JECL: Joint Embedding and Cluster Learning for Image-Text Pairs
This addresses the problem of clustering unstructured image-text data for applications where structured training data is expensive, offering an incremental improvement over existing methods.
The paper tackled clustering image-caption pairs by proposing JECL, a method that jointly learns embeddings and cluster assignments using regularized objectives, resulting in outperforming single-view and multi-view methods on large benchmarks with robustness to missing captions and varying data sizes.
We propose JECL, a method for clustering image-caption pairs by training parallel encoders with regularized clustering and alignment objectives, simultaneously learning both representations and cluster assignments. These image-caption pairs arise frequently in high-value applications where structured training data is expensive to produce, but free-text descriptions are common. JECL trains by minimizing the Kullback-Leibler divergence between the distribution of the images and text to that of a combined joint target distribution and optimizing the Jensen-Shannon divergence between the soft cluster assignments of the images and text. Regularizers are also applied to JECL to prevent trivial solutions. Experiments show that JECL outperforms both single-view and multi-view methods on large benchmark image-caption datasets, and is remarkably robust to missing captions and varying data sizes.