Matches in SemOpenAlex for { <https://semopenalex.org/work/W4367693790> ?p ?o ?g. }
- W4367693790 abstract "Abstract This work is on constrained large-scale non-convex optimization where the constraint set implies a manifold structure. Solving such problems is important in a multitude of fundamental machine learning tasks. Recent advances on Riemannian optimization have enabled the convenient recovery of solutions by adapting unconstrained optimization algorithms over manifolds. However, it remains challenging to scale up and meanwhile maintain stable convergence rates and handle saddle points. We propose a new second-order Riemannian optimization algorithm, aiming at improving convergence rate and reducing computational cost. It enhances the Riemannian trust-region algorithm that explores curvature information to escape saddle points through a mixture of subsampling and cubic regularization techniques. We conduct rigorous analysis to study the convergence behavior of the proposed algorithm. We also perform extensive experiments to evaluate it based on two general machine learning tasks using multiple datasets. The proposed algorithm exhibits improved computational speed, e.g., a speed improvement from $$12% :text {to} :227%$$ <mml:math xmlns:mml=http://www.w3.org/1998/Math/MathML> <mml:mrow> <mml:mn>12</mml:mn> <mml:mo>%</mml:mo> <mml:mspace /> <mml:mtext>to</mml:mtext> <mml:mspace /> <mml:mn>227</mml:mn> <mml:mo>%</mml:mo> </mml:mrow> </mml:math> , and improved convergence behavior, e.g., an iteration number reduction from $$mathcal{O}left(maxleft(epsilon_g^{-2}epsilon_H^{-1},epsilon_H^{-3}right)right) ,text {to}: mathcal{O}left(maxleft(epsilon_g^{-2},epsilon_H^{-3}right)right)$$ <mml:math xmlns:mml=http://www.w3.org/1998/Math/MathML> <mml:mrow> <mml:mi>O</mml:mi> <mml:mfenced> <mml:mo>max</mml:mo> <mml:mfenced> <mml:msubsup> <mml:mi>ϵ</mml:mi> <mml:mi>g</mml:mi> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>2</mml:mn> </mml:mrow> </mml:msubsup> <mml:msubsup> <mml:mi>ϵ</mml:mi> <mml:mi>H</mml:mi> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>1</mml:mn> </mml:mrow> </mml:msubsup> <mml:mo>,</mml:mo> <mml:msubsup> <mml:mi>ϵ</mml:mi> <mml:mi>H</mml:mi> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>3</mml:mn> </mml:mrow> </mml:msubsup> </mml:mfenced> </mml:mfenced> <mml:mspace /> <mml:mtext>to</mml:mtext> <mml:mspace /> <mml:mi>O</mml:mi> <mml:mfenced> <mml:mo>max</mml:mo> <mml:mfenced> <mml:msubsup> <mml:mi>ϵ</mml:mi> <mml:mi>g</mml:mi> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>2</mml:mn> </mml:mrow> </mml:msubsup> <mml:mo>,</mml:mo> <mml:msubsup> <mml:mi>ϵ</mml:mi> <mml:mi>H</mml:mi> <mml:mrow> <mml:mo>-</mml:mo> <mml:mn>3</mml:mn> </mml:mrow> </mml:msubsup> </mml:mfenced> </mml:mfenced> </mml:mrow> </mml:math> , compared to a large set of state-of-the-art Riemannian optimization algorithms." @default.
- W4367693790 created "2023-05-03" @default.
- W4367693790 creator A5009213856 @default.
- W4367693790 creator A5048659336 @default.
- W4367693790 date "2023-05-02" @default.
- W4367693790 modified "2023-09-30" @default.
- W4367693790 title "Faster Riemannian Newton-type optimization by subsampling and cubic regularization" @default.
- W4367693790 cites W147998453 @default.
- W4367693790 cites W2012231377 @default.
- W4367693790 cites W2015627422 @default.
- W4367693790 cites W2032089608 @default.
- W4367693790 cites W2045512849 @default.
- W4367693790 cites W2045989120 @default.
- W4367693790 cites W2053799091 @default.
- W4367693790 cites W2063690908 @default.
- W4367693790 cites W2066214692 @default.
- W4367693790 cites W2076669446 @default.
- W4367693790 cites W2110662303 @default.
- W4367693790 cites W2112796928 @default.
- W4367693790 cites W2117354486 @default.
- W4367693790 cites W2149532724 @default.
- W4367693790 cites W2151591509 @default.
- W4367693790 cites W2156005216 @default.
- W4367693790 cites W2402588523 @default.
- W4367693790 cites W2462905479 @default.
- W4367693790 cites W2486565058 @default.
- W4367693790 cites W2556460356 @default.
- W4367693790 cites W2789558982 @default.
- W4367693790 cites W2798738034 @default.
- W4367693790 cites W2900789157 @default.
- W4367693790 cites W2952200000 @default.
- W4367693790 cites W2963307318 @default.
- W4367693790 cites W2964089577 @default.
- W4367693790 cites W2983044386 @default.
- W4367693790 cites W3010910782 @default.
- W4367693790 cites W3034438741 @default.
- W4367693790 cites W3098671196 @default.
- W4367693790 cites W3101159735 @default.
- W4367693790 cites W3102317997 @default.
- W4367693790 cites W3104501706 @default.
- W4367693790 cites W3135262686 @default.
- W4367693790 cites W3135904811 @default.
- W4367693790 cites W3166589644 @default.
- W4367693790 cites W4231407710 @default.
- W4367693790 cites W4281382771 @default.
- W4367693790 cites W4320800818 @default.
- W4367693790 cites W615893297 @default.
- W4367693790 cites W2790379190 @default.
- W4367693790 doi "https://doi.org/10.1007/s10994-023-06321-0" @default.
- W4367693790 hasPublicationYear "2023" @default.
- W4367693790 type Work @default.
- W4367693790 citedByCount "0" @default.
- W4367693790 crossrefType "journal-article" @default.
- W4367693790 hasAuthorship W4367693790A5009213856 @default.
- W4367693790 hasAuthorship W4367693790A5048659336 @default.
- W4367693790 hasBestOaLocation W43676937901 @default.
- W4367693790 hasConcept C11413529 @default.
- W4367693790 hasConcept C119857082 @default.
- W4367693790 hasConcept C126255220 @default.
- W4367693790 hasConcept C134306372 @default.
- W4367693790 hasConcept C154945302 @default.
- W4367693790 hasConcept C162324750 @default.
- W4367693790 hasConcept C195065555 @default.
- W4367693790 hasConcept C2524010 @default.
- W4367693790 hasConcept C2681867 @default.
- W4367693790 hasConcept C2776135515 @default.
- W4367693790 hasConcept C2777127463 @default.
- W4367693790 hasConcept C2777303404 @default.
- W4367693790 hasConcept C2779593128 @default.
- W4367693790 hasConcept C33923547 @default.
- W4367693790 hasConcept C41008148 @default.
- W4367693790 hasConcept C50522688 @default.
- W4367693790 hasConceptScore W4367693790C11413529 @default.
- W4367693790 hasConceptScore W4367693790C119857082 @default.
- W4367693790 hasConceptScore W4367693790C126255220 @default.
- W4367693790 hasConceptScore W4367693790C134306372 @default.
- W4367693790 hasConceptScore W4367693790C154945302 @default.
- W4367693790 hasConceptScore W4367693790C162324750 @default.
- W4367693790 hasConceptScore W4367693790C195065555 @default.
- W4367693790 hasConceptScore W4367693790C2524010 @default.
- W4367693790 hasConceptScore W4367693790C2681867 @default.
- W4367693790 hasConceptScore W4367693790C2776135515 @default.
- W4367693790 hasConceptScore W4367693790C2777127463 @default.
- W4367693790 hasConceptScore W4367693790C2777303404 @default.
- W4367693790 hasConceptScore W4367693790C2779593128 @default.
- W4367693790 hasConceptScore W4367693790C33923547 @default.
- W4367693790 hasConceptScore W4367693790C41008148 @default.
- W4367693790 hasConceptScore W4367693790C50522688 @default.
- W4367693790 hasFunder F4320322725 @default.
- W4367693790 hasLocation W43676937901 @default.
- W4367693790 hasLocation W43676937902 @default.
- W4367693790 hasOpenAccess W4367693790 @default.
- W4367693790 hasPrimaryLocation W43676937901 @default.
- W4367693790 hasRelatedWork W1482630493 @default.
- W4367693790 hasRelatedWork W2033071165 @default.
- W4367693790 hasRelatedWork W2051848413 @default.
- W4367693790 hasRelatedWork W2066844763 @default.
- W4367693790 hasRelatedWork W2082990920 @default.
- W4367693790 hasRelatedWork W2771077211 @default.
- W4367693790 hasRelatedWork W2804907179 @default.