multiparadigmd atas cience - swiss re group | swiss reconrad+wolfram.pdf2009:c omputationm eetsk...

Multiparadigm data science:AI, analytics and answers

Multiparadigm data science: AI, analytics and answers | 1

More data ⟶

better decisions?

2 | Multiparadigm data science: AI, analytics and answers

Problem: Information overload


Bigger problem: personal + systemic data


Today’s data answers challenge:

Available ⟶ Accessible?(esp. Disparate ⟶ Combined)Averaged ⟶ Personalised


Key solution:

Computation for Everyone (⟹ smart automation)Everyone for Computation (⟹ computational thinking)


AI, data Computation for Everyone: ImageIdentify »Dynamic[img = CurrentImage[];imgassembled = Magnify[Column[{Magnify[img, 2], ImageIdentify[img]}], 2]]

��


How to achieve:(highest-level, insightful)Computation for Everyone?


Review:what’s worked well for computation?

Traditionally computation domains

Scaling-up pencil and paper

Well-structured models

Simple-to-specify questions of data


Review:what’s still to develop?

Fuzzy questioning

Rapid deployment of new ideas

Communicating with computation

Utility of modern (non-rectangular) data structures


Need Multiparadigm approachfor computation, for data science »

ie. Let the job lead the toolset;Data Science toolset ≠ Statistics toolset


http://mpdatascience.com

Computation for Everyone:Start with interface; ....work backwards


Example: linguistic access


Dynamic[CurrentImage[]]


Example: knowledge apps��

��

��

��

��

��

�� /�� /�

��

��

��

��

��

♡

�� % ��

��

�� - ��

��

- ��

��

6.0

4.6

0 10 20 30 40

��

��

- �� % ��

��

- ��

��


Example: interactive report(eg. computable patient record) »


Different processes of data automationSearch (✓ powerful for eking out existing info, × generating new knowledge, quantification)

AI curation (✓ work from existing data, × promise yet realised)

Computational (✓ new knowledge, insights, × data needs to be in shape)


Build-in a Computable Data Layer


Shi� in data qualityCurrent: for human readabilityFuture: for “citizen” computation


Example: immediate data computation�� GenomeLookup["CTCTCTAACTAAACT"]

�� {{{Chromosome1, 1}, {108939073, 108939087}},{{Chromosome1, -1}, {138309610, 138309624}},{{Chromosome5, -1}, {139640264, 139640278}},{{Chromosome8, 1}, {72019948, 72019962}},{{Chromosome9, 1}, {110092060, 110092074}}}


Knowledge assets:Wolfram’s »


http://reference.wolfram.com/language/guide/LifeSciencesAndMedicineDataAndComputation.html

Knowledge assets:Industry (eg. insurance, pharma)


Knowledge assets:Health system—finance (Public or Private)


Knowledge assets:Publicly funded R&D


Knowledge assets:Clients’/Patients’


Data Repository »


https://datarepository.wolframcloud.com/category/Health

What’s for the human,What’s for the computer ?

Computer role: crunch data ⟶ suggest insightHuman role: be the boss ⟶ manage the

human-computer team;

Example: Patient Liver Spreadsheet »


http://www.wolframalpha.com/input/?fileinput=%2FWolframAlphaProUploads%2FIndian+Liver+Patient+Dataset+(ILPD).csv&i=+&fileinputtype=file

Example case:Management Decisions:instinct or analysis?

Eras of management: instinct, simple targets, agile-targets?


Example use:Should Texasbuy hurricaneinsurance? »


Barriers to ubiquitouscomputation workflows

Computational Data

Technology EcosystemInterfaces (all levels)Ubiquitous deploymentComputational powerCoherence

Maths/Coding/Computational Thinking Education


Wolfram: infrastructure for injecting

computation everywhere


Multiparadigm Data Science Examples: Thrust SSC / Bloodhound »Biovotion (quick cut) »


Key unification technology:symbolic expressions


Key Technology: Wolfram Language

LiterateMultifaceted but unifiedPrinciple: human trumps computerAlgorithms included


Key Technology:Enterprise Private Cloud »


http://www.wolfram.com/enterprise-private-cloud/

Example Workflow: Enterprise Private Cloud »


https://www.wolframcloud.com/objects/dvs/wepc/wepc-finance.html

Enterprise Computation: Why?

Prevent knowledge silos

Reduce code replication and liability

Manage valuable computational assets; centralised quality

Control your Data Security

Horizontal and Vertical Integration

One codebase, multiple interfaces eg. Excel

Key component of meeting regulations...

“Accurate computation” cf. Excel...


Departmental siloing: Same computations,different departments?

��

Time to build Enterprise Computation...


Paradox: change education

computerbasedmath.org: rebuild the maths curriculum assuming computers exist »


http://www.computerbasedmath.org

Computational Thinking Process1.

2.

3.

4.


Example Module:Should I Insure My Laptop? »


History so far...1988: Maths (Mathematica)1990s: Maths ⟶ Computation2000s: Prototyping ⟶ Complete Workflow/Deployment2009: Computation meets Knowledge (Wolfram|Alpha; Apple’s Siri, 2011)2011: Computation meets Documents (CDF)2012: Finance Platform2012: Wolfram Technical Services2014: Wolfram Language (Programming Cloud)2016: Wolfram Enterprise Private Cloud2017: Multiparadigm Data Science (MPDS)


How will you:Empower the knowledge consumerto generate usable knowledge?


What’s yourEnterprise Computation Strategy?

www.wolfram.comwww.wolframalpha.comwww.conradwolfram.com

@conradwolfram


Basic Copyright Notice & Disclaimer

©2017 This presentation is copyright protected. All rights reserved. You may download or print out a hard copy for your private or internal use. You are not permitted to create any modifications or derivatives of this presentation without the prior written permission of the copyright owner.

This presentation is for information purposes only and contains non-binding indications. Any opinions or views expressed are of the author and do not necessarily represent those of Swiss Re. Swiss Re makes no warranties or representations as to the accuracy, comprehensiveness, timeliness or suitability of this presentation for a particular purpose. Anyone shall at its own risk interpret and employ this presentation without relying on it in isolation. In no event will Swiss Re be liable for any loss or damages of any kind, including any direct, indirect or consequential damages, arising out of or in connection with the use of this presentation.