A Study on Word Embeddings in Chinese Word Sense Embeddings – In our dissertation, we discuss the task of translating from Chinese using a low-rank version of WordNet (WordNet). We suggest that this work is a first step towards translating word embeddings in Chinese. This work is a first step towards this goal. In this paper we propose methods to translate word vectors to their high-dimensional representations. To our knowledge, we have not proposed any technique for translating word vectors. In this thesis we will discuss how we can use the high-dimensional features for translation to improve the translation quality of WordNet. We will discuss various techniques that can be used to translate WordNet vectors with high-dimensional features which are commonly used by machine translation systems. To our knowledge, we do not have the knowledge about the algorithm used for translating various word vectors in an end-to-end fashion. So, our work is also a first step towards this goal.

We propose a new probabilistic and regularized Graph model for Graph Embedding (GED) that captures the interplay between the structure, graph, and the form of the data. In our model, the model is designed to maximize the uncertainty involved in embeddings of data, and the embedding is designed to perform minimally important operations for the data. In particular, the embedding can be defined as a set of conditional and undirected graphs, and can be modeled as a non-convex optimization problem. Our experiments show that GED is more accurate than previous SGD models for embedding graph models.

Learning the Block Kernel for Sparse Subspace Analysis with Naive Bayes

Learning Deep Transform Architectures using Label Class Regularized Deep Convolutional Neural Networks

A Study on Word Embeddings in Chinese Word Sense Embeddings

Unsupervised Video Summarization via Deep Learning

Probabilistic and Regularized Graph Models for Graph EmbeddingWe propose a new probabilistic and regularized Graph model for Graph Embedding (GED) that captures the interplay between the structure, graph, and the form of the data. In our model, the model is designed to maximize the uncertainty involved in embeddings of data, and the embedding is designed to perform minimally important operations for the data. In particular, the embedding can be defined as a set of conditional and undirected graphs, and can be modeled as a non-convex optimization problem. Our experiments show that GED is more accurate than previous SGD models for embedding graph models.