Adam Gartenberg's Blog

Business Analytics and Optimization, IBM and Social Marketing

Recap: Insight 2014 Day 2 General Session


There were some great demos in today's Insight general session (and this technology is always better in live demos than pre-recorded ones, isn't it?  I've included the rough time stamps should you want to fast forward the replay once available to just see the demos.)  There was also a lot shared about how Watson technology is being applied to help oncologists, police investigators, and even chefs.  (See the link below for a chance to become a beta tester for Chef Watson).  We also heard first hand from Terry Jones, the founder of Travelocity and co-founder of Kayak, who is now starting a new venture, Wayblazer, built on Watson, and got a live demo geared around the Internet of Things.

Here are my notes from today's general session, and as a reminder, head over to the Insight GO site to watch tomorrow's session with Kevin Spacey and Captain Phillips (in the flesh, not Tom Hanks).

Mike Rhodin, SVP, Watson Group

How do we think?  Each of us in this room are experts in our own fields... but we're only humans.  We need help.  We're overrun by information.  Need analytic capabilities that turn big data into information, and then turn that information into knowledge.  Imagine a tool, coach, concierge that can help us see the world in new ways.  Strip out bias we have ingrained in our own cognitive frameworks.  Make decisions based on facts and research of others.  Cognitive computing can help us do that.

What makes Watson cognitive?

  • Understands and operates in natural language - just as if having conversation with another human being.  Watson understands the request in the complete context of the question.
  • Makes evidence based recommendations with clear levels of confidence.  No guesswork, just unbiased facts.
  • Not bound by memory or format.

Announcing new portfolio of Watson applications.
Designed for specific use cases, specific industries, users.
  • IBM Watson for Oncology:  To help doctors provide tailored cancer treatment
  • IBM Watson for Wealth Management:  Personalized financial advice to clients
  • Chef Watson: Spurring innovation in culinary arts

Can expect to see many more coming for the fields of legal, healthcare, finance, education and many more from IBM and partners

Demo:  Watson Oncology - Mark Megerian
  • Will demo a working, live system.  Partnered with Memorial Sloane Kettering on this solution, bringing together the cognitive power of Watson and expertise of Memorial Sloane Kettering.
  • Example:  67 year old female with stage 2 breast cancer.
  • Watson has access to all of her history and notes.  Oncologist can go through and look at summary extracted from those notes by Watson.  First, have to verify all the info that's been collected.
  • Then, Watson has taken all knowledge in Memorial Sloane Kettering and ranked/scored treatment plan options, along with supporting evidence all provided in one place.  (For example, imagine doctors walking from patient to patient with all of this information available in front of them on an iPad as they walk around.)
  • It's as if any doctor in the world can call up the world-renowned specialist at Memorial Sloane Kettering that focuses on this particular form of cancer and take advantage of their experience, along with all the information shared in thousands of journals, etc.
  • Watson Oncology also brings in the eligibility for clinical trials based on that patient's criteria, something much more difficult to do manually.
  • Watson then will display drug info for counter-indications and side effects to see which matches the patient.
  • The clinician makes the judgement based on this information.  Build from there a customized plan for this patient.

IBM Watson Discovery Advisor
Unveiling hidden connections and leading down new lines of thinking, learning language of your industry or profession.

Demo:  Anne Peterson - How Watson can work fighting crime (~25 min into session)
Watson and Identity Analytics
How investigator interacts with Watson to find patterns, quickly derive insights from data
Traditionally, investigators spent 80% of time sifting through data and only 20% analyzing data.  Want to be able to turn those percentages around.

ex, modeled on TV show Breaking Bad.  Investigating a new drug on the street:  Blue Meth
  • First, look into own data repository - data that others have put in.  Have found arrest for Jesse Pinkman associated with it.  Also have a person named Heisenberg, but don't know his real name or more about him.
  • Could send agents back out on street, but time consuming.  Could sort through more info, look through notes, etc., but also time consuming.  Or could interact with Watson.
  • Will ask Watson to look through notes, public records, etc.
  • Get back list of suggestions with probabilities.  It's about generating new leads and possibilities.  Can see why Watson surfaced each of these possible answers.
  • e.g., brought back Physicist, but with only 2% probability.  Walter White is the top result - 80% probability.  
  • Shows a visualization of all the information pulled on Walter White, e.g., spotted at lab, has lung cancer, etc.  Trying to get connection back to Jesse Pinkman, so add him into the picture, and the visualization expands to show the linkages between the two.

Demo:   Dr. Steve Abrams - Chef Watson
Watson is working with the Institute for Culinary Education and Bon Appetit Magazine

Chef Watson uses as input:
  • 9,000 kitchen tested recipes from Bon Appetit
  • Which ingredients work well together, which associated with different types of cuisine
  • Food chemistry
  • Psychology of what people find pleasant and not pleasant
  • Can put in ingredients want to use and ingredients that want to avoid
  • Will suggest a list of things that might work well
  • Will suggest styles of cuisine to choose from
  • Generate list of possible dishes, ranking from more common to less.
  • App is in beta, thousands of people taking through paces and helping to refine it.
  • Can try it by signing up for beta at bonappetit.com/chefwatson.
  • Have created cookbook:  Cognitive Cooking with Chef Watson, which will be available later this year.


Terry Jones - Founder of Travelocity, Cofounder of Kayak
New venture:  Wayblazer

Back when started Travelocity, information was only flowing one way.  But then users started creating content, and became two-way street.  Sites like Trip Advisor, with 80 million reviews.  And getting data on Facebook, Twitter, Instragram, YouTube, etc.

Travel can and should be fun, but travel planning isn't as much fun.  The average consumer searches over 20 sites to plan their trip, because what's missing is expert advise.  The web doesn't give advise.  Today, it comes from books, articles, magazines, reviews, Facebook.  But the #1 source remains our friends, because online everything has to fit in a box (e.g., reduced down to a # of stars).

Take structured, social, demographic, historic, location, deals data - and give it to Watson to create a totally new product that understands natural language, can provide confidence levels, gives insider insights, and then commerce from those insights.  That's Wayblazer.

First customer is Austin visitor's bureau.
You can tell it that you're heading to Austin with three guys, and will get recommendation for Formula 1 Grand Prix, a UT football game, and Austin City Limits... and that you should sit on turn 15 at the Grand Prix because that's where the action is.  Will not only recommend a particular food truck... but can also tell you to ask for the secret menu.  It will recommend that you stay at the Omni... because that's where the fans all stay.

Rhodin:
APIs:  Now have 8 services/APIs, scaling to more in coming months
Content:  Introduced Watson Content Store to provide info developers need to jumpstart their applications
Skills:  Stanford, Berkeley, Carnegie Mellon, NYU, U of Michigan, UT Austin, Univ of Toronto, among others.  Classes started this fall.  By next year, expect over 100 schools around the world.

Announced Watson on BlueMix a few weeks ago, and already have 1500 developers working on it, making more than 1M API calls.

Watson Solutions:
  • Watson Analytics - As announced yesterday
  • Watson Explorer - 360 degree view into a company's information to make better decisions
  • Watson Curator - Locating and curating content for Watson is frequent question.  Addresses question of how you update the information while keeping a clean audit trail.
  • Watson Engagement Advisor 2.0 - with cognitive conversation services to make interaction more natural and nuanced
  • Watson Discovery Advisor
  • Apps (like Chef Watson)
  • Incorporated into IBM portfolio:  IBM Kenexa, i2, Connections


Jake Porway and Anthony Thomas, CIO Vodaphone India
  • 170 million subscribers
  • Telecom environment in India has been primarily pre-paid, about $3/month.  Many people buying on daily basis.
  • In voice world, knew exactly what getting (e.g., how many minutes).  In data world, though, very hard to do though - can't correlate money spending and bandwidth using.
  • How handle?
  • Keep customers informed of how much used (50% used, 80% used).  
  • Started making offers for off-peak hours, say, or try to move to post-paid
  • Seeing positive reports - # of calls coming in with complaints coming down drastically.  Net Promoter score increasing.

Beth Smith, General Manager, Information Systems
Internet of Things
15 years ago, her husband lost part of his leg.  Just received new prosthetic with computer and sensors in knee that can adjust his gait.  So he no longer needs to adapt to device; now the device fits the way he wants to walk.  A whole range of activities open up that weren't possible before.

IBM is working with Bionik labs - exoskeleton to help disabled patient, with sensors in every joint.  Using Bluemix to analyze data collected from exoskeleton in real time, coupled with data from rehab doctor, so that all data available to doctors and investigators during clinical trials.  End result, better clinical trials.

Examples:
  • Peugeot:  Enabled vehicles with connected services so that drivers can avoid danger zones, find parking, find fuel.  Also partnering with fleet operators and insurers.
  • A Chinese utility is combining weather data with statistical models and satellite imagery, to more accurately forecast energy needs in an area where the eweather is highly unpredictable

IoT is about more than just instrumentation.  Data from local events is being combined with other data, such as enterprise, public, social.  Leads to world where IoT becomes Internet of Insights.  

Three simple steps:
  • Capture:  Secure the data streaming from devices and gateways
  • Analyze: Develop new insights based on integration of info from devices, enterprises, public data
  • Act:  Initiate new processes and actions to deliver business outcomes where they matter.  At speed, in real time, at massive scale.

Example:  Improving safety in vehicles.  Data from sensors in steering wheels, cameras in cars, sensor-connected clothing, mobile phone.  Need to analyze that data, and very little room for latency.  Need to integrate it and diagnose relevant event (is the driver falling asleep, or is it a more serious medical event?)  Decision must be made to take appropriate action.  If driver falling asleep, car can wake him with vibration, audio, flashing lights, or pull over and park itself.  In more serious situation, can pull car over and automatically contact emergency services.

How enable all this?
Bluemix:  Open standards based cloud platform for developing and running applications.

Michael Curry, VP WebSphere Foundation Services
Demo:  How sensor can sense phone and keep safe from equipment not trained to be around in industrial scenario (Starts around ~75 minutes into session)

Using new capabilities Inhi announced yesterday:

Use Cloudant for the capture piece.  
Use DashDB for the analytics from sensors
Build rules and logic to keep person safe - use Node RED (part of Foundation services)

Already built application using IoT foundation starter application boilerplate - connect physical device into cloud

Mobile phone receiving signal from an iBeacon attached to equipment.  Phone will connect into Bluemix, and will build logic to see if authorized to be near it.  Very easy to build out logic, and can take action such as alerting user, rather than shutting down expensive equipment.

Also receiving analytics in DashDB so can see every time an incident occurs.  Is there an issue with a particular person (or in this case, that everyone on a particular team is having problems, and the whole team needs more training.)


Jake Porway and Josh Cooper, CEO, Hildebrand
Collecting about 1B data points a day

Processing it through machine learning and then putting into some Bluemix tools and cloud platforms that IBM is hosting.  Helps retailers optimize what selling in shopping malls.
Have APIs tied into digital signage in mall.


Doug Balog, VP, Systems & Technology Group
Does infrastructure still matter?  Client study says yes.
Even though 70% of organizations with leading IT infrastructure practices recognize that IT infrastructure enables competitive advantage, less than 10% believe their existing infrastructure is prepared to address the future requirements.
  • NC State University:  Power Systems, running Big Insights for Hadoop.  Analytics queries running 37x faster, and 4x shrinkage in data center.
  • Coca-Cola Bottling Co. Consolidated - 20x improvement in supply chain analytics on System Storage
  • Swiss Mobiliar - 80% improvement in access to critical data on System z, bringing together what had been islands of data

Infrastructure also lets you create differentiation

What is IBM doing:
  • Big Data needs big systems to run and crunch analytics algorithms, running across multiple cores and threads
  • Memory needs space, cache close to processor
  • System Z - integrates real time analytics, bringing all data together on same footprint
  • Storage - fast access to information.  Flash storage.