Out-of-distribution learning is concerned with developing methods where the distribution of the data processed at test time may be different from the distribution of the data used during the training of a model. These techniques go beyond the simple fine-tuning of models. They include techniques such as domain adaptation (Motiian et al., 2017) (Motiian et al., 2017), where models undergo additional training with special losses to adapt to novel training data distributions. OOD techniques enable the deployment of models in open-set settings, where there is the need to identify whether an input data sample belongs to the training distribution or not, by detecting novelties (Pidhorskyi et al., 2018). OOD techniques aim to go beyong domain generalization (Motiian et al., 2017) where models are made robust to any distortion of the training data distribution, and attempt to produce models that can generalize the original task to something new, sometimes in a compositional fashion (Yao & Doretto, 2010) by leveraging what the model has learned so far. We apply these techniques to several computer vision tasks, including object detection, objects clssification, medical imaging applications, and immersive analytics.
- NeurIPSGenerative probabilistic novelty detection with adversarial autoencoders In Neural Information Processing Systems, 2018.
- NIPSFew-shot adversarial domain adaptation In Advances in Neural Information Processing Systems, 2017.
- ICCVUnified deep supervised domain adaptation andgeneralization In Proceedings of IEEE International Conference on Computer Vision, 2017.
- CVPRBoosting for transfer learning with multiple sources In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2010.