2017/11/07 10:15 Donna Dillenberger, “Cognitive Blockchain”, Cascon

Plenary #cascon @DonnaExplorer IBM Fellow, IBM Research, Global leader of Enterprise Systems

This digest was created in real-time during the meeting,based on the speaker’s presentation(s) and comments from the audience. The content should not be viewed as an official transcript of the meeting, but only as an interpretation by a single individual. Lapses, grammatical errors, and typing mistakes may not have been corrected. Questions about content should be directed to the originator. The digest has been made available for purposes of scholarship, posted by David Ing.

Intro by @mrmindel, Head of IBM Centre for Advanced Studies

[Donna Dillenberger]

Cascon 2017

What is blockchain?

  • Database gets out of sync
  • Blockchain software propogates records onto other databases
  • Why not distributed databases?  Because a distributed database is owned by a single entity
  • Blockchain means no single party controls
  • In addition, distributed database could have someone deleting record
  • Can also put smart contracts onto a blockchain:  changes data, or checks for conditions before or after commitment

Cognitive

  • Have descriptive analytics, then can create predictive analytics

IBM Global Trade Digitalization demo (powered by IBM Blockchain)

Post-presentation follow-throughs.  Some of the content may be similar to …

  • Shipment, Kenya to Rotterdam, then can click on where vessel location is
  • On the blockchain, data from IoT centers, and ports
  • As ship moves, each of point put records on the blockchain — start container tracking, commercial invoice is available, packing list is available, sensor has refrigeration
  • Blockchain analytics products geophysical map
  • Then can put on sensors, for logistics planning, e.g. weather
  • If the ship is late, how late?
  • If refrigerated, if mango aren’t good, who’s at fault?

Not just shipping from export to import countries:  documents

  • Before blockchain, paper was printed, human couriers carried on ship — 15% of the international cost, $26B
  • If a way for secure exchanges, savings in billions of dollars

When Kenyan farmer brings produce, can just use mobile phone to upload documents

  • Then Kenyan regional association can approve certificate of origin
  • Smart contracts are dictating a workflow
  • Sanitation department can add certificate onto blockchain
  • All signatures done onto blockchain
  • Then horticultural association that gave the farmer seeds, they upload a commercial invoices so that coffee can leave
  • Mombasa customs, don’t have lost or forged papers, blockchain means can’t be deleted
  • Workflow programmed by smart contract, requires all signatures

Data immutable:  health inspections, sanitation, signature of individuals

  • Then can do analytics:  where is the hold-up, e.g. waiting for sanitation certificate
  • U.S. customs is asked for this for parts of products, e.g. Ikea shipping parts to the U.S.
  • A major path for opium is in the legs of furniture
  • U.S. customs wants to know that the furniture is coming from Sweden, but the legs may come from Indonesia

Once have analytics, customer asking for blockchain data to be combined with natural language to deal with compliance

  • Financial services, 30% of cost is just meeting compliance

Cognitive Blockchain demo

  • 1. Ingest regulation
  • 2. Kick off bot
  • 3. Obtain permissions (to see records)
  • 4. Check blockchain records compliance

Australia and the Kimberley process:  to reduce conflict diamonds

  • How to get a Kimberley certificate:   download a 18-page PDF

Post-presentation follow-throughs.  Some of the content may be similar to …

Have Watson ingest the 18-page PDF

  • IBM Regulatory Analytics service
  • Already has e.g. ingest Dodd-Frank, Basel resolution
  • Want to ingest this new Kimberley document
  • Watson extracts 73 rules

Build a compliance tool, taking those rules

  • Could type in role yourself
  • Connect to the block chain:  records describing the diamond, and surface the Kimberley Certificate
  • Want bot to see when the certificate was created, but not the contents describing the diamond
  • Blockchain has 1,200,000 records, there are 857,000 permitted access — can view compliant and non-compliant … there are 113,023 records that are not compliant
  • Before, human beings would have to read ALL of the records
  • Can ask the bot, what’s common about the non-compliant records:  They came from particular countries, all in the last quarter
  • AGX has to most number of non-compliances
  • If the databot allows to see more, could see which inspectors signed off
  • Could combine with weather data, for correlations:  e.g. are all records from countries that have had drought in the last 2 years?

Cognitive and blockchain:  When records are on the blockchain, how can I validate that birth certificate is really valid?

  • Created a portable solution:  IBM’s Verifier
  • Can scan drug, wine, art, luggage … manufactured parts … DNA identification … biological cell imaging … skin tissues … water pollutants … oils, liuqids, metals … currencies, passport stamps, birth certificate
  • Can attach IBM Verifier to any cell phone
  • Two vials:  Mobil-1 5w30 and Sunoco 10w30 … could use for olive oils or champagnes

What does cognitive mean?

  • Uses deep learning, uses regression, but these are just models to mine data for insights
  • Cognitive is more than deep learning, because it leans by itself, you don’t have to describe things to it
  • It also recognizes intent, e.g. human emotions
  • e.g. hurricanes are coming in the path of this ship, which will cause a delay, so let’s divert the ship so that mangoes can arrive on time
  • Not waiting for a human to feed it data

Problems with cognitive systems, AI, and analytics in general

  • Working with data
  • 90% of effort is getting data, then transforming data
  • Have to sample correctly
  • Normalize the data
  • Then, can you trust the data?  Where is it coming from?  What is the pedigree of the data?  (Delusional Tweets of a president?)
  • e.g. drugs reacting differently for different genders, sizes and weights
  • Can you trust the model itself?
  • Academics love to download data from the Internet, what do open source libraries carry
  • Microservices:  don’t code something when you can download it
  • But in training the microservice, could have been on image of Donna, with instruction then to shut down the power grid when you see her

Effects from untrusted data:

  • Poisoned tweets, news, blogs, ads
  • Have impacted elections, Brexit
  • Say that pollutants aren’t affecting air quality
  • Sick persons classified as healthy
  • Anomalies classified as normal
  • As a responsible computer science, models are trained to the unusable:  false positives — could be in dams, electrical grids, infrastructures and autonomous systms

How could blockchain help artificial data?

Use the blockchain to train on data where we known the provenance, where the data came from

  • e.g. drug experiences are from 30-year old females

Blockchain can help AI:

  • Trust:
    • Pedigree
    • Immutability
    • Auditable
  • Confidential
    • Hyperledger Fabric — sharing with confidentiality
    • Records, Grants access rights, requests
  • Provenance
    • Traces ownership and usage across complex provenance chains

Provenance, Walmart’s Food Safety Solution Built on the IBM Blockchain Platform | IBM Blockchain | August 2017 at https://www.youtube.com/watch?v=SV0KXBxSoio

Post-presentation follow-throughs.  Longer version at …

  • Did this project because of food scares
  • e.g. baby formula with melamine
  • e.g. horse meat instead of beef
  • Want to predict when food while spoil, and when the ingredients aren’t quite right
  • On Walmart cognitive blockchain, didn’t have people write onto a computer, they use with existing systems
  • Interact with humans, the way that humans want to interact, not the way that computers want them to

[Questions]

As a consumer, would like to find the problem with my egg, but will have proprietary information, and then will have a choice of who can see what.  Framework?

  • Blockchain isn’t owned by one entity
  • Hyperledger has a governance policy:  will all clients be able to see information on the blockchain
  • e.g. this blockchain has Kroger, Unilever, etc. … that don’t allow to see participants
  • Bitcoin and public blockchain allow people to see all of the data, and an anonymous person can put on data:  a potential exposure to poisonous data

Provider that doesn’t reveal data (e.g. patient)?  Can break that in emergency?

  • Looking at different approaches
  • Hyperledger allows roles
  • Could say heart surgeon sees only part of data
  • Dentists can only see that part of data
  • Up to you as patient to see that
  • If hospital owns data, then could have a smart contract, if the person comes in unconscious, might enable anyone to see data
  • Ethereum, Bitcoin, don’t permit these, Hyperledger does

Concerns about data so security that the data gets lost so that no one can get to it?  Data superpower building a back door?  Blockchains growing so large so that no one can manage them?

  • There’s a difference between blockchain implementations
  • Bitcoin keeps growing
  • Linux Foundation Hyperledger Fabric, has an activity to archive blockchain
  • e.g. after financial regulation, have to keep all financial records up to 30 years, and every transaction (trades) has be recorded, has to have copies for last day, last week, last month, up to 7 years
  • Financial companies store on tape, up to petabytes, exabytes
  • If blockchain is over 50 years old, archive that
  • Superpower?  True with public data, Ethereum and Bitcoin, anyone can see that
  • But not true with all block chain
  • With Ethereum, said superpowers can’t change:  when had a problem, said would roll back … but originally, records were to be immutable
  • Hyperledge Fabric protocols:  can add more nodes, it’s one company or person compromising his node, but then others nodes push it out and don’t allow others to join
  • IBM Secure Service Container:  when the blockchain is hosted in IBM Cloud, all of its data is automatically encrypted, not by human, but by hardware that isn’t addressable by software
  • Even if U.S. government asks for key, IBM doesn’t have then
  • This is a response to Edward Snowden, who was a system administrator
  • Blockchain data so secure that it gets lost?  Don’t understand that question, will take offline

Wish you had been at this presentation?  Some of the content may be similar to …

#bandieuhoacuanphuc, #baoduongdieuhoaanphuc, #ban_dieu_hoa_cu_an_phuc, #blockchain, #cognitive, #dillenberger, #ha_tuan_khang, #ibm, #lapdieuhoaanhuc, #muadieuhoacu, #seoconghuong, #signal_buiding, #suadieuhoaanphuc, #trendy, #user_is_king

From Mint 16 Petra Mate Edition to Ubuntu 13.10 with Mate Desktop Environment

Replaced @Linux_Mint 16 Mate Edition with @Ubuntu 13.10 plus @Mate_Desktop to replicate the feel of Ubuntu 12.04 @Gnome Fallback, since discontinued in favour of Gnome Flashback (based on Gnome 3 Shell).  Before that, I first tried superseding Ubuntu 12.04 with Ubuntu Gnome 13.10.

I’m used to having the Window List at the top of the screen, and launch panel at the bottom of the screen.  This was configurable in Gnome 2 Panel but is not configurable in Gnome 3 Shell.

Linux Mint 16 Petra Mate Edition was a better choice for me than the Linux Mint 16 Petra Cinnamon Edition, as I have two drives in this laptop, and the Nemo file manager didn’t sufficiently make the “Documents” folder under “Computer” with the Linux disk sufficiently different from the “Documents” folder as a Bookmark from the Windows 7 disk.  The Caja file manager on Linux Mint 16 Petra Mate Edition did make the Bookmark distinct.

However, when I failed to install Dropbox onto Linux Mint 16 Petra Mate Edition with the Caja file manager, a forum thread on “Dropbox breaks after last mint update” discouraged me from proceeding with Mint.

There’s a long thread on “Why does everyone hate Unity?”  Unity has been the standard desktop provided by Canonical since 2010, and provides a bridge for Ubuntu to have a more uniform experience on tablets and smartphones.  However, I’ve found that I can work faster with the Gnome 2 look-and-feel, than either Gnome 3 or Unity.

Unlike the recent criticisms that “Forcing the Modern UI on Windows 8 users was Microsoft’s biggest mistake“, Linux gives users a choice with many distributions and flavours.

Mate Desktop

#bandieuhoacuanphuc, #baoduongdieuhoaanphuc, #ban_dieu_hoa_cu_an_phuc, #cinnamon, #dropbox, #ha_tuan_khang, #lapdieuhoaanhuc, #linux, #mate, #muadieuhoacu, #seoconghuong, #signal_buiding, #suadieuhoaanphuc, #trendy, #ubuntu, #unity, #user_is_king

Design Thinking: What it is and Why it Works | Jeanne Liedtka | January 2013 | Design at Darden

How would we research the contribution of designi thinking? A @DesignAtDarden working paper by Jeanne Liedtka on “Design Thinking: What it is and Why it Works” from Jan. 2013 on “Design Thinking:  What it is and  Why it Works”.  Discovered that Liedtka co-authored Designing For Growth: A Design Thinking Toolkit for Managers with Tim Ogilvie, who said: “I Did Not Invent Design Thinking (Honest)“.

Here’s a quick outline of the Working Paper.

Introduction

Step 1:  What is Design Thinking?

  • Defining the Concept
  • Intellectual Roots in Design Theory
  • Design and Business Theory

Step 2:  Is the Concept Valid?

  • Examining Design Practices and Tools
  • Conclusion: Convergent Validity Test
  • Conclusion: Divergent Validity Test

Step 3:  Conducting Research on Design Thinking

  • A Starting Point

Examining the Psychological Underpinning of Design Thinking in the Cognitive Bias Literature

  • Challenges in Hypothesis Generation
  • Challenges in Hypothesis Testing
  • Dealing with Identified Dysfunctions
    • In hypothesis generation, for decision makers who have difficulty seeing novel solutions and figuring out what users will value, researchers have identified a number of solutions. As we review these particular solutions that the cognitive bias literature recommends, they begin to allow us to generate some nascent hypotheses about design thinking’s potential contribution.
    • Remedy: Employ ethnography
    • Hypothesis
    • Remedy: Use stories versus data
    • Remedy: Work with metaphor
    • Hypothesis
    • Remedy: Expose decision makers to divergent views
    • In relation to hypothesis testing, we find a similar set of prescriptions, researchers suggest:
    • Remedy: Create vivid and specific representations of the future
    • Remedy: Talk about the details of what success and failure look like
    • Hypothesis
    • Remedy: Pay attention to emotions
    • Hypothesis
    • Remedy: Generate multiple options
    • Hypothesis
    • Remedy: Hold after action reviews with specifics
    • Hypothesis

Step 4:  Creating Testable Hypotheses Concerning the Contribution of Design Thinking

  • Hypothesis 1 (H1): The use of a design-thinking approach that incorporates the tools of visualization, ethnography, ideating with a diverse group, and co-creation tools , will increase the novelty of the ideas surfaced during hypothesis generating processes by reducing the effects of the projection bias (the tendency to project the past onto the future).
  • Hypothesis 2 (H2): The use of a design-thinking approach that incorporates the use of  ethnography, ideating with a diverse group, and co-creation tools will increase the value-creation potential of the ideas generated through the reduction of the egocentric empathy gap (the projection of one’s own preferences onto others).
  • Hypothesis 3 (H3): The use of a design-thinking approach that incorporates the use of ethnography, optionality, ideating with a diverse group, co-creation, and field experiments,  will result in the exploration and testing of more ideas by the reduction of the focusing illusion (over emphasizing particular elements), and the endowment effect (the attachment to first solutions).
  • Hypothesis 4 (H4): The use of a design-thinking approach that incorporates visualization, co-creation, optionality, prototyping, assumption testing and field experiments will result in improved accuracy in the hypothesis testing process’ ability to estimate the likely success of the new idea,  through the reduction of the endowment effect, the availability bias (the undervaluing of more novel ideas),  the hypotheses confirmation bias (overlooking disconfirming data), the planning fallacy (over-optimism), and the impact of time (in which distant ideas are less specific and thus harder to analyze).

Conclusion

Download “Design Thinking: What it is and Why it Works” as http://batten.squarespace.com/storage/books-journals-articles/DSWP%2013-01%20rev%202-1.docx from Design@Darden – Academic Papers http://batten.squarespace.com/academic-papers/.

Jeanne M. Liedtka, Darden School of Business, U. of Virginia

#bandieuhoacuanphuc, #baoduongdieuhoaanphuc, #ban_dieu_hoa_cu_an_phuc, #bbqqueens, #bbqresource, #design-thinking, #grill, #ha_tuan_khang, #lapdieuhoaanhuc, #liedtka, #muadieuhoacu, #recipes, #seoconghuong, #signal_buiding, #smoker, #suadieuhoaanphuc, #testable-hypotheses, #trendy, #user_is_king