Software: MIT Researchers Develop More Intuitive Programming Method Based on Probability
Researchers at MIT are working to create a new generation of programming languages that are much more compact and declarative, compared to typical programming languages that are very detailed and prescriptive in describing how a task must be performed. They’re calling this new style of programming language “probabilistic code”.Tejas Kulkarni, a member of the MIT team working on the new languages, says that “the whole hope is to write very flexible models, both generative and discriminative models, as short probabilistic code, and then not do anything else. General-purpose inference schemes solve the problems”.Josh Tenenbaum, an MIT professor of computational cognitive science, said that “in pure machine learning, you drive performance increases by just collecting more and more data and just letting machine learning do the work. In probabilistic programming, the underlying system is more knowledge-based, using the causal process of how images are formed.As a prototype of the new type of probabilistic language, researchers created a language they’re calling ‘Picture’. Using Picture they were able to perform a complex image processing task in just 50 lines of code that would have required many thousands of lines using conventional programming languages.