SetFlow: Generating Structured Sets of Representations for Multiple Instance Learning

TL;DR

SetFlow is a novel generative model that creates permutation-invariant, semantically coherent sets of representations for MIL data, improving classification performance and data augmentation in data-scarce scenarios.

Contribution

It introduces SetFlow, a flow-based architecture that models entire MIL bags in the representation space, capturing intra-bag dependencies and enabling effective data augmentation.

Findings

Generated samples closely match original data distribution.

Synthetic data training yields competitive classification results.

SetFlow improves downstream MIL classification performance.

Abstract

Data scarcity and weak supervision continue to limit the performance of machine learning models in many real-world applications, such as mammography, where Multiple Instance Learning (MIL) often offers the best formulation. While recent foundation models provide strong semantic representations out of the box, effective augmentation of such representations of MIL data remains limited, as existing methods operate at the instance level and fail to capture intra-bag dependencies. In this work, we introduce SetFlow, a generative architecture that models entire MIL bags (i.e., sets) directly in the representation space. Our approach leverages the flow matching paradigm combined with a Set Transformer-inspired design, enabling it to handle permutation-invariant inputs while capturing interactions between instances within each bag. The model is conditioned on both class labels and input scale, allowing it to generate coherent and semantically consistent sets of representations. We evaluate SetFlow on a large-scale mammography benchmark using a state-of-the-art MIL-PF classification pipeline. The generated samples are shown to closely match the original data distribution and even improve downstream performance when used for augmentation. Furthermore, training on synthetic data alone shows competitive results, demonstrating the effectiveness of representation-space generative modeling for data-scarce and privacy-sensitive tasks.