Training of Convolutional Neural Networks – As a powerful tool, deep learning can be used to discover the underlying structure of a computer’s input, and thus to model the dynamics of the input. In this work, we develop an iterative strategy for the deep learning to map input states into the input, as well as an iterative strategy for learning the output structure. To achieve this goal, in this work we construct an ensemble of deep network models, with weights on each model. Experimental results demonstrate that the weights have significantly different roles in the output structure and learned weights are more effective than other weights when applied to the same task.

We present a framework for predicting the trajectory of a moving object from its point of interest and then inferring whether the object moved or not. Unlike other existing methods relying on hand-crafted and learned features, our approach relies on the model-structured representations of the objects’ motion trajectories. To efficiently learn the model-structured representations, we propose a neural network based on convolutional neural networks (CNN). We first use a set of trajectories in a model to model the trajectory. Each trajectory is estimated with a convolutional neural network, which is fed a set of hand-crafted features. Then, a trajectory model is constructed that exploits those trajectories to infer which trajectories are associated with the object, and which trajectories are associated with the trajectory. Based on this learning, we propose two learning algorithms and their optimization strategies. We also propose an online learning procedure to automatically update trajectories and achieve better target locations. The algorithm is evaluated on both standard and synthetic datasets.

Training a Sparse Convolutional Neural Network for Receptive Field Detection

Modeling and Analysis of Non-Uniform Graphical Models as Bayesian Models

# Training of Convolutional Neural Networks

Fully Automatic Saliency Prediction from Saline Walors

Bidirectional Multiple Attractor Learning for Multi-Target Tracking and TrackingWe present a framework for predicting the trajectory of a moving object from its point of interest and then inferring whether the object moved or not. Unlike other existing methods relying on hand-crafted and learned features, our approach relies on the model-structured representations of the objects’ motion trajectories. To efficiently learn the model-structured representations, we propose a neural network based on convolutional neural networks (CNN). We first use a set of trajectories in a model to model the trajectory. Each trajectory is estimated with a convolutional neural network, which is fed a set of hand-crafted features. Then, a trajectory model is constructed that exploits those trajectories to infer which trajectories are associated with the object, and which trajectories are associated with the trajectory. Based on this learning, we propose two learning algorithms and their optimization strategies. We also propose an online learning procedure to automatically update trajectories and achieve better target locations. The algorithm is evaluated on both standard and synthetic datasets.