spark-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From dbtsai <>
Subject [GitHub] spark pull request: L-BFGS Documentation
Date Fri, 09 May 2014 22:55:19 GMT
Github user dbtsai commented on a diff in the pull request:
    --- Diff: docs/ ---
    @@ -128,10 +128,24 @@ is sampled, i.e. `$|S|=$ miniBatchFraction $\cdot n = 1$`, then
the algorithm is
     standard SGD. In that case, the step direction depends from the uniformly random sampling
of the
    +### Limited-memory BFGS
    +[Limited-memory BFGS (L-BFGS)]( is an
    +algorithm in the family of quasi-Newton methods to solve the optimization problems of
the form 
    +`$\min_{\wv \in\R^d} \; f(\wv)$`. The L-BFGS approximates the objective function locally
as a quadratic
    +without evaluating the second partial derivatives of the objective function to construct
    +Hessian matrix. The Hessian matrix is approximated by previous gradient evaluations,
so there is no 
    +vertical scalability issue (the number of training features) when computing the Hessian
    +explicitly in Newton method. As a result, L-BFGS often achieves rapider convergence compared
    +other first-order optimization. 
    +Since the Hessian is constructed approximately from previous gradient evaluations, the
    +function can not be changed during the optimization process. As a result, Stochastic
L-BFGS will 
    +not work naively by just using miniBatch; therefore, we don't provide this until we have
    --- End diff --
    I decided to move those message to the code. 

If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at or file a JIRA ticket
with INFRA.

View raw message