Nearest Local Average Post-Processing for Online Linear Learning – We propose a method for online linear learning. The linear learning algorithm is based on a random walk algorithm where the objective is to minimize the sum of all the weights that are positive, and each positive weight is estimated in advance. In the linear learning setting, the objective is to find the least sum of all weight vectors that fit a non-negative matrix. The algorithm is efficient, easy to implement and generalizable. On the other hand, the linear learning algorithm is not very suitable for a data-driven learning environment. We prove that the linear learning algorithm is a non-linear learning formulation within an online learning framework. The formulation is a matrix-based linear algorithm which is not suitable for use in a data-driven setting. The implementation of the algorithm requires some computations and is not suitable for a data-driven setting. We demonstrate that the linear learning algorithm can be improved by a linear learning algorithm.

We present a novel approach for learning deep neural networks (DNNs) on-the-fly. The approach addresses two distinct challenges: (1) is the DNN not only trained and optimized for all inputs at each time step, but also all layers are trained in all layers and learn to discriminate between inputs in a coherent representation; and (2) is the DNN trained on the learned representations of the input. The DNN training is accomplished by using a deep architecture and utilizes the data structure to capture the learned discriminative representation of the input, which is then used to train a DNN with the discriminative representation. Experiments on various challenging datasets demonstrate that our approach outperforms the state-of-the-art deep neural network architectures.

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

MorphNet: A Python-based Entity Disambiguation Toolkit

# Nearest Local Average Post-Processing for Online Linear Learning

Multi-view Deep Reinforcement Learning with Dynamic CodingWe present a novel approach for learning deep neural networks (DNNs) on-the-fly. The approach addresses two distinct challenges: (1) is the DNN not only trained and optimized for all inputs at each time step, but also all layers are trained in all layers and learn to discriminate between inputs in a coherent representation; and (2) is the DNN trained on the learned representations of the input. The DNN training is accomplished by using a deep architecture and utilizes the data structure to capture the learned discriminative representation of the input, which is then used to train a DNN with the discriminative representation. Experiments on various challenging datasets demonstrate that our approach outperforms the state-of-the-art deep neural network architectures.