BudgetedSVM Icon

BudgetedSVM

BudgetedSVM: A C++ Toolbox for Large-scale, Non-linear Classification

Add a Review
13 Downloads (This Week)
Last Update:
Download BudgetedSVM_v1.1.zip
Browse All Files

Description

We present BudgetedSVM, a C++ toolbox containing highly optimized implementations of three recently proposed algorithms for scalable training of Support Vector Machine (SVM) approximators: Adaptive Multi-hyperplane Machines (AMM), Budgeted Stochastic Gradient Descent (BSGD), and Low-rank Linearization SVM (LLSVM). BudgetedSVM trains models with accuracy comparable to LibSVM in time comparable to LibLinear, as it allows solving highly non-linear classi fication problems with millions of high-dimensional examples within minutes on a regular personal computer. We provide command-line and Matlab interfaces to BudgetedSVM, efficient API for handling large-scale, high-dimensional data sets, as well as detailed documentation to help developers use and further extend the toolbox.

BudgetedSVM Web Site

Features

  • We provide efficient implementations of algorithms for highly-scalable non-linear SVM training.
  • The toolbox can handle large, high-dimensional data sets that cannot be loaded into memory.
  • The toolbox requires constant memory to train models that solve highly non-linear problems.
  • We provide command-line and Matlab interfaces to BudgetedSVM.
  • We provide an efficient API that provides functionalities for handling large, high-dimensional data sets. Using BudgetedSVM API, data sets with millions data points and/or features are easily handled.
  • For more details, please see the documentation included in the download package.
  • Published under industry-friendly Modified BSD licence.

Update Notifications





Write a Review

User Reviews

Be the first to post a review of BudgetedSVM!

Additional Project Details

Languages

English

Intended Audience

Information Technology, Science/Research, End Users/Desktop

Programming Language

C++

Registered

2013-05-06

Icons must be PNG, GIF, or JPEG and less than 1 MiB in size. They will be displayed as 48x48 images.