DARPA Awards Charles River Analytics $5.7 M
- To Advance Machine Learning through Probabilistic Programming
Charles River Analytics Inc. a developer of intelligent systems solutions, has announced a contract awarded by the Defense Advanced Research Projects Agency (DARPA) to extend the capabilities of machine learning. Charles River is leveraging probabilistic programming to create a powerful new machine learning system under the Automated Probabilistic Programming Representation and Inference Languages (APPRIL) program. The broad agency announcement contract was awarded as part of DARPA’s Probabilistic Programming for Advancing Machine Learning (PPAML) program. The contract is valued at over $5.7 million over a forty-six month period.
Machine learning is a branch of artificial intelligence that focuses on programming computer systems to automatically learn, act, and improve with experience. It has led to developments such as more effective web searches, an improved understanding of the human genome, and even improved robots.
Antenna Transducer and Radome Market - Analysis & Global Forecasts to 2020
“We want to do for machine learning what the advent of high-level program languages 50 years ago did for the software development community as a whole,” said Dr. Kathleen Fisher, DARPA program manager. “Our goal is that future machine learning projects won’t require people to know everything about both the domain of interest and machine learning to build useful machine learning applications. Through new probabilistic programming languages specifically tailored to probabilistic inference, we hope to decisively reduce the current barriers to machine learning and foster a boom in innovation, productivity and effectiveness.”
“In probabilistic programming, an individual describes a probabilistic model of the domain and the system automatically creates algorithms to reason with the model,” explained Dr. Avi Pfeffer, Principal Scientist at Charles River. “This model is expressed using programming language concepts, which can include complex data structures and control flow constructs.” Dr. Pfeffer continued, “Under APPRIL, we will expand our Figaro probabilistic programming language into a robust system with advanced algorithms and automatic problem-solving capabilities.”
Figaro has been used in a number of efforts at Charles River, including prototyping a system for DARPA on Probabilistic and Relational Inferences in Dynamic Environments (PRIDE). Figaro was used to develop a probabilistic-based framework for monitoring complex, dynamic situations and assessing the likelihood of a given plan succeeding in these situations.
Source : Charles River Analytics