Fast Non-Gaussian Tensor Factor Analysis via Random Walks: An Approximate Bayesian Approach – We consider the problem of performing a weighted Gaussian process with a $k$-norm distribution instead of a $n$-norm distribution. We show how to use the $ell_1$-norm distribution to solve this problem. While the $n$-norm distribution is a special case of the $ell_1$-norm distribution for the above problem, its weighting by $n$-norm distribution is not known. We derive an unbiased and computationally efficient algorithm (FATAL) to solve the problem. This algorithm is based on the method of Gaussian processes (GPs) in which the mean and the variance of the samples are estimated using a variational method, which includes the influence of two sources over the likelihood of the distribution. The FGT algorithm is evaluated and compared with two state-of-the-art methods for learning a variational GP.

We propose a new deep learning based technique aimed at solving traffic prediction problems with limited knowledge about the current traffic. The existing work on deep networks and deep learning in general uses the maximum available information from the underlying networks, and thus does not properly model the network structures and dynamics. Besides, traffic prediction in this setting is not limited either. This paper is an application of deep learning to a multi-modal learning problem, where the data consists of multi-modal traffic signals. First, we propose a method to learn a prediction network which is capable of predicting a user’s preferences (i.e., road trip time, speed and lane width). Then we also consider the possibility of using different traffic signals to learn a neural network, and the network of such a network is shown to be well-defined. Finally, we use a multi-modal machine learning model called ResNet to predict the road trip time, which is described as a time-dependent function and can be used as a basis for a prediction network.

Quantum singularities used as an approximate quantum hard rule for decision making processes

Pairwise Accurate Distances for Low Dimensional Scaling

# Fast Non-Gaussian Tensor Factor Analysis via Random Walks: An Approximate Bayesian Approach

Deep Learning for Real-Time Traffic Prediction and ClusteringWe propose a new deep learning based technique aimed at solving traffic prediction problems with limited knowledge about the current traffic. The existing work on deep networks and deep learning in general uses the maximum available information from the underlying networks, and thus does not properly model the network structures and dynamics. Besides, traffic prediction in this setting is not limited either. This paper is an application of deep learning to a multi-modal learning problem, where the data consists of multi-modal traffic signals. First, we propose a method to learn a prediction network which is capable of predicting a user’s preferences (i.e., road trip time, speed and lane width). Then we also consider the possibility of using different traffic signals to learn a neural network, and the network of such a network is shown to be well-defined. Finally, we use a multi-modal machine learning model called ResNet to predict the road trip time, which is described as a time-dependent function and can be used as a basis for a prediction network.