Microsoft Research Group Releases “OWL-QN”

I obviously monitor a great many Microsoft resources — and of course, I subscribe to the Microsoft Research Group (where people smarter than "us" work).
So when the Microsoft Research Group released OWL-QN today — I giggled that I had NO clue what the heck this is . . . but it sure sounds interesting! ha ha
The Orthant-Wise Limited-memory Quasi-Newton algorithm (OWL-QN) is a numerical optimization procedure for finding the optimum of an objective of the form {smooth function} plus {L1-norm of the parameters}. It has been used for training log-linear models (such as logistic regression) with L1-regularization. The algorithm is described in "Scalable training of L1-regularized log-linear models" by Galen Andrew and Jianfeng Gao. To use OWLQN, one must define the (unregularized) objective function by deriving a subclass of the provided abstract base class "DifferentiableFunction", implementing the method "ValueAt". ValueAt takes as arguments a const vector respresenting the current point and a modifiable vector which the implementation must fill in with the gradient of the function at that point. It should return the value of the function at that point. It is assumed that the function is convex and differentiable. Then the user passes an instance of their DifferentiableFunction to the Minimize method. Minimize takes two other mandatory arguments– the initial point for optimization, and a reference to a vector which OWLQN will fill in with the optimal point. It also has three optional arguments with the following default values: double l1weight (1.0) : weight of L1 regularizer — double tol (1e-4) : convergence tolerance — int m (10) : memory parameter for L-BFGS

About blakehandler

BLAKE was a Microsoft MVP and award winning programmer with over 20+ years experience providing complete Windows and networking support for small to medium sized businesses. BLAKE is also Jazz Musician and Instructor for residential clients on the Los Angeles West Side.
This entry was posted in Microsoft Must Haves. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s