Parametric Local Metric Learning for Nearest Neighbor Classification
This work addresses overfitting issues in metric learning for classification, offering a scalable solution that improves predictive power for large-scale datasets.
The paper tackles the problem of overfitting in local metric learning for nearest neighbor classification by proposing a parametric method that learns a smooth metric matrix function over the data manifold, using manifold regularization and linear combinations of basis metrics. It demonstrates excellent performance, outperforming state-of-the-art global and local metric learning methods and SVM with automatic kernel selection on large-scale classification problems with tens of thousands of instances.
We study the problem of learning local metrics for nearest neighbor classification. Most previous works on local metric learning learn a number of local unrelated metrics. While this "independence" approach delivers an increased flexibility its downside is the considerable risk of overfitting. We present a new parametric local metric learning method in which we learn a smooth metric matrix function over the data manifold. Using an approximation error bound of the metric matrix function we learn local metrics as linear combinations of basis metrics defined on anchor points over different regions of the instance space. We constrain the metric matrix function by imposing on the linear combinations manifold regularization which makes the learned metric matrix function vary smoothly along the geodesics of the data manifold. Our metric learning method has excellent performance both in terms of predictive power and scalability. We experimented with several large-scale classification problems, tens of thousands of instances, and compared it with several state of the art metric learning methods, both global and local, as well as to SVM with automatic kernel selection, all of which it outperforms in a significant manner.