We propose discriminative adversarial networks (DAN) for semi-supervised learning and loss function learning. Our DAN approach builds upon generative adversarial networks (GANs) and conditional GANs but includes the key differentiator of using two discriminators instead of a generator and a discriminator. DAN can be seen as a framework to learn loss functions for predictors that also implements...

Topic model is a popular tool for visual concept learning. Most topic models are either unsupervised or fully supervised. In this paper, to take advantage of both limited labeled training images and rich unlabeled images, we propose a novel regularized Semi-Supervised Latent Dirichlet Allocation (r-SSLDA) for learning visual concept classifiers. Instead of introducing a new complex topic model,...

In computer vision, fine-grained classification has become an important issue in recognizing objects with slight visual differences. Usually, it is challenging to generate good performance when solving problems using traditional convolutional neural networks. To improve the accuracy and training time of networks problems, this paper proposes a tree-structured framework by eliminating effect dif...

High-dimensional regression problems are becoming more and more common with emerging technologies. However, in many cases data are constrained to a low dimensional manifold. The information about the output is hence contained in a much lower dimensional space, which can be expressed by an intrinsic description. By rst nding the intrinsic description, a low dimensional mapping can be found to gi...

We propose a weakly-supervised framework for action labeling in video, where only the order of occurring actions is required during training time. The key challenge is that the per-frame alignments between the input (video) and label (action) sequences are unknown during training. We address this by introducing the Extended Connectionist Temporal Classification (ECTC) framework to efficiently e...

Recently the research on supervised term weighting has attracted growing attention in the field of Traditional Text Categorization (TTC) and Sentiment Analysis (SA). Despite their impressive achievements, we show that existing methods more or less suffer from the problem of over-weighting. Overlooked by prior studies, over-weighting is a new concept proposed in this paper. To address this probl...

Wikipedia is a Web based, freely available multilingual encyclopedia, constructed in a collaborative effort by thousands of contributors. Wikipedia articles on the same topic in different languages are connected via interlingual (or translational) links. These links serve as an excellent resource for obtaining lexical translations, or building multilingual dictionaries and semantic networks. As...

Recent advances in wearable devices and Internetof-Things (IoT) have led to massive growth sensor data generated edge devices. Labeling such for classification tasks has proven be challenging. In addition, by different users bear various personal attributes heterogeneity, rendering it impractical develop a global model that adapts well all users. Concerns over privacy communication costs also p...