Learning Machines

Recounting explorations in the art and science of #ML and #AI

  • Best Practices in Conversational UI

    The chatbot space is still in it's infancy and one of the key areas of learning over the last year has been around techniques for designing conversational interfaces.  This is a brave new world and it differs quite a bit from our existing forms-based user interface paradigm, whether it be on a desktop, web or mobile.  The information below is certainly not my original content, but rather the collected tidbits of wisdom that I've gathered over the past year on the topic by reading various articles and blogs.

  • Is Conversation the new UI?

    At the recent Denver Dev Days conference, on June 24th, 2016, I explored the topic of whether "Conversation is the New UI"?  The presentation came about as a reaction to the announcment of Microsoft's Bot Framework - https://dev.botframework.com.  Initially, when I watched a few of the sessions from the Microsoft Build conference (https://build.microsoft.com), my thoughts were that using messaging systems such as SMS, Skype, and Facebook Messenger seemed to be a tedious way for a user to go about achieving a particular task that they wish to perform with a computer.  Much better and more familiar to work within a traditional forms based interface that the user was accustomed to than introducing some new paradigm that is unlikely to stick.  However, as I dug deeper, I quickly realized that I was completely missing the point.  The end game for this technology is clearly not the interface, in this case a Messaging system, but the idea of conversation and computerized Natural Language Understanding.  Conversation is a familar way that human beings interact.  In fact, it has been the holy grail of human computer interaction for many years.

  • Collapse Rows of Dates into a Single, Comma-Delimited Column in SQL Server

    The Internet is an amazing thing.  I remember back in the early days of my career, when you hit a problem the only resources at your disposal were your colleages and documentation.  Today you simply do a Google search and often wind up on StackOverflow.  That's how I found the solution the problem I faced last night.  Here's the StackOverflow that got me over the hump:

  • Azure ML at the Data Science Association Machine Learning Forum

    The Machine Learning Forum took place Saturday, November 7th at University of Colorado Denver.  There was a great crowd of machine learning and data science practitioners in attendance and I was honored to share an intro to Azure Machine Learning with them.  The Forum was put together to promote the Hackski machine learning contest, which I really hope is a great success in helping Colorado Front Range skiers drive less and ski more!  Check out all the great speakers at: http://www.datascienceassn.org/content/2015-11-07-data-science-machine-learning-forum

  • Operationalizing R with Azure ML

    Had a great time discussing Azure ML with the Denver R User Group meeting this evening.  We covered ways to use R in Azure ML to clean, transform and process data and to build machine learning models.  This is a great meetup if you have a chance to attend a future meeting!

  • Smart Apps with Azure ML

    I recently presented at the Denver Dev Day on Smart Apps with Azure ML:  In the words of Marc Andreessen, "Software is eating the world".  Industries are being disrupted at an alarming rate due to intelligent software.  Azure Machine Learning enables developers to easily add intelligence to their Apps.  In this session we'll look at the recently GA'd Azure ML service and see how it's easy to make your Apps smart!

  • Named Entity Recognition and Generating Keys for your Datasets

    Recently I've been exploring the Text Analytics capabilities of Azure ML, in particular "Entity Recognition".  Entity Recognition is a valuable Natural Language Processing technique that allows you to extract named entities (think Proper Nouns) from a body of text.  Azure ML Named Entity Recognition goes a step further by allowing you to identify the type of entity.  Currently 3 types of entities are supported, Name, Location and Organition (labelled "NAM", LOC", and "ORG" respectively).  Additionally the Named Entity Module gives you the entity name, the location in the original text, and the length of the string.

  • Adventures in Authenticode

    I'm currently working on renewing our Microsoft Partner Network membership.  As part of this process, we can optionally either pass certification exams or test our product against the Microsoft Platform Readiness tests.  Naturally, I chose to test our product since this is the basis of our partner relationship with Microsoft.  Last time around it was easy, our QA executed the tests and our product passed with no modifications.  This time around, however, that was not the case.  I ran the tests and hit a number of snags.  It appears these tests have become much more rigorous.  Overall, this is a good thing for the Microsoft community.  For me, however, I was hoping for a lay up and instead got a center court shot at the buzzer to tie the game.  Such is the life of a product developer - the only way out is through.

  • Stop Moving Leaves from the Front Yard to the Back Yard!

    I've been hearing a lot lately about the challenges moving from Exchange On-Prem to an Office 365 deployment.  At Integro, we did a cold cutover from Lotus Notes to Office 365, leaving our legacy baggage behind (but properly classified, managed and accessible).  Our friends at Transvault have a nice solution for this kind of migration and Integro can help you implement it.  However, I still ask the question, "Why move the leaves from the front yard to the backyard?"  To give proper credit, this poignant quip was first phrased by my buddy Jim Morganstern.  It so aptly applies to this challenge with email migration.  You see at Integro, with IEM, we're on a mission to eliminate ROT from your email environment.  Storage is cheap, but this problem proves why maintaining excessive ROT on your mail servers is expensive.  With IEM, our customers are seeing a reduction of up to 85% or more of their total mail volume.  However, it's selective reduction, eliminating the unnecessary and useless, while keeping the valuable and important.  But the key element with IEM is that you, the email end user, are in in total control.  We utilize machine learning technology to reduce the burden without subverting your critical judgment.  Your judgement helps the system improve over time.  Giving email users ease of use and control is the key to acceptance and to ensure that the 20% of important records and business documents are maintained and readily accessible.  The nice thing is that IEM works for both Exchange On-Prem, Office 365 and Domino environments.  We also have some technology in the works to cleanup your legacy content using our patent-pending approach we call "Auto-classification with Human Oversight".  Contact us to learn more about how we can help eliminate the mess and stop moving leaves from your front yard to your backyard: http://www.integro.com/contact-us 

  • 1