Return-Path: X-Original-To: apmail-flink-issues-archive@minotaur.apache.org Delivered-To: apmail-flink-issues-archive@minotaur.apache.org Received: from mail.apache.org (hermes.apache.org [140.211.11.3]) by minotaur.apache.org (Postfix) with SMTP id 3818C17E02 for ; Thu, 21 May 2015 13:49:03 +0000 (UTC) Received: (qmail 15385 invoked by uid 500); 21 May 2015 13:49:03 -0000 Delivered-To: apmail-flink-issues-archive@flink.apache.org Received: (qmail 15291 invoked by uid 500); 21 May 2015 13:49:03 -0000 Mailing-List: contact issues-help@flink.apache.org; run by ezmlm Precedence: bulk List-Help: List-Unsubscribe: List-Post: List-Id: Reply-To: dev@flink.apache.org Delivered-To: mailing list issues@flink.apache.org Received: (qmail 15124 invoked by uid 99); 21 May 2015 13:49:02 -0000 Received: from arcas.apache.org (HELO arcas.apache.org) (140.211.11.28) by apache.org (qpsmtpd/0.29) with ESMTP; Thu, 21 May 2015 13:49:02 +0000 Date: Thu, 21 May 2015 13:49:02 +0000 (UTC) From: "Theodore Vasiloudis (JIRA)" To: issues@flink.apache.org Message-ID: In-Reply-To: References: Subject: [jira] [Updated] (FLINK-1994) Add different gain calculation schemes to SGD MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit X-JIRA-FingerPrint: 30527f35849b9dde25b450d4833f0394 [ https://issues.apache.org/jira/browse/FLINK-1994?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Theodore Vasiloudis updated FLINK-1994: --------------------------------------- Labels: ML Starter (was: ML) > Add different gain calculation schemes to SGD > --------------------------------------------- > > Key: FLINK-1994 > URL: https://issues.apache.org/jira/browse/FLINK-1994 > Project: Flink > Issue Type: Improvement > Components: Machine Learning Library > Reporter: Till Rohrmann > Priority: Minor > Labels: ML, Starter > > The current SGD implementation uses as gain for the weight updates the formula {{stepsize/sqrt(iterationNumber)}}. It would be good to make the gain calculation configurable and to provide different strategies for that. For example: > * stepsize/(1 + iterationNumber) > * stepsize*(1 + regularization * stepsize * iterationNumber)^(-3/4) > See also how to properly select the gains [1]. > Resources: > [1] http://arxiv.org/pdf/1107.2490.pdf -- This message was sent by Atlassian JIRA (v6.3.4#6332)