Abstract:Distance metric learning (DML) is a critical factor for image analysis and pattern recognition. To learn a robust distance metric for a target task, we need abundant side information (i.e., the similarity/dissimilarity pairwise constraints over the labeled data), which is usually unavailable in practice due to the high labeling cost. This paper considers the transfer learning setting by exploiting the large quantity of side information from certain related, but different source tasks to help with target metric learning (with only a little side information). The state-of-the-art metric learning algorithms usually fail in this setting because the data distributions of the source task and target task are often quite different. We address this problem by assuming that the target distance metric lies in the space spanned by the eigenvectors of the source metrics (or other randomly generated bases). The target metric is represented as a combination of the base metrics, which are computed using the decomposed components of the source metrics (or simply a set of random bases); we call the proposed method, decomposition-based transfer DML (DTDML). In particular, DTDML learns a sparse combination of the base metrics to construct the target metric by forcing the target metric to be close to an integration of the source metrics. The main advantage of the proposed method compared with existing transfer metric learning approaches is that we directly learn the base metric coefficients instead of the target metric. To this end, far fewer variables need to be learned. We therefore obtain more reliable solutions given the limited side information and the optimization tends to be faster. Experiments on the popular handwritten image (digit, letter) classification and challenge natural image annotation tasks demonstrate the effectiveness of the proposed method.

A Fast Diagonal Distance Metric Learning Approach for Large-Scale Datasets

A Distributed Approach Toward Discriminative Distance Metric Learning

Efficient Distance Metric Learning by Adaptive Sampling and Mini-Batch Stochastic Gradient Descent (SGD)

Towards Making High Dimensional Distance Metric Learning Practical

Scalable Large-Margin Mahalanobis Distance Metric Learning

Distance-Dependent Metric Learning.

Efficient Dual Approach to Distance Metric Learning

Metric Learning with Adaptive Density Discrimination

Fast Generalization Rates for Distance Metric Learning

A scalable algorithm for learning a mahalanobis distance metric

A Scalable Dual Approach to Semidefinite Metric Learning

Decomposition-based transfer distance metric learning for image classification.

Large Scale Metric Learning

Transfer Metric Learning: Algorithms, Applications and Outlooks.

Efficiently Learning a Distance Metric for Large Margin Nearest Neighbor Classification.

Improve distance metric learning by learning positions of class centers

Distance Metric Learning Via Iterated Support Vector Machines

An Efficient Sparse Metric Learning in High-Dimensional Space Via L1-Penalized Log-Determinant Regularization

Data-Adaptive Metric Learning with Scale Alignment.

Efficient Stochastic Optimization for Low-Rank Distance Metric Learning.

Discrete Metric Learning for Fast Image Set Classification