplex dynamics Development of embryo Molecular interactions within cell Neural networks1/11/12571/11/12Neural Netshighhighlocalshort/long termhighbrain wavescommonfrom Solé & Goodwin: Signs of Life, p. 14958Four Ingredients of Self-OrganizationSelf-Organization Concept originated in physics and chemistry– emergence of macroscopic patterns– out of microscopic processes & interactions Activity amplification by positive feedback Activity balancing by negative feedback “Self-organization is a set of dynamicalmechanisms whereby structures appear atthe global level of a system frominteractions among its lower-levelcomponents.” — Bonabeau, Dorigo & Theraulaz, p. 91/11/12Ant Colonieshighhighlocalshort-termweaktrailsobserved Amplification of random fluctuations Multiple Interactions— Bonabeau, Dorigo & Theraulaz, pp. 9-11591/11/126010

1/11/12Characteristics of Self-Organized SystemTwo Approaches to theProperties of Complex Systems Creation of spatiotemporal structures ininitially homogeneous medium Multistability Bifurcations when parameters are varied1/11/12— Bonabeau, Dorigo & Theraulaz, Swarm Intelligence, pp. 12-14611/11/12Focal Issue: Emergence Focal Issue: ComplexityDeals with: elements & interactionsBased on: relation between parts & wholeEmergent simplicityEmergent complexityImportance of scale (level)1/11/1262 Deals with: information & description Based on: relation of system to itsdescriptions Information theory & computation theoryare relevant Must be sensitive to level of description631/11/1264Additional Bibliography1.2.3.4.1/11/12Fig. from NECSI651/11/12Solé, Ricard, & Goodwin, Brian. Signs of Life: HowComplexity Pervades Biology. Basic Books, 2000.Bonabeau, Eric, Dorigo, Marco, & Theraulaz, Guy.Swarm Intelligence: From Natural to Artificial Systems.Oxford, 1999.Gordon, Deborah. Ants at Work: How an Insect SocietyIs Organized. Free Press, 1999.Johnson, Steven. Emergence: The Connected Lives ofAnts, Brains, Cities, and Software. Scribner, 2001. Apopular book, but with many good insights.Part II6611

Biology applied to computation! – biologically-inspired computation! – apply them in CS (bio-inspired computing)neural networks! – artificial life! – etc.! 1/11/12! 14! Natural Computation! “Computation occurring in nature or inspired by that occurring in nature”! Information processing occurs in natural

Related Documents:

420-3-1-.31 Prefabricated Septic Tank, Grease Trap And Holding Tank Permit 420-3-1-.32 Tank Manufacturer Records 420-3-1-.33 Effluent Filter Specifications . 420-3-1-.41 Sewage Tank Pumping Permit 420-3-1-.42 OSS Requiring Pumping Of Effluent Systems And Criteria For Special Sites 420-3-1-.43 Non-Waterborne Systems: Pit Privies

The 2012 version of ISO 527 October 2018 cb-fz 26 Menu End Nominal strain The solution is the use of nominal strain. ISO 527-1 defines two ways of measuring nominal strain. Used in ISO 527 since 1993 Needs two graphics to show all results correctly in stress-strain diagrams

Handbook of Green Chemistry ISBN (12 volumes): 978-3-527-31404-1 Set 1 Green Catalysis 978-3-527-31577-2 Volume 1: Homogeneous Catalysis Set 2 Green Solvents 978-3-527-31574-1 Volume 2: Heterogeneous Catalysis Set 3 Green Processes 978-3-527-31576-5 Volume 3: Biocatalysis Set 4 Green Product

e d i u G r e s 6U ShoreTel IP Phone 420 1 Getting Started Overview of the ShoreTel IP Phone 420 Overview of the ShoreTel IP Phone 420 Welcome to your ShoreTel IP phone! Figure 1 provides an overview of the IP420 phone components. Figure 1: ShoreTel IP Phone 420 Components Handset with finger rest Speaker Delivers high-quality sound Audio .

Energy Facilities Contractor’s Group (EFCOG) 9 1.3.1 10 CFR 851; § 851.27, Reference Sources. 1.3.2 DOE Order 420.1C, Facility Safety 1.3.3 DOE Guide 420.1-1 Nonreactor Nuclear Safety Design Guide For Use With DOE O 420.1C, Facility Safety 1.3.4 DOE G-420.1-3, Implementation Guide for DOE Fire Protection and Emergency

Office system bizhub 420 and finishing Excellent handling The bizhub 420 is a leading exponent in the art of quality printing and finishing alternatives combined with effective office workflow. Gone are the days of inconvenient and costly out-sourcing. The bizhub 420 just keeps on running

a 2 2 8 3 9330 1————5 10 420 5 5 11 2 140ue-h1c3-c32 — a 0 0 8 4 9430 1————5 10 420 5 5 11 2 140ue-h1c3-c40 — a 0 0 8 5 9530 1————5 10 420 5 5 11 2 140ue-h1c3-c50 — a 3 3 8 6 9630 1————5 13 420 6 5 11 2 140ue-h1c3-c63 — a 0 0 8 8 9

No de cours : 420-KHJ-LG à l'enseignement régulier Plan de cours Titre du cours : Intégration de techniques nouvelles en informatique industrielle Programme : 420.A0 Pondération : 2-2-2 Préalables : 420-KH6-LG Cours associés : 420-KH9-LG Discipline : Informatique Professeur Bureau Téléphone Courriel Patrice Roy F-314 poste 2780 atrice.roy@clg.qc.ca Cooordonnateurs Bureau Téléphone .