Recently, learning equivariant representations has attracted considerable research attention. Dieleman et al. introduce four operations which can be inserted into convolutional neural network to learn deep representations equivariant to rotation. However, feature maps should be copied and rotated four times in each layer in their approach, which causes much running time and memory overhead. In order to address this problem, we propose Deep Rotation Equivariant Network consisting of cycle layers, isotonic layers and decycle layers. Our proposed layers apply rotation transformation on filters rather than feature maps, achieving a speed up of more than 2 times with even less memory overhead. We evaluate DRENs on Rotated MNIST and CIFAR-10 datasets and demonstrate that it can improve the performance of state-of-the-art architectures.

Nearly all recent face recognition algorithms have been evaluated on the Labeled Faces in the Wild (LFW) dataset and many of them achieved over 99% accuracy. However, the performance is still not enough for real-world applications. One problem is the data bias. The faces in LFW and other web-collected datasets come from celebrities. They are quite different from the faces of a normal person captured in the daily life. In other words, they are different in the face distribution. Replacing the training data with the same distribution is a simple solution. However, the photos of common people are much harder to collect because of the privacy concerns. So it is useful to develop a method that transfers the knowledge in the data of different face distribution to help improving the final performance. In this paper, we crawl a large face dataset whose distribution is different from LFW and show the improvement of LFW accuracy with a simple domain adaptation technique. To the best of our knowledge, it is the first time that domain adaptation is applied in the unconstrained face recognition problem with a million scale dataset. Besides, we incorporate face verification threshold into FaceNet triplet loss function explicitly. Finally, we achieve 99.33% on the LFW benchmark with only single CNN model and similar performance even without face alignment.

Community-based question answering platforms have attracted substantial users to share knowledge and learn from each other. As the rapid enlargement of community-based question answering (CQA) platforms, quantities of overlapped questions emerge, which makes users confounded to select a proper reference. It is urgent for us to take effective automated algorithms to reuse historical questions with corresponding answers. In this paper, we focus on the problem with question retrieval, which aims to match historical questions that are relevant or semantically equivalent to resolve one’s query directly. The challenges in this task are the lexical gaps between questions for the word ambiguity and word mismatch problem. Furthermore, limited words in queried sentences cause sparsity of word features. To alleviate these challenges, we propose a novel framework named HSIN which encodes not only the question contents but also the asker’s social interactions to enhance the question embedding performance. More specifically, we apply random walk based learning method with recurrent neural network to match the similarities between asker’s question and historical questions proposed by other users. Extensive experiments on a large-scale dataset from a real world CQA site Quora show that employing the heterogeneous social network information outperforms the other state-of-the-art solutions in this task.

The genome-wide association study (GWAS) is a popular approach to identify disease-associated genetic factors for Alzhemer's Disease (AD). However, it remains challenging because of the small number of samples, very high feature dimensionality and complex structures. To accurately identify genetic risk factors for AD, we propose a novel method based on an in-depth exploration of the hierarchical structure among the features and the commonality across related tasks. Specifically, we first extract and encode the tree hierarchy among features; then, we integrate the tree structures with multi-task feature learning (MTFL) to learn the shared features-that are predictive of AD-among related tasks simultaneously. Thus, we can unify the strength of both the prior structure information and MTFL to boost the prediction performance. However, due to the highly complex regularizer that encodes the tree structure and the extremely high feature dimensionality, the learning process can be computationally prohibitive. To address this, we further develop a novel safe screening rule to quickly identify and remove the irrelevant features before training. Experiment results demonstrate that the proposed approach significantly outperforms the state-of-the-art in detecting genetic risk factors of AD and the speedup gained by the proposed screening can be several orders of magnitude.

One of the most active areas in computer vision is object detection, which has made significant improvement in recent years. Current state-of-the-art object detection methods mostly adhere to the framework of the regions with convolutional neural network (R-CNN). However, they only take advantage of the local appearance features inside object bounding boxes. Since these approaches ignore the contextual information around the object proposals, the outcome of these detectors may generate a semantically incoherent interpretation of the input image. In this paper, we propose a novel object detection system which incorporates the local appearance and the contextual information. Specifically, the contextual information comprises the relationships among objects and the global scene based contextual feature generated by a convolutional neural network. The whole system is formulated as a fully connected conditional random field (CRF) defined on object proposals. Then the contextual constraints among object proposals are modeled as edges naturally. Furthermore, a fast mean field approximation method is utilized to infer in this CRF model efficiently. The experimental results demonstrate that our algorithm achieves a higher mean average precision (mAP) on PASCAL VOC 2007 datasets compared with the baseline algorithm Faster R-CNN.