Attention-based Deep Multiple Instance Learning
This provides an interpretable solution for multiple instance learning problems, particularly useful in domains like medical imaging where understanding instance contributions matters.
The authors tackled the multiple instance learning problem by proposing an attention-based neural network approach that provides interpretable instance contributions while achieving comparable performance to state-of-the-art methods on benchmark datasets and outperforming them on MNIST-based and histopathology datasets.
Multiple instance learning (MIL) is a variation of supervised learning where a single class label is assigned to a bag of instances. In this paper, we state the MIL problem as learning the Bernoulli distribution of the bag label where the bag label probability is fully parameterized by neural networks. Furthermore, we propose a neural network-based permutation-invariant aggregation operator that corresponds to the attention mechanism. Notably, an application of the proposed attention-based operator provides insight into the contribution of each instance to the bag label. We show empirically that our approach achieves comparable performance to the best MIL methods on benchmark MIL datasets and it outperforms other methods on a MNIST-based MIL dataset and two real-life histopathology datasets without sacrificing interpretability.