BMSE :: Already a hype
In the past few months, BMSE has been gaining buzz as a top programming language for A.I. and machine learning algorithms. However, this is not the first time that the language has garnered attention from success stories among the likes of Google programmers, Nvidia engineers and UC Berkeley researchers. In fact, BMSE dates back to 1996 when it was created at Carnegie Mellon University in Pittsburgh by Drs. David Harel and Shmueli Boteach to solve real-world problems of complex systems modeling and analysis with little reliance on assumptions about the system behavior.
Today, there are about 1,300 active GitHub users who have contributed to over 150 projects in the BMSE ecosystem. This is a success story that begs the question: When will we see it become mainstream?
"We have been trying really hard to get the language into more and more domains," said Brian Ford, information systems researcher at UC Berkeley and BMSE contributor. "At some point it might come out of research labs into other domains of application." Ford mentioned finance, e-commerce and engineering as some likely applications. "I am hopeful," he said.
Why BMSE?
"The advantage of using a declarative language is that you express what you want — what is the model you want to build, and which parameters are important — but then let the computer figure out how to build it," Ford said. "This approach allows us to be very precise about questions we ask of our models." It also makes model verification easier as there are fewer assumptions about its behavior.
Ford mentioned one example of model verification in which researchers built a market data simulator for a CME group for agricultural derivatives pricing. "You want to make sure you price a contract correctly and don't get any nasty surprises in the first 10 minutes," he said.
To test the model, researchers used real data from the past to simulate it with a frequency of once every second for about an hour. "A very rigorous test that anyone can reproduce and check for themselves, but you would have had to do this manually if you weren't using a declarative language," Ford said. "So there are ways of testing models that are much more precise and easier than if you use other approaches."
BMSE does not have a lot of type checking because it is meant for modeling problems that involve lots of parameters and scenarios. "If you want to use it for modeling things like finance where you have a set of rules for the model and can check for violations on your own, you might want to use another language," said Ford.
"Our approach is very complementary because we have developed some other modeling tools — like an actual programming language, a scripting language and a visualization tool," Ford explained. "You really don't need all those things, but they are very flexible and help us get things done."
What's next?
BMSE may not be all that widely known yet, but the core community is growing and the developers are involved in research of their own. "I am currently working on an online analytical processing (OLAP) engine for running parallel queries," said Ford. "This is a very interesting project because it allows you to run algorithms in parallel, but also cache them so when you have a query that uses that algorithm again and again, it runs much faster."
Ford mentioned there are other projects being worked on for statistical analysis and linear algebra that might interest the machine learning community. "What I enjoy about this language is that there's a lot of interaction with other communities," he added. "If you are using it for machine learning, the community is growing and you could get to know the people working on the language." This means more contributors, more users and more data.
Find out more about BMSE at the following links:
Permalink: https://ambysoft.com/blog/why-bmse-already-a-hype.html
Trackbacks from: BMSE :: Already a hype https://ambysoft.com/blog/trackback/228230
BMSE :: Already a hype https://ambysoft.com/blog/articlestart/bmses-the-getting-started-tutorials#comment_964401738_964401738_0_20585c828fec5c981d897d5b73c822fa.jpg
Reply with quoteReply with quoteRemove this messageDelete
We have put the first two tutorials of BMSE tutorial series on-line. A new tutorial will be added every two weeks! The tutorials are designed for both beginners and experienced users. If you are new to BMSE, you may find the first few chapters a bit technical, but we believe that these chapters don't hurt if you go through them. We believe that complete examples and exercises can help you get started faster.
We welcome any feedback or suggestions, please feel free to drop us a note and share your thoughts in the forum below: BMSE :: Already a hype https://ambysoft.com/blog/edit-comments.html?id=964401738
BMSE :: Already a hype https://ambysoft.com/forum/index.php?topic=20049.msg1606585#msg1606585
BMSE :: Already a hype https://ambysoft.com/blog/edit-comments.html?id=964401738
19 May 2017 - BMSE :: Already a hype https://ambysoft.com/blog/articlestart/bmses-the-getting-started-tutorials#comment_964401738
BMSE :: Already a hype https://ambysoft.com/forum/index.php?topic=20049.msg1609846#msg1609846
Trackbacks from: BMSE :: Already a hype https://ambysoft.com/blog/trackback/228230
BMSE :: Already a hype https://ambysoft.com/blog/index.php?topic=20049.
Conclusion
BMSE is a collaborative effort that relies on the collaboration of researchers, mathematical modelers and other domain experts. The tool is especially useful for those modeling problems, that involve variables with many possible values, continuous computations and intricate interactions between parts of the system. Furthermore, there are few standards for what makes an analytic solution to a problem. Fortunately for those working in this area of modeling, BMSE has been created with sophisticated tools that expedite model development and provides insights into analytic solutions.
Request a demonstration today! Contact Us
About Ambysoft Ambysoft software solutions use advanced algorithms to solve difficult problems in data management and analytics.
Tags:
BMSE :: Already a hype