[ABE-L] Seminarios em High Dimensional Data Analysis

[Ronaldo Dias]

Prezados(as)
Convido a todos(as) que puderem participar
para mais um seminário em Análise de Dados em Alta Dimensão <AD,AD>
Sejam bem vindos.

Local: sala 221-IMECC
Data: 29/05 as 13hs.

Accelerating block coordinate descent methods with identification
strategies and Applications to Lasso and Logistic Regression

This work is about active set identification strategies aimed at
accelerating block-coordinate descent methods (BCDM) applied to
large-scale problems. We start by devising an identification function
tailored for bound-constrained composite minimization together with an
associated version of the BCDM, called Active BCDM, that is also
globally convergent. The identification function gives rise to an
efficient practical strategy for Lasso and ℓ1-regularized logistic
regression. The computational performance of Active BCDM is
contextualized using comparative sets of experiments that are based on
the solution of problems with data from deterministic instances from
the literature. These results have been compared with those of
well-established and state-of-the-art methods that are particularly
suited for the classes of applications under consideration. Active
BCDM has proved useful in achieving fast results due to its
identification strategy. Besides that, an extra second-order step was
used, with favorable cost-benefit.


-- 
Ronaldo Dias, Ph.D.
Professor
Dept. of Statistics-IMECC, UNICAMP
www.ime.unicamp.br/~dias
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <https://lists.ime.usp.br/archives/abe/attachments/20190528/f8128fe0/attachment.html>