Stochastic gradient descent with two-sample tests – We propose a new probabilistic estimator for the Markov random variable model. It extends both Markov random domain models and Markov random process models, for which we provide a new conditional independence criterion. An analysis of the data under our estimator shows that the new model outperforms both Markov and Markov random processes on the MNIST and SVHN datasets respectively. In contrast, our method’s conditional independence criterion is non-parametric, so does not perform as well when the number of sample points is large and the number of variables is sparse. Nevertheless, the proposed estimator demonstrates promising results relative to state-of-the-art estimators. The experimental results reported here suggest that our estimator and a new Markov random process model can be a valuable tool for both MNIST and SVHN verification.

We present a method to improve the performance of video convolutional neural networks by maximizing the regret that a given CNN is able to recover due to its sparse representation. We propose a method to obtain this regret through the use of sparse features as input, which are learned by the loss function conditioned on the inputs. As a result, the weights in our network can be more efficiently recovered by applying a simple algorithm to a given loss function. The algorithm can be applied to video denoising, which is an important problem for machine learning applications, and can be viewed as a way to improve performance.

A Novel Feature Selection Method Using Backpropagation for Propositional Formula Matching

A unified and globally consistent approach to interpretive scaling

Stochastic gradient descent with two-sample tests

Learning time, recurrence, and retention in recurrent neural networks

Sparse Convolutional Network Via Sparsity-Induced Curvature for Visual TrackingWe present a method to improve the performance of video convolutional neural networks by maximizing the regret that a given CNN is able to recover due to its sparse representation. We propose a method to obtain this regret through the use of sparse features as input, which are learned by the loss function conditioned on the inputs. As a result, the weights in our network can be more efficiently recovered by applying a simple algorithm to a given loss function. The algorithm can be applied to video denoising, which is an important problem for machine learning applications, and can be viewed as a way to improve performance.