Variational Learning of Probabilistic Generators – Learning a large class of estimators (e.g., Gaussian process models) is a challenging problem. For the past decade, there has been much interest in generating estimators that achieve consistent improvement. In this work, we consider the problem of learning an estimator for a large class of estimators. In this paper we propose a novel estimator for several large class of estimators including Markov chains and conditional random fields. We use a modified version of the Residual Recurrent Neural Network (RRCNN) model, which is able to learn a conditional probability density estimator from data, without relying on the input of any estimator. Our model achieves state-of-the-art performance and is able to achieve better performance with less computation with the same model complexity. We apply our algorithm to a variety of large data sets generated by Bayesian networks and to a large-scale model classification problem.

Despite the recent success of learning structured classifiers, the main challenge is to find the right balance between classification performance and training data quality, which in turn requires large amounts of manual annotation. Many previous efforts to address the difficulty of labeling training examples in a single action, using machine learning, have focused on dealing with a single task. However, learning a complex feature vector of a data set can be time consuming, and to deal with it, feature vectors are often pre-trained to do the same task. In this work, we address these issues by leveraging deep semantic learning to extract more complex features from a dataset for classification tasks. In particular, we design a novel framework to extract semantic feature predictions with the goal of reducing the computational cost of feature extraction. We demonstrate how this approach can speed up the classification process by up to an order of magnitude.

Highly Scalable Bayesian Learning of Probabilistic Programs

Graph Clustering and Adaptive Bernoulli Processes

Variational Learning of Probabilistic Generators

Learning TWEless: Learning Hierarchical Features with Very Deep Neural Networks

Learning a Modular Deep Learning Model with Online CorrectionDespite the recent success of learning structured classifiers, the main challenge is to find the right balance between classification performance and training data quality, which in turn requires large amounts of manual annotation. Many previous efforts to address the difficulty of labeling training examples in a single action, using machine learning, have focused on dealing with a single task. However, learning a complex feature vector of a data set can be time consuming, and to deal with it, feature vectors are often pre-trained to do the same task. In this work, we address these issues by leveraging deep semantic learning to extract more complex features from a dataset for classification tasks. In particular, we design a novel framework to extract semantic feature predictions with the goal of reducing the computational cost of feature extraction. We demonstrate how this approach can speed up the classification process by up to an order of magnitude.