Charles River Analytics has developed a programming language that is built to teach a machine to adapt its behavior to changing experiences and unpredictable data sets.
The company presented the Automated Probabilistic Programming Representation and Inference Languages as a case study to the Defense Advanced Research Projects Agency, Charles River said Monday.
In March 2014, DARPA awarded a$5.7 million, 46-month broad agency announcement contract to the Cambridge-based company under PPAML, an initiative to advance machine learning through probabilistic programming.
Avi Pfeffer, principal scientist at Charles River, said the firm plans to modify its proprietary framework for the APPRIL system.
“We will expand our Figaro probabilistic programming language into a robust system with advanced algorithms and automatic problem-solving capabilities,” Pfeffer said on Mar. 10.
According to Charles River, the APPRIL program has already tackled several challenges that were presented within and outside of the PPAML spectrum, such as developing a model to predict bird migration patterns and learning the lineages of malware families.