Researchers from Massachusetts Institute of Technology Report Findings in Machin e Learning (Using Taylor-approximated Gradients To Improve the Frank-wolfe Metho d for Empirical Risk Minimization)
Researchers from Massachusetts Institute of Technology Report Findings in Machin e Learning (Using Taylor-approximated Gradients To Improve the Frank-wolfe Metho d for Empirical Risk Minimization)
By a News Reporter-Staff News Editor at Robotics & Machine Learning DailyNews Daily News - Data detailed on Machine Learning have been presented. According to news originatingfrom Cambridge, Massachusetts, by NewsRx correspondents, research stated, “The Frank-Wolfe method has become incr easingly useful in statistical and machine learning applications due to the stru ctureinducingproperties of the iterates and especially in settings where linea r minimization over the feasible setis more computationally efficient than proj ection. In the setting of empirical risk minimization-one of thefundamental opt imization problems in statistical and machine learning-the computational effecti venessof Frank-Wolfe methods typically grows linearly in the number of data obs ervations n. This is in starkcontrast to the case for typical stochastic projec tion methods.”
Key words
Cambridge/Massachusetts/United States/North and Central America/Cyborgs/Emerging Technologies/Machine Learning/Ma ssachusetts Institute of Technology
NETL
NSTL