of 15

Boldly Innovating Innovation via Creative Machine Intelligence

Jun 14, 2015




Innovation Demands Boldness Conference, Arizona State University, 2010.

  • 1. Boldly Innovating InnovationCreativity Machine Simple & Elegantvia Creative MachineParadigmA noise-drivendialog between atleast two artificial Intelligencehopping synapticneural networksgenerates new (A Feigned Retreat from Complexity)ideas. perturbations Makes ItselfArbitrarily Complex Stephen L. Thaler, Ph.D.Self-organization President & CEO,inherent toImagination Engines, Inc.artificial neuralnetworks makes andthis so.Founder,Scalable Machine In Its Image, Inc.ideasIntelligence& ConsciousnessNeural Innovation Demands Boldnessbrainstorming Scottsdale, AZ 30 September, 2010session is thebasis of human __________________________________cognition.A radically new form of totally self-organizing, autonomous,Is that all there is?opinionscontemplative, and creative machine intelligence may possibly beRequires utmostthe poster child for all disruptive technologies since it may, byboldness to definition, generate all subsequent technologies. The lessonsadvance,cumulatively learned from the advancement of what might arguablyThe Ultimate Idea realizing itsphilosophicalbe called the ultimate idea should be especially interesting to thisworkshop since the approaches used have boldly exploited thebecause it generates all others! implications.current retreat from complexity, bypassed conventional reviewmethods, and harnessed the concepts inherent disruptiveness, 2010, Imagination Engines, Inc. both technologically and philosophically, as an extremely effectivemarketing vehicle. 2010, Imagination Engines, Inc.

2. THE PERCEPTRONraw environmental input patterns associated output patterns (i.e., opinions) 2010, Imagination Engines, Inc. 3. THE IMAGITRON 10.9 memories + confabulationsinput optimal memoryPmem , probability of activating a memory0.8 generation rate0.7 highest constraint satisfaction bymemories0.6 confabulations output0.5 N0.4 (w w ) 2i=1 i i0.3 N0.2wi = ith weightN = no. of weights0.1 000.1 0.2 0.3 0.40.5 0.6 0.70.8 0.9 1 root-mean-square transient synaptic fluctuation 2010, Imagination Engines, Inc. 4. DEVICE FOR THE AUTONOMOUS GENERATION OF USEFUL INFORMATION (DAGUI) Imagitron quickly absorbs Zen of conceptual space, without human auxiliary sensory inputsinvolvement. Perceptron quickly learns to form opinions about novel patternsIMAGITRONemerging from imagitron, without human involvement.weight fluctuations Perceptron manages weightpotential ideasfluctuations within imagitron until solution pattern is obtained. Can be used to invent significance PERCEPTRONto raw sensory input patterns (i.e., sense making). Compound DAGUIs emulate juxtapositional invention, logical opinionsdeduction/induction, and create theories from analogy modules. US Patent 5,659,666, DAGUI (Creativity Machine Paradigm) 2010, Imagination Engines, Inc. 5. SELF-TRAINING ARTIFICIAL NEURAL NETWORK OBJECT (STANNO) No explicit, human-conceived trainingalgorithm.data patterns Trainer absorbed into trainee so as tocreate a monolithic neural net capableof autonomously learning. TRAINEE Class wrapper supplied to producemost efficient object-oriented neuralweight updatesnet in world.error patterns 10 million attribute networks enabledfor PCs and GPUs, completing trainingcycles on millisecond time scales. TRAINER At core of extremely advancedautomotive machine vision systems. Automates DAGUIs by allowing weight update strategy component nets to learn fromsuccesses and failures. US Patent 5,845,271, Non-Algorithmically Implemented ANNs 2010, Imagination Engines, Inc. 6. DEVICE FOR THE AUTONOMOUS BOOTSTRAPPINGOF USEFUL INFORMATION (DABUI) Untrained STANNO-based imagitronauxiliary sensory inputs generates potential idea. Untrained STANNO-based perceptron generates some figure of merit to it.STANNO-BASED IMAGITRONweight fluctuations / learning If figure of merit (opinion) is sufficient, reinforcement takes place in all networks.potential ideas If figure of merit (opinion) is insufficient, training is weakened in imagitron, with reinforcement learning occurring in perceptron. STANNO-BASED PERCEPTRON Bootstrapping is continued until ideas have matured. opinions Thereafter, derivative ideas may be generated on demand.US Patent 7,454,388, DABUI 2010, Imagination Engines, Inc. 7. SUPERNETS Any number of sensors and actuators may be tied together via an adaptive/creative synthetic brain. Myriad STANNO modules may interconnect into supernets involving self-forming group membership filters (GMF) and Creativity Machines. Various neural correlates to the human brain automatically form. Synthetic intelligence shapes itself in response to environmental and corporeal demands. We are skipping simulation of human brain and striving for trans-human level intelligence.US Patent 7,454,388, DABUI 2010, Imagination Engines, Inc. 8. CREATIVE ROBOT MASTERS SAND Main problem isaccumulatingmound of sandin front ofrobot. Robot developsits own squatand leapstrategy viaunderlyingDABUI system. Side-to-sidemotiondisperses sandaccumulation. Behaviordeveloped inless than twominutes, tabularasa. 2010, Imagination Engines, Inc. 9. CREATIVE ROBOT MASTERS SOIL Main problemagain isaccumulation ofsoil in front ofrobot. DABUI systemdevelops itsown strategy forcrawlingthrough pottingsoil, levelingmaterial withfront legs, whilethrusting withrear legs. Behaviordeveloped inless than aminute! 2010, Imagination Engines, Inc. 10. CREATIVE ROBOT LEARNS TO DREAD ROCKS Main problem isnon-deterministicbehavior of rockbeneath robot. DABUI systemdevelops astrategy ofstabilizationwithin the rockthrough in-place rocking,so that legspenetrate asdeeply aspossible. Garden rockdefinitely offersa highercrawlingimpedance thansand or soil. 2010, Imagination Engines, Inc. 11. CONTEMPLATIVE TERRAIN-SENSING ROBOTS Visionpathways,navigation fieldgenerators,creative motorcontrol, andpathoptimizationmodules knitthemselvestogether into atargetsingle Supernet. Robothigh impedancecontemplatesterrain, decidesrock patchupon path ofleast resistance,and selectsappropriategaits along wayto reach target,avoiding rockypatch. 2010, Imagination Engines, Inc. 12. CONTEMPLATIVE TERRAIN-SENSING ROBOTS When rockypatch isreplaced bysand, DABUI-based controlsystem picks a targetmore directroute towardthe target onleft. Project wasreborn asautonomouslow impedancerendezvous anddocking for sand patchNASA. Later, off-worldrobotsprototyped forNASA, utilizingtabula rasabehavioraldevelopment. 2010, Imagination Engines, Inc. 13. STIFF CHALLENGES / BOLD SOLUTIONS Retreat from Complexity Emphasize simplicity andelegance. Disruptive Technology Identify proper point of entry intoindustry. Academic Adaptation Treat university researchers asesteemed knowledge workers. Timid Incrementalismcortical imagitron Seek out the more visionary manin uniform. Poverty of Imagination Stimulate imagination via potential thalamic perceptronapplications and the viability ofmachine consciousness. 2010, Imagination Engines, Inc. 14. STIFF CHALLENGES / BOLD SOLUTIONS let a thousand flowers bloom Allowed IEI in door. There is vast duplication of effort. Grand Challenge / X-Prizes Small companies with limited budgets and vanguard technologies cant afford to drop everything. Too many simplifications create grand challenges overabundant competition. too simplified. Become engineering and not scientific competitions. Greatest Curse This is an unanticipated technology that does everything. We are drowning in the possibilities. There is hope in collaboration. 2010, Imagination Engines, Inc. 15. IEI DECADE (1997-2007) IN A NUTSHELL 2010, Imagination Engines, Inc.