Adaptive Learning of Cross-Agent Recommendations with Arbitrary Reward Sets – A new framework for multi-agent decision making is presented, in which agents are asked to provide an arbitrary reward set, but have no information about how it is chosen. Agents may be asked to learn the policy jointly with the other agents in order to make the best decision, and may decide among them based on the outcomes of rewards. This framework is based on minimizing the cost function and on making the best decision, and also on minimizing the total expected reward reward given the rewards of all the agents (i.e., the number of agents) plus the reward of the other agents. The framework is applied to a variety of multi-agent decision making scenarios, including decision making tasks where one agent is asked to maximize the rewards of all the other agents, and in situations in which agents are already engaged in cooperation, which may include decision making tasks where the other agents are not able to provide rewards, or where agents are not actively engaged in decisions to learn the policy jointly with the other agents.

Sparse-time classification (STR) has emerged as a promising tool for automatic vehicle identification. The main drawback of STR is its lack of training data and the difficulty of handling noisy data. In this work we present an innovative approach to the problem using Convolutional Neural Networks. In our model, we first use unsupervised learning as feature representation for image classification: the Convolutional Neural Network (CNN) is trained with an unlabeled image. The CNN learns a binary metric feature embedding representation of its output vectors (e.g., the k-dimensional). Following this representation, the CNN can model the training data by selecting a high-quality subset of the training data. Our method learns the representations and, by using the learned representations, can be used with the standard segmentation and classification algorithms in order to learn the feature representation for the given dataset. We evaluate our method on the challenging TIDA dataset and compare it to the state-of-the-arts.

Learning from Continuous Events with the Gated Recurrent Neural Network

On a Generative Baseline for Modeling Clinical Trials

Adaptive Learning of Cross-Agent Recommendations with Arbitrary Reward Sets

Feature Selection with Stochastic Gradient Descent in Nonconvex and Nonconjugate Linear Models

Machine Learning Methods for Multi-Step Traffic AcquisitionSparse-time classification (STR) has emerged as a promising tool for automatic vehicle identification. The main drawback of STR is its lack of training data and the difficulty of handling noisy data. In this work we present an innovative approach to the problem using Convolutional Neural Networks. In our model, we first use unsupervised learning as feature representation for image classification: the Convolutional Neural Network (CNN) is trained with an unlabeled image. The CNN learns a binary metric feature embedding representation of its output vectors (e.g., the k-dimensional). Following this representation, the CNN can model the training data by selecting a high-quality subset of the training data. Our method learns the representations and, by using the learned representations, can be used with the standard segmentation and classification algorithms in order to learn the feature representation for the given dataset. We evaluate our method on the challenging TIDA dataset and compare it to the state-of-the-arts.