Neural Embeddings for Sentiment Classification – We present an innovative approach for embedding Chinese and English texts in a unified framework, which is able to process and embed sentences in an end-to-end fashion.
We present a unified framework for automatic feature learning systems. In particular, it is proposed to utilize features from text-to-text to predict a human-level representation of a word in the English language. The proposed approach is based on a recurrent neural network (RNN) that is trained with a novel recurrent language model that represents word vectors over a sequence of text-structured data. Then, the RNN-RNN encodes the data using a convolutional neural network to perform word prediction. Our experiments on a dataset of English sentences show that the proposed method is effective and comparable to RNN-RNN’s performance for predicting speech words in Chinese text. We also describe a new approach for automatic feature learning in sentence embeddings.
We present a method to automatically identify unlabeled and labeled objects from video that are likely to be labeled with a particular label. The identification of such instances is a challenging task in computer vision, which has an interesting dynamic due to multiple factors. To tackle the problem, we propose a joint model framework called K-CNN and N-CNN. Extensive evaluation on a challenging dataset, CIFAR-10 and CIFAR-100, shows that N-CNN outperforms CNN based approaches by a large margin, with near-optimal classification performance.
Learning how to model networks
Fast Riemannian k-means, with application to attribute reduction and clustering
Neural Embeddings for Sentiment Classification
Linear Time Approximation and Spatio-Temporal Optimization for Gaussian Markov Random Fields
Using Deep Learning and Deep Convolutional Neural Networks For Deformable Object RecognitionWe present a method to automatically identify unlabeled and labeled objects from video that are likely to be labeled with a particular label. The identification of such instances is a challenging task in computer vision, which has an interesting dynamic due to multiple factors. To tackle the problem, we propose a joint model framework called K-CNN and N-CNN. Extensive evaluation on a challenging dataset, CIFAR-10 and CIFAR-100, shows that N-CNN outperforms CNN based approaches by a large margin, with near-optimal classification performance.