© 2012 international business machines corporation wolfgang singer, honorary member of ibm...

© 2012 International Business Machines Corporation

Wolfgang Singer, Honorary Member of IBM IT-Specialist [email protected]

IBM Technology Focus for the Next 10 Years

ADLUG Annual Meeting, Florence, Italy September 19-21, 2012

Intellectual Property and Standards

© 2012 IBM CorporationIBM Patent Leadership2

USPTO Top 10 Patentees – 2011 IBM has the patent leadership

for the 19th consecutive year.

Hundreds of patents donated.

IBM has the patent leadership

for the 19th consecutive year.

Hundreds of patents donated.

© 2012 International Business Machines Corporation 3

IBM Research: The World is Our Lab

China

WatsonAlmaden

Austin

TokyoHaifaZurich

India

Dublin

Melbourne

Brazil

IBM Research labs

Labs added since 2010

Other IBM Research presence


Leading IBM: Eras of computing

Co

mp

ute

r In

telli

gen

ce

Time

Tabulating Systems Era

Programmable Systems Era

Cognitive Systems Era

5 © 2012 IBM Corporation

Nano Devices

1 Billion Transistors

Power7 chip

Workload Optimized

Systems

Data

Compute+ Natural Language+

Analytics

Deep Q&A Computers

Exascale (Datacenter-in-a-box) Massive parallelism Flexible system

optimization

1 Trillion Devices

Nano Systems (Systems-on-a-chip)

Cognitive Computing “Synapse” devices

DNA Transistor Nano Medicine

BIG/Fast Data Exa / zettabytes Milli / microseconds Unstructured Noisy

Four Technologies that Will Change the World


Nano Devices

1Billion Transistors

Power7 chip

Workload Optimized

Systems

Data


Analytics

Deep Q&A Computers


optimization

1Trillion Devices





From Nano Devices to Nano Systems

IBM Research - Zurich

[email protected] © 2012 IBM Corporation

Nanotechnology – Everywhere

Scratch resistive Lacquer

Laser Etching

Self-cleaning paints and Textiles

Regenerative Medicine

Photovoltaics

Filtration und purification

Nano-Pore

DNA

DNA Sequencing


Medicine at the Nanoscale

13

Anti-bacterial Nanoparticles

DNA TransistorDNA Transistor Fast, cost-effective method to use

genetic info in healthcare

DNA strands pass through ‘nanopore’

Electric sensor reads genetic information

Detects and destroys antibiotic-resistant bacteria

Fights infectious diseases

Biodegradable


Nano Devices


Power7 chip

Workload Optimized

Systems

Data


Analytics

Deep Q&A Computers

Exascale (Datacenter-in-a-box) Massive

parallelism Flexible system

optimization

1Trillion Devices





From Petascale to Exascale


1 PetaFlop = 1/3 rack

Software

Phase Change Memory

CPU Silicon Photonics

3D

1 PetaFlop 72 BG/P Racks

Overall Performance = 1000X Performance / watt = 135X Performance / $ = 1000X Footprint = <2%

Referenced to one-petaflop system

The Next Ten Years

2009

Exascale ComputingA billion calculations in a billionth of a second

2019

15


Progress in the Programmable Systems Era

Scale out Scale out

Scale up

Scale in


Scale In

A new paradigm for fast systems

Memory

Switching

Storage

Processor

13

© 2012 IBM Corporation17

24- to 48-hour forecasts

1 - 2 km resolution

3 hours to 3 days lead time

Weather data coupled with analytics

Natural Disaster Prediction and Response Deep Thunder

14

© 2012 IBM Corporation

Nano Devices


Power7 chip

Workload Optimized

Systems

Data


Analytics

Deep Q&A Computers


optimization

1Trillion Devices





From Big Data to Big Analytics


IBM DeepQA 100s of gigabytes for Deep Analytics 3 seconds per decision

Big Data – Very Big and Very Fast

Merging real-time analytics with deep analytics

Meaning from text at record volumes and speed

Massive scale analytics with minimal code Super fast indexing of thousands of facets Optimized and integrated system

Telco Promotions100k records per second, 8 billion per day10 milliseconds per decision270 terabytes for Deep Analytics

Homeland Security600k records per second50 billion per day1-2 milliseconds per decision320 terabytes for Deep Analytics


Big Data: Why we must move to a new era of computing

Dimensions of data growth

Terabytes to exabytes of existing datato process

Structured, unstructured, text, multimedia

Streaming data,

milliseconds to seconds to

respond

Uncertaintyfrom inconsistency,

ambiguities, etc.

VarietyVariety

VolumeVolume VelocityVelocity

VeracityVeracity

Uncertain data on the rise

2010

Vo

lum

e in

Exa

byt

es

Sensors & Devices

VoIP

9000

8000

7000

6000

5000

4000

3000

Enterprise Data

Social Media

2015


Forecasting a hurricane(www.noaa.gov)

Fitting a curve to data

Model UncertaintyAll modeling is approximate

Process UncertaintyProcesses contain

“randomness”

Uncertainty arises from many sources

Uncertain travel times

Semiconductor yield

Intended Spelling Text Entry

Actual Spelling

GPS Uncertainty

??

?

RumorsContaminated?

{John Smith, Dallas}{John Smith, Kansas}

Data UncertaintyData input is uncertain

Ambiguity

{Paris Airport}Testimony

Conflicting Data

??

?

18

© 2012 IBM Corporation

Nano Devices


Power7 chip

Workload Optimized

Systems

Data

Compute+ Natural

Language+ Analytics Deep Q&A

Computers


optimization

1Trillion Devices





From Programming to Systems that Learn


Cognitive Systems: A new era of computing

Programmable Systems Era

Processor-centric Fixed calculation Scale up/out Manual systems management


Data-centric Statistical analytics Scale in Automated systems/ workload management


Exploring the future of cognitive systems


Core Technology

Applications

ArchitecturesFar-

reaching research

Fully Homomorphic Encryption

21

Fully homomorphic encryption is a privacy enabling technology

Allows encrypted user data to be processed without the server knowing the content

Results returned to authorized user for decryption

Privacy-enhanced cloud services

Craig Gentry, a 35-year-old IBM researcher, solved this 30-year cryptographic problem

2010 ACM Distinguished Dissertation Award2010 Best Paper Award – IACR Crypto2010 Privacy Enhancing Technology Award2009 Privacy Innovation Award from the Intl.

Association of Privacy Professionals


Cognitive Systems: Atomic storage

Atomic limits of magnetic storage

96 iron atoms storeone byte of data

IBM Research explores the boundaries of science and technology


Cognitive Systems: Quantum computing

Extraordinary capabilities are expected…factoring a 3,000 digit number 1040 faster than today

Silicon chip with three qubits

Superconducting qubits use established manufacturing techniques for silicon technology

Potential for dramatic advances in cryptanalysis, database sorting, pharmaceutical research

IBM is the world leader in this technology

http://www.computerworld.com/s/article/9224670/IBM_touts_quantum_computing_breakthrough


Cognitive Systems: Applications

Sensor Networks

Social Business Smarter Cities

Watson

Human and knowledge capital analytics Optimize sales through insights

about sellers, teams, deals, clients, and performance.

Social / Demographic

Economic

Environmental

Technological

Feeder line

Intelligent building

Electric car

Substation

Transmission line

Automated meters

Millions of end points

100K+ elements,10ms latency

Multiple feedback time-scales

John Kelly: In 2014 Watson could be able to fit in your hand ... tremendous opportunities await.


LearningUnderstanding Interacting Explaining

Specific Questions

The type of murmur associated with this condition is harsh,

systolic, and increases in intensity with

Valsalva

From specific questions

to rich, incomplete problem

scenarios(e.g. EHR)

Rich ProblemScenarios

Entire Medical Record

Question-In/Answer-Out

Evidence analysis and look-ahead,

drive interactive dialog to refine

answers and evidence

Interactive Dialog Teach Watson

Refined Answers, Follow-up Questions

Input, Responses

Dialog

Batch Training Process

Scale domain learning and

adaptation rate and efficiency

Continuous Training& Learning Process

Answers, Corrections, Judgements

Responses, Learning Questions

Precise Answers& Accurate Confidences

Move fromquality answers

to quality answers and

evidence

ComparativeEvidence Profiles

Taking Watson beyond Jeopardy!


Cognitive Systems: SyNAPSE

“Neuron” and “synapse”-like computing model

Systems learn through analytics / experience

Advantages: Ultra energy-efficient, flexible, learning

Wiring diagram – monkey brain

‘True North’ Character ID

Learning Pong

© 2012 international business machines corporation wolfgang singer, honorary member of ibm...

Documents