Multi-View Representation Lasso through Constrained Random Projections for Image Recognition – We present a novel multi-view feature representation learning method for automatic segmentation of facial landmarks in images. We show that the proposed algorithm outperforms baseline approaches, with significant improvement of performance compared to the traditional approach. Additionally, we present a new benchmark dataset for automatically segmenting landmarks in images at human and machine levels using multi-view convolutional neural networks. Extensive evaluation on two standard benchmark datasets for facial landmarks segmentation shows that our framework significantly outperforms baseline approaches.

We propose a novel, principled strategy of learning sparse representations that is well suited in domains with large amounts of labels and features. We also develop a learning algorithm that can be generalized to sparse representations and then to any sparse representation. We also design and evaluate a large-scale learning algorithm for learning sparse representations without the need to explicitly label them. We demonstrate the power of the proposed algorithm on large-scale and distributed dataset of image classification problems with labeled and unlabeled data.

A Deep Learning Approach for Image Retrieval: Estimating the Number of Units Segments are Unavailable

On the Scope of Emotional Matter and the Effect of Language in Syntactic Translation

Multi-View Representation Lasso through Constrained Random Projections for Image Recognition

Segmental Low-Rank Matrix Estimation from Pairwise Similarities via Factorized Matrix Factorization

An Adaptive Classification Method for Learning Sparse RepresentationsWe propose a novel, principled strategy of learning sparse representations that is well suited in domains with large amounts of labels and features. We also develop a learning algorithm that can be generalized to sparse representations and then to any sparse representation. We also design and evaluate a large-scale learning algorithm for learning sparse representations without the need to explicitly label them. We demonstrate the power of the proposed algorithm on large-scale and distributed dataset of image classification problems with labeled and unlabeled data.