From the March 2014 issue

Good Practice in Large-Scale Learning for Image Classification

By Zeynep Akata, Florent Perronnin, Zaid Harchaoui, Cordelia Schmid

Featured article thumbnail imageWe benchmark several SVM objective functions for large-scale image classification. We consider one-versus-rest, multiclass, ranking, and weighted approximate ranking SVMs. A comparison of online and batch methods for optimizing the objectives shows that online methods perform as well as batch methods in terms of classification accuracy, but with a significant gain in training speed. Using stochastic gradient descent, we can scale the training to millions of images and thousands of classes. Our experimental evaluation shows that ranking-based algorithms do not outperform the one-versus-rest strategy when a large number of training examples are used. Furthermore, the gap in accuracy between the different algorithms shrinks as the dimension of the features increases. We also show that learning through cross-validation the optimal rebalancing of positive and negative examples can result in a significant improvement for the one-versus-rest strategy. Finally, early stopping can be used as an effective regularization strategy when training with online algorithms. Following these "good practices," we were able to improve the state of the art on a large subset of 10K classes and 9M images of ImageNet from 16.7 percent Top-1 accuracy to 19.1 percent.

download PDF View the PDF of this article      csdl View this issue in the digital library


Editorials and Announcements

Announcements

Editorials

Guest Editorials

Reviewers List

Annual Index


Access recently published TPAMI articles

RSSSubscribe to the RSS feed of latest TPAMI content added to the digital library

Mail Sign up for the Transactions Connection newsletter.

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) is a scholarly archival journal published monthly. This journal covers traditional areas of computer vision and image understanding, all traditional areas of pattern analysis and recognition, and selected areas of machine intelligence.
Read the full scope of TPAMI