Matches in SemOpenAlex for { <https://semopenalex.org/work/W2417107316> ?p ?o ?g. }
- W2417107316 abstract "A rank-r matrix X in R^{m x n} can be written as a product UV', where U in R^{m x r} and V in R^{n x r}. One could exploit this observation in optimization: e.g., consider the minimization of a convex function f(X) over rank-r matrices, where the scaffold of rank-r matrices is modeled via the factorization in U and V variables. Such heuristic has been widely used before for specific problem instances, where the solution sought is (approximately) low-rank. Though such parameterization reduces the number of variables and is more efficient in computational speed and memory requirement (of particular interest is the case r << min{m, n}), it comes at a cost: f(UV') becomes a non-convex function w.r.t. U and V. In this paper, we study such parameterization in optimization of generic convex f and focus on first-order, gradient descent algorithmic solutions. We propose an algorithm we call the Bi-Factored Gradient Descent (BFGD) algorithm, an efficient first-order method that operates on the U, V factors. We show that when f is smooth, BFGD has local sublinear convergence, and linear convergence when f is both smooth and strongly convex. Moreover, for several key applications, we provide simple and efficient initialization schemes that provide approximate solutions good enough for the above convergence results to hold." @default.
- W2417107316 created "2016-06-24" @default.
- W2417107316 creator A5024280658 @default.
- W2417107316 creator A5038315348 @default.
- W2417107316 creator A5040127066 @default.
- W2417107316 creator A5053978837 @default.
- W2417107316 date "2016-06-10" @default.
- W2417107316 modified "2023-09-27" @default.
- W2417107316 title "Finding Low-rank Solutions to Matrix Problems, Efficiently and Provably." @default.
- W2417107316 cites W1506690472 @default.
- W2417107316 cites W1529624360 @default.
- W2417107316 cites W1568698519 @default.
- W2417107316 cites W1573301526 @default.
- W2417107316 cites W1574851760 @default.
- W2417107316 cites W1587887312 @default.
- W2417107316 cites W1604604340 @default.
- W2417107316 cites W1667938101 @default.
- W2417107316 cites W1673735735 @default.
- W2417107316 cites W1736339626 @default.
- W2417107316 cites W1775587472 @default.
- W2417107316 cites W1798398164 @default.
- W2417107316 cites W1834987204 @default.
- W2417107316 cites W1836708065 @default.
- W2417107316 cites W1874232560 @default.
- W2417107316 cites W1877469910 @default.
- W2417107316 cites W1884359871 @default.
- W2417107316 cites W1967921672 @default.
- W2417107316 cites W1968157422 @default.
- W2417107316 cites W1976618413 @default.
- W2417107316 cites W1977361006 @default.
- W2417107316 cites W1987570579 @default.
- W2417107316 cites W2000345836 @default.
- W2417107316 cites W2023101933 @default.
- W2417107316 cites W2042465463 @default.
- W2417107316 cites W2054141820 @default.
- W2417107316 cites W2059283452 @default.
- W2417107316 cites W2060204507 @default.
- W2417107316 cites W2068019581 @default.
- W2417107316 cites W2078455576 @default.
- W2417107316 cites W2098044836 @default.
- W2417107316 cites W21006490 @default.
- W2417107316 cites W2102019642 @default.
- W2417107316 cites W2103972604 @default.
- W2417107316 cites W2104780023 @default.
- W2417107316 cites W2106005123 @default.
- W2417107316 cites W2110451175 @default.
- W2417107316 cites W2111297856 @default.
- W2417107316 cites W2116600879 @default.
- W2417107316 cites W2116680824 @default.
- W2417107316 cites W2117377454 @default.
- W2417107316 cites W2117756735 @default.
- W2417107316 cites W2118550318 @default.
- W2417107316 cites W2119058682 @default.
- W2417107316 cites W2122090912 @default.
- W2417107316 cites W2125027820 @default.
- W2417107316 cites W2125238156 @default.
- W2417107316 cites W2126418337 @default.
- W2417107316 cites W2130185724 @default.
- W2417107316 cites W2135001774 @default.
- W2417107316 cites W2140336578 @default.
- W2417107316 cites W2143075842 @default.
- W2417107316 cites W2143701678 @default.
- W2417107316 cites W2145080587 @default.
- W2417107316 cites W2145179194 @default.
- W2417107316 cites W2145962650 @default.
- W2417107316 cites W2146502635 @default.
- W2417107316 cites W2147542713 @default.
- W2417107316 cites W2156637418 @default.
- W2417107316 cites W2158923808 @default.
- W2417107316 cites W2161374719 @default.
- W2417107316 cites W2162451874 @default.
- W2417107316 cites W2164214353 @default.
- W2417107316 cites W2189224591 @default.
- W2417107316 cites W2256553158 @default.
- W2417107316 cites W2282361021 @default.
- W2417107316 cites W2296616510 @default.
- W2417107316 cites W2396019715 @default.
- W2417107316 cites W2399402693 @default.
- W2417107316 cites W2403076655 @default.
- W2417107316 cites W2405479119 @default.
- W2417107316 cites W2406099690 @default.
- W2417107316 cites W2461939514 @default.
- W2417107316 cites W2529189650 @default.
- W2417107316 cites W2557168633 @default.
- W2417107316 cites W2595191246 @default.
- W2417107316 cites W2611328865 @default.
- W2417107316 cites W2950176250 @default.
- W2417107316 cites W2962737134 @default.
- W2417107316 cites W2962904925 @default.
- W2417107316 cites W2963014255 @default.
- W2417107316 cites W2963099043 @default.
- W2417107316 cites W2963262327 @default.
- W2417107316 cites W2963449128 @default.
- W2417107316 cites W2963687412 @default.
- W2417107316 cites W2964200481 @default.
- W2417107316 cites W3141595720 @default.
- W2417107316 cites W826455576 @default.
- W2417107316 cites W3142913680 @default.
- W2417107316 hasPublicationYear "2016" @default.
- W2417107316 type Work @default.