Boston, MA – When Matt Berseth chose the computing language “Julia” as the computing platform to build the NLP Logix machine learning development kit called LogixStudio, he did so primarily because of the promise of high performance. What he did not know, was that decision, made about two years ago, would land him a guest teaching invitation at the birthplace of the language, the Computer Science Department at MIT.
“We are always looking for ways to increase both our computing performance as well as our skills here at NLP Logix, and when I saw the high-performance and distributive computing capabilities of Julia coupled with the syntactical elegance I knew I found a winner” said Berseth.
Once NLP Logix chose to develop its machine learning application in Julia, it began a two-year journey of developing and fine-tuning the latest algorithms so NLP Logix could not only deliver cutting edge predictive models, but also the platform to integrate models directly into a customers operating environment. This complete, end-to-end machine learning solution is the first to be written entirely in the Julia Computing language and one of the first examples of the language being used to solve business problems.
But it was not LogixStudio that gained the attention of the Julia community, but rather how well NLP Logix performed in the international data science contest called the KDD Cup 2015. The company placed 13th out of 821 teams from across the world using their computing tools to predict which students would drop out of a massive open online course.
“We were proud to see a team using our language placing so high in such a prestigious contest” said Alan Edelman, MIT Computer Science Professor and Co-Author of the Julia Computing Language. “After we found out how Matt and the team at NLP Logix have taken our language and not only performed so well on the international stage, but are also delivering business solutions, I knew I wanted my students at MIT to hear how they did it.”
The work and communications resulted in Mr. Berseth teaching a 90 minute class to the MIT students on how NLP Logix was able to take the Julia computing language, develop a world-class machine learning solution and build a successful business delivering value to its customers. The class was given on the Cambridge, MA campus on the afternoon of Monday, September 26th to a class of approximately 20 computer science graduate students.
“It was an honor to present our tools to the students” said Berseth. “They had a lot of questions that were not only technical in nature but also focused on how data science can solve real-world problems.”