A Unified Batch Selection Policy for Active Metric Learning
This addresses the challenge of efficient data annotation in metric learning for applications like image retrieval, though it is incremental as it builds on prior batch decorrelation strategies.
The paper tackles the problem of selecting batches of training data in active metric learning to avoid redundancy and improve efficiency, presenting a method that uses maximum joint entropy and achieves provable optimality with consistent outperformance of state-of-the-art methods in experiments.
Active metric learning is the problem of incrementally selecting high-utility batches of training data (typically, ordered triplets) to annotate, in order to progressively improve a learned model of a metric over some input domain as rapidly as possible. Standard approaches, which independently assess the informativeness of each triplet in a batch, are susceptible to highly correlated batches with many redundant triplets and hence low overall utility. While a recent work \cite{kumari2020batch} proposes batch-decorrelation strategies for metric learning, they rely on ad hoc heuristics to estimate the correlation between two triplets at a time. We present a novel batch active metric learning method that leverages the Maximum Entropy Principle to learn the least biased estimate of triplet distribution for a given set of prior constraints. To avoid redundancy between triplets, our method collectively selects batches with maximum joint entropy, which simultaneously captures both informativeness and diversity. We take advantage of the submodularity of the joint entropy function to construct a tractable solution using an efficient greedy algorithm based on Gram-Schmidt orthogonalization that is provably $\left( 1 - \frac{1}{e} \right)$-optimal. Our approach is the first batch active metric learning method to define a unified score that balances informativeness and diversity for an entire batch of triplets. Experiments with several real-world datasets demonstrate that our algorithm is robust, generalizes well to different applications and input modalities, and consistently outperforms the state-of-the-art.