Mailing-List: contact dev-help@reef.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@reef.apache.org
Date: Thu, 27 Jul 2017 00:55:03 +0000 (UTC)
From: "Julia (JIRA)" <jira@apache.org>
To: dev@reef.apache.org
Message-ID: <JIRA.13090390.1501116868000.23294.1501116903435@Atlassian.JIRA>
In-Reply-To: <JIRA.13090390.1501116868000@Atlassian.JIRA>
References: <JIRA.13090390.1501116868000@Atlassian.JIRA> <JIRA.13090390.1501116868035@jira-lw-us.apache.org>
Subject: [jira] [Created] (REEF-1842) Making IMRU task and input data
 association deterministic
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Thu, 27 Jul 2017 00:55:13 -0000

Julia created REEF-1842:
---------------------------

             Summary: Making IMRU task and input data association deterministic
                 Key: REEF-1842
                 URL: https://issues.apache.org/jira/browse/REEF-1842
             Project: REEF
          Issue Type: Task
            Reporter: Julia
            Priority: Critical


Currently, the tasks added to the typology on communication group is based on the sequence of contexts received. While the input data assigned to context can be random especially when some evaluators fail and new evaluators/contexts are requested. This makes the training result undeterministic given same input data set. 
To make the result deterministic for given the input data, we would like to the tasks in the typology keep the same order with the same input data file. To achieve it, the mapping between the context id and partitionDescriptor id should be fixed once assigned. And when adding tasks to the communication group, the order should be based on the sorted id to ensure it is deterministic. 


--
This message was sent by Atlassian JIRA
(v6.4.14#64029)