@pieroleo Computing, Data, Algorithms, and, Problem Solving Pietro Leo Executive Architect - IBM Italy CTO for Artificial Intelligence IBM Academy of Technology Leadership Head of IBM Italy Center of Advanced Studies A guest lecture prepared for Information Technology & Digital Strategy @pieroleo www.pieroleo.com
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
@pieroleo
Computing, Data, Algorithms, and, Problem SolvingPietro LeoExecutive Architect - IBM Italy CTO for Artificial IntelligenceIBM Academy of Technology LeadershipHead of IBM Italy Center of Advanced Studies
A guest lecture prepared for
Information Technology & Digital
Strategy
@pieroleo
www.pieroleo.com
@pieroleo
@pieroleowww.pieroleo.com
@pieroleo
ALGORITHMSMy dentist told me how to perfectly clean your teeth…..
Data is there and we need to make the best out of it@pieroleo
www.pieroleo.com
@pieroleo
We produce and consume Data for a specific purpose@pieroleo
www.pieroleo.com
@pieroleo
Source: http://datacoup..com
Value of Data
Pietro Leo's SecondIncome!
@pieroleowww.pieroleo.com
@pieroleo
For Science, Big Data is the microscope of the 21st century
Wine DNA Tracing
@pieroleowww.pieroleo.com
@pieroleo
Source: Cornell University - Maize kernal infected with Aspergillus flavus, which producedaflatoxin.http://www.plantpath.cornell.edu/labs/milgroom/Research_aflatoxin.html And http://www.special-clean.com/special-clean/en/mold/mold-lexicon-1.php
For science, Big Data is the microscope of the 21st century
The largest accommodation company owns no real estates
@pieroleowww.pieroleo.com
Marriott CIO acknowledged Airbnbas a new competitor
@pieroleowww.pieroleo.com
35
You shared your position with me and can guess your mobility need. I can take you where you need to be
Just enjoy your new experience. Stay safe as in your home
I know what is needed for you, even before you order it
Please, come with me and stay by me.I know your content I can take care of all your digital life
@pieroleowww.pieroleo.com
Automation will bring big shifts to the world of work, as AI and robotics change or replace some jobs, while others are created.
Source: McKinsey & co: Jobs Lost, Jobs created: workforce transactions in a time of automation, 2017
By 2030, 75 million to 375 million workers will need to switch occupational categories.
60% percent of occupations have at least 30% of constituent work activities that could be automated
@pieroleowww.pieroleo.com
Leveraging the Explosion of Data in Medicine An Impossible Task Without Analytics and New advanced Artificial Intelligence Computing Models
1000
Factsp
erDecision
10
100
1990 2000 2010 2020
HumanCognitiveCapacity
ElectronicHealthRecords(ClinicalData)
InternetofThings(Exogenous Data)
TheHumanGenome(GenomicData)
CapturingtheValueofData:BigChangesAhead
Medical error—the third leading cause of death in the US
Source:BMJ 2016; 353 doi: http://dx.doi.org/10.1136/bmj.i2139 (Published 03 May 2016) Cite this as: BMJ 2016;353:i2139
38
Body Mass Index (BMI)
Mass (weight - Kg) / height (cm) x height (cm)
You are “Normal” if your BMI is between18.5 and 24.99
Adolphe Quetelet, 1832@pieroleo
www.pieroleo.com
39
Practice Pearls:• BMI - Body mass index is a strong and independent risk factor for being diagnosed with type 2 diabetes mellitus• Type 2 diabetes risk may be incrementally higher in those with a higher body mass index• Understanding the risk factors helps to shorten the time to diagnosis and treatment
How precise could be a “simple” signal
@pieroleowww.pieroleo.com
Assistant
Tools
Collaborator
Coach
Mediator
Emerging types of Cognitive Systems
Augment Decision Making is opening to new forms of collaboration between humans and machines
to solve problems
@pieroleowww.pieroleo.com
Assistant
Tools
Collaborator
Coach
Mediator
Augment Decision Making is opening to new forms of collaboration between humans and machines
Emerging types of Cognitive Systems @pieroleowww.pieroleo.com
86%Accuracy
Assistant
Tools
Collaborator
Coach
Mediator
Augment Decision Making is opening to new forms of collaboration between humans and machines
Emerging types of Cognitive Systems @pieroleowww.pieroleo.com
@pieroleo
2011 2015
2016 - AlphaGO=4 Lee Se-Dol=1
1997 - IBM=2.5 Kasparov=2.5
1997
AlphaGO uses self-trained net to evaluate positions and moves on 30M historical games
DeepBlue uses a hard-coded objective function written by a human coupled with High Performance Computing
2016
10
10170
1040
Applying or having wisdom in real world is not only an AI game
COMPUTING & MATH WISDOM
IBM Watson – Jeopardy!
SEMANTICS
Watson OncologyA collaboration between IBM and Memorial Sloan Kettering (MSK). Watson for Oncology utilizes MSK curated literature and rationales, as well as over 290 medical journals, over 200 textbooks, and 12 million pages of text to support decisions.
• Analyzes the patient's medical record• Identifies potential evidence-backed treatment options• Finds and provides supporting evidence from a wide variety of sources
46
TheMedicalSieve § Build a fast anomaly detection engine
• Insulin, Meal, Activity, Context• Standalone, for MDI patients
• Watson Health Cloud & Analytics• Pattern recognition, insights &
Predictions• Engagement & Gamification
• Real-time insights, coaching platform
• Assists in daily diabetes mgmt• Aggressive capabilities
roadmap
Insights GlucoseForecasts
Hype-hyperPredictions Ask
Watson
@pieroleowww.pieroleo.com
Sugar.IQ app
@pieroleowww.pieroleo.com
Assistant
Tools
Collaborator
Coach
Mediator
Types of Cognitive Systems
Augment Decision Making is opening to new forms of collaboration between humans and machines
@pieroleowww.pieroleo.com
ViTA Advisor: it is a conversational multi-modal agent to support older as well as a tool to collect meaningful data about the context of an individual
ViTA : Virtual Trainer for cognitive impaired patients
Sustain Independence and Dignity with affect and purpose, preserve and reinforce individuals and social memories
Vita Memory Coach: a system that supports caregivers to collect meaningful facts and memories of an individual and his context
Vita
Memories
@pieroleowww.pieroleo.com
Memories are a bridge among generations
12/15/17Tales@pieroleo
www.pieroleo.com
@pieroleo
Source: Cognitoys @pieroleowww.pieroleo.com
Assistant
Tools
Collaborator
Coach
Mediator
Augment Decision Making is opening to new forms of collaboration between humans and machines
Emerging types of Cognitive Systems @pieroleowww.pieroleo.com
Dinner Planner
@pieroleowww.pieroleo.com
Assistant
Tools
Collaborator
Coach
Mediator
Augment Decision Making is opening to new forms of collaboration between humans and machines
Emerging types of Cognitive Systems @pieroleowww.pieroleo.com
Source: IBM Research automatic sport highlights generation https://www.ibm.com/blogs/research/2017/06/scaling-wimbledons-video-production-highlight-reels-ai-technology/ @pieroleo
www.pieroleo.com
@pieroleo
71
Source: IBM Research Food Recognition - https://www.ibm.com/blogs/research/2017/05/training-watson-see-whats-plate @pieroleowww.pieroleo.com
@pieroleo
72Source: IBM Research Image Caption generation paper - https://arxiv.org/pdf/1612.00563.pdf
“a blue boat is sitting on the side of a building” “a person holding a giraffe in their hand”
ConversationIntegrate diverse conversation technology into your application.
KnowledgeGet insights through accelerated data optimization capabilities.
VisionIdentify and tag content then analyze and extract detailed information found in an image.
SpeechConvert text and speech with the ability to customize models.
LanguageAnalyze text and extract meta-data from unstructured content.
EmpathyUnderstand tone, personality, and emotional state.
Practical exercise: explore how an AI set of basic services looks like
@pieroleowww.pieroleo.com
@pieroleo
ALGORITHMS PROBLEMSDATACOMPUTING
• QUANTUM LAWS
• DIGITAL
• ANALOG
High Complexity
• WISDOM
• KNOWLEDGE
• INFORMATION
• NUMBERS
• TALK WITH A DIGITAL HUMAN
• DGITAL TRANSFORMATON
• ASK FOR A LOAN
• 2+2=4
• MACHINE LEARNS WITHOUT A TRAINING
• SUPERVISE AND TRAIN A MACHINE
• PROGRAM
LowComplexity
Dimensions of IT systems and their complexity
@pieroleowww.pieroleo.com
@pieroleo
“Nature isn’t classical, dammit, and if you want to make a simulation of nature, you’d better make it quantum mechanical, and
by golly, it’s a wonderful problem, because it doesn’t look so easy.”
-RichardP.Feynman
NATURE ISN’T CLASSICAL, DAMMIT, AND IF YOU WANT TO MAKE A SIMULATION OF NATURE, YOU’D BETTER MAKE IT QUANTUM MECHANICAL, AND BY GOLLY, IT’S A WONDERFUL PROBLEM, BECAUSE IT DOESN’T LOOK SO EASY.”
RICHARD P. FEYNMAN
“
QuantumAlgorithmsA quantum computer can solve certain problems much faster.
Deck with99 black cards
1 red Ace Problem:find red Ace
Average-case scenario:
Classical Algorithm: 51 card draws
Grover Quantum Algorithm: 10 card draws
@pieroleowww.pieroleo.com
QuantumAlgorithms
A quantum computer can solve certain problems much faster.
This laptop could simulate a 25 electron system, Titan a 43 electron system but no classical computer ever built could simulate a 50 electron system exactly.
Caffeine is a moderately sized molecule. It is more complex than water, but simpler than DNA or proteins.The complexity involved to simulate its energy, structure and interactions is beyond the capability of any computer with current technology.To simulate Caffeine on Quantum might take 160 Qubits, doing so with classical with current methods require around 10^48 bits (as there are atoms on planet Earth)
@pieroleo
Some of the earliest examples of these include optimization, chemistry, and machine learning.
Quantum Computing holds the potential to solve today’s intractable computational challenges
@pieroleo
How quantum mechanics laws (such as superposition, entaglement, interference, etc..) could help to build new computational methods to solve these problems more efficiently?
@pieroleowww.pieroleo.com
@pieroleo
0
1
QuantumQUBIT
‘1’
‘0’
‘0’ + ‘1’
ClassicBITS
Superposition:eachqubit in2statessimultaneously
How Quantum laws can speedup Information representation?
@pieroleo Whereas a 56-qubit quantum computer can theoretically perform 2^56 operations simultaneously, IBM's accomplishmentinvolved dividing this task into 2^19 slices that each essentially consisted of 2^36 operations. This strategy meant the researchers only needed about 3 terabytes of memory for their simulated quantum computer.
In contrast, earlier in 2017, a 45-qubit simulation at the Swiss Federal Institute of Technology in Zurich required 500 terabytes of memory. @pieroleo
Pietro LeoExecutive Architect IBM Italy CTO for Artificial IntelligenceHead of IBM Italy Center of Advanced StudiesIBM Academy of Technology Leadership