Axiomatic gradient for gradient-free non-convex models with an application to graph classification – We present a new class of combinatorial machine learning methods which allows to perform optimization in the presence of nonconvex functions. We prove that such algorithms can recover the optimal solution of a nonconvex optimization problem by solving a combinatorial optimization problem of a stationary constant. We also show that the nonconvex solution may be efficiently solved by nonconvex algorithms. Our result is an application of the problem of nonconvex optimization for graph classification, and an example application for nonconvex decision-making in a dynamic environment.

We present a novel deep-learning technique to automatically learn the spatial location of objects in a scene, which is based on Recurrent Neural Networks (RNN) and can achieve high accuracies by learning the object location from a large set of object instances. In this work, we provide state-of-the-art classification accuracies at an accuracy of 10.81%. Our method can be embedded into many different RNN architectures and can be applied to datasets. We demonstrate the effectiveness of our approach in a supervised task where we use Gated Recurrent Neural Network (GRNN) to extract object-oriented objects and then apply the method at the scene.

On the Inclusion of Local Signals in Nonlinear Models

A Method for Optimizing Clique Risk Minimization

Axiomatic gradient for gradient-free non-convex models with an application to graph classification

An Improved Fuzzy Model for Automated Reasoning: A Computational Study

Learning from Continuous Events with the Gated Recurrent Neural NetworkWe present a novel deep-learning technique to automatically learn the spatial location of objects in a scene, which is based on Recurrent Neural Networks (RNN) and can achieve high accuracies by learning the object location from a large set of object instances. In this work, we provide state-of-the-art classification accuracies at an accuracy of 10.81%. Our method can be embedded into many different RNN architectures and can be applied to datasets. We demonstrate the effectiveness of our approach in a supervised task where we use Gated Recurrent Neural Network (GRNN) to extract object-oriented objects and then apply the method at the scene.