Multi-Domain Adversarial Learning
This addresses the challenge of handling multiple datasets with overlapping classes and biases in domains like bioimaging, though it appears incremental as it builds on existing adversarial learning methods.
The paper tackled the problem of multi-domain learning (MDL) to minimize average risk across domains, particularly for automated microscopy data with experimental biases, and resulted in improved state-of-the-art performance on two standard image benchmarks and a novel bioimage dataset.
Multi-domain learning (MDL) aims at obtaining a model with minimal average risk across multiple domains. Our empirical motivation is automated microscopy data, where cultured cells are imaged after being exposed to known and unknown chemical perturbations, and each dataset displays significant experimental bias. This paper presents a multi-domain adversarial learning approach, MuLANN, to leverage multiple datasets with overlapping but distinct class sets, in a semi-supervised setting. Our contributions include: i) a bound on the average- and worst-domain risk in MDL, obtained using the H-divergence; ii) a new loss to accommodate semi-supervised multi-domain learning and domain adaptation; iii) the experimental validation of the approach, improving on the state of the art on two standard image benchmarks, and a novel bioimage dataset, Cell.