Unsupervised Domain Adaptation (UDA) is quite challenging due to the large distribution discrepancy between the source domain and the target domain. Inspired by diffusion models which have strong capability to gradually convert data distributions across a large gap, we consider to explore the diffusion technique to handle the challenging UDA task. However, using diffusion models to convert data distribution across different domains is a non-trivial problem as the standard diffusion models generally perform conversion from the Gaussian distribution instead of from a specific domain distribution. Besides, during the conversion, the semantics of the source-domain data needs to be preserved to classify correctly in the target domain. To tackle these problems, we propose a novel Domain-Adaptive Diffusion (DAD) module accompanied by a Mutual Learning Strategy (MLS), which can gradually convert data distribution from the source domain to the target domain while enabling the classification model to learn along the domain transition process. Consequently, our method successfully eases the challenge of UDA by decomposing the large domain gap into small ones and gradually enhancing the capacity of classification model to finally adapt to the target domain. Our method outperforms the current state-of-the-arts by a large margin on three widely used UDA datasets.

GANs are designed to learn a single distribution, though multiple distributions can be modeled by treating them separately. However, this naive implementation does not consider overlapping distributions. We propose Mixed Membership Generative Adversarial Networks (MMGAN) analogous to mixed-membership models that model multiple distributions and discover their commonalities and particularities. Each data distribution is modeled as a mixture over a common set of generator distributions, and mixture weights are automatically learned from the data. Mixture weights can give insight into common and unique features of each data distribution. We evaluate our proposed MMGAN and show its effectiveness on MNIST and Fashion-MNIST with various settings.

Recent work on curvilinear structure segmentation has mostly focused on backbone network design and loss engineering. The challenge of collecting labelled data, an expensive and labor intensive process, has been overlooked. While labelled data is expensive to obtain, unlabelled data is often readily available. In this work, we propose SemiCurv, a semi-supervised learning (SSL) framework for curvilinear structure segmentation that is able to utilize such unlabelled data to reduce the labelling burden. Our framework addresses two key challenges in formulating curvilinear segmentation in a semi-supervised manner. First, to fully exploit the power of consistency based SSL, we introduce a geometric transformation as strong data augmentation and then align segmentation predictions via a differentiable inverse transformation to enable the computation of pixel-wise consistency. Second, the traditional mean square error (MSE) on unlabelled data is prone to collapsed predictions and this issue exacerbates with severe class imbalance (significantly more background pixels). We propose a N-pair consistency loss to avoid trivial predictions on unlabelled data. We evaluate SemiCurv on six curvilinear segmentation datasets, and find that with no more than 5% of the labelled data, it achieves close to 95% of the performance relative to its fully supervised counterpart.

We examine two key questions in GAN training, namely overfitting and mode drop, from an empirical perspective. We show that when stochasticity is removed from the training procedure, GANs can overfit and exhibit almost no mode drop. Our results shed light on important characteristics of the GAN training procedure. They also provide evidence against prevailing intuitions that GANs do not memorize the training set, and that mode dropping is mainly due to properties of the GAN objective rather than how it is optimized during training.

This paper presents the techniques employed in our team's submissions to the 2016 Emotion Recognition in the Wild contest, for the sub-challenge of group-level emotion recognition. The objective of this sub-challenge is to estimate the happiness intensity of groups of people in consumer photos. We follow a predominately bottom-up approach, in which the individual happiness level of each face is estimated separately. The proposed technique is based on geometric features derived from 49 facial points. These features are used to train a model on a subset of the HAPPEI dataset, balanced across expression and headpose, using Partial Least Squares regression. The trained model exhibits competitive performance for a range of non-frontal poses, while at the same time offering a semantic interpretation of the facial distances that may contribute positively or negatively to group-level happiness. Various techniques are explored in combining these estimations in order to perform group-level prediction, including the distribution of expressions, significance of a face relative to the whole group, and mean estimation. Our best submission achieves an RMSE of 0.8316 on the competition test set, which compares favorably to the RMSE of 1.30 of the baseline.