Spectral Clamping by Matrix Factorization – With increasingly complex environments, many methods have been proposed to address the problem of object manipulation. However, existing approaches mainly aim at modeling the object motion along with its interactions, such as pose, orientation, etc. In this paper, we propose an unsupervised and fast online method for object manipulation in the visual space. To this end we learn object-level object pose from images and train a convolutional network (CNN) to model the pose-vector representation. The model is trained with object transformations from the objects and the transformations are extracted from the bounding boxes. Our approach, which achieves state-of-the-art accuracy on 3DOF datasets, is based on the idea of learning rich semantic representations from 3D images. Extensive experiments on both synthetic and real images demonstrate that our method is comparable to the baselines, outperforming most methods.

We investigate the use of deep neural network in machine learning. The main focus of this work is on the Deep Belief Network (DBN) which can learn an abstract representation from a low-level, but high-level representation, for classification. DBNs have the capability of learning abstract representations, but learning only the abstract representation is not feasible. We propose a method to learn a dictionary representation by learning the dictionary-level representation. It is shown that the dictionary-level representation achieves some performance improvement with the DBN.

A Comparison of SVM Classifiers for Entity Resolution

Probability Sliding Curves and Probabilistic Graphs

Spectral Clamping by Matrix Factorization

Fast k-means using Differentially Private Low-Rank Approximation for Multi-relational Data

Using Deep Belief Networks to Improve User Response Time PredictionWe investigate the use of deep neural network in machine learning. The main focus of this work is on the Deep Belief Network (DBN) which can learn an abstract representation from a low-level, but high-level representation, for classification. DBNs have the capability of learning abstract representations, but learning only the abstract representation is not feasible. We propose a method to learn a dictionary representation by learning the dictionary-level representation. It is shown that the dictionary-level representation achieves some performance improvement with the DBN.