ibm systems cognitive systems
play

IBM Systems Cognitive Systems Dr. Wolfgang Maier Director HW - PowerPoint PPT Presentation

IBM Systems Cognitive Systems Dr. Wolfgang Maier Director HW Development IBM Research & Development wmaier@de.ibm.com 9/2017 IBM Research & Development Forschung Hardware-Entwicklung Software-Entwicklung Hardware- und


  1. IBM Systems Cognitive Systems Dr. Wolfgang Maier Director HW Development IBM Research & Development wmaier@de.ibm.com 9/2017

  2. IBM Research & Development Forschung Hardware-Entwicklung Software-Entwicklung Hardware- und Software-Entwicklung Watson Dublin Burlington Endicott Moskau Toronto Minsk East Fishkill Peking Poughkeepsie China Rochester Fujisawa Vancouver Tokio Boulder Yasu Beaverton Yamato Raleigh Santa Teresa Shanghai Foster City Taipei San Jose Almaden Costa Mesa Perth Tucson Gold Coast Austin Sydney Greenock Hursley Paris Böblingen Zürich Indien Bangalore La Gaude Krakau Pune Rom Haifa São Paolo Kairo

  3. Digitale Transformation

  4. Powered by data

  5. Photo - Lithography UV – Lithography 193 nm eUV – Lithography 13 nm

  6. Artificial Intelligence c i Output Input

  7. Artificial Intelligence neuromorphic True North ? Zeroth Spinnaker Architecture von Neumann CAL HTM CNN supervised unsupervised Training

  8. Neural Networks

  9. Stochastic Gradient Descent

  10. Quantum Computing

  11. Spiking Neurons

  12. Prominent Features of Spiking Neurons after Izhikevich

  13. Neuron Function • Emulation of analog behaviour by +/- 255 INT variable • 2-dimensional on-chip synaptic weighted network and off-chip packet based thru-neuron routing for multi-chip scaling • Update of Synaptic network every ms (logical / biological clock), internal processing ~ 1MHz • Neuron fires a spike (45 pJ) to the network if in the last update cycle a threshold was reached or exceeded • Stochastic and leak behaviour configurable Membrane potential for neuron j at time t 255 V j (t) = V j (t-1) + S A i (t) * z ij * [(1 - b j ) * s j + sign(s j ) * b j * F( | s j | , q j )] + Leak i=0 Leak +1 V j (t) = V j (t-1) + * [(1 - c j ) * k j + sign(k j ) * c j * F( | k j | , q j )] ( ) 0 -1

  14. “ Universal Cortical Engine “ Sparse Distributed Representations (SDR) OUTPUTS: Predictions Context Stable Concepts (SDR) Motor commands INPUT: Spatial-temporal data streams of any kind

  15. Find semantic similarities of words in Wikipedia 100K “Word SDRs ” Document corpus (e.g. Wikipedia) 128 x 128 Apple Fruit Computer runners up were Macintosh Microsoft = minus minus minus minus Mac Linux Operating system …. 17

  16. IBM Watson System Specifications IBM Technology Depth 2880 Processing Cores Content Analytics 90 IBM P750 Servers Business Analytics 16 Terabytes Memory (RAM) – 20TB Disk Big Data 80 Teraflops (80 trillion operations per second) Databases / Data Warehouses Workload Optimized Systems

  17. Synapse Hardware TrueNorth Technology 28nm Year 2012 Transistor 5.4 billion Count Power 0.05W

  18. Intelligence

  19. Errorfunctions C ro s s -E n tro p y E rro r (C ) d u rin g tra in in g T es t-E rro r d u rin g tra in in g

  20. Quantum Computing

Recommend


More recommend