Hierarchical Gaussian Process Models – Learning to predict future events is challenging because of the large, complex, and unpredictable nature of the data. Despite the enormous volume of available data, supervised learning has made great progress in recent years in learning to predict the future rather than in predicting the past. In this paper, we present a framework for modeling and predicting the future of data by non-Gaussian prior approximating latent Gaussian processes. The underlying assumptions are to be established in the context of non-Gaussian prior approximating learning, and we further elaborate on these assumptions in a neural-network architecture. We evaluate this network on two datasets: the Long Short-Term Memory and the Stanford Attention Framework dataset, where we show that the model achieves state-of-the-art performance with good accuracy.
This paper presents a simple model-based approach for predicting future facial poses by combining a pair of convolutional-based deep Convolutional Neural Networks (CNNs). Our approach outperforms previous models that use only a single convolutional-bijection network to achieve accurate detection of facial pose. In addition, we show that it is possible to perform a CNN to predict future pose with small training samples. The proposed approach is applicable to different applications, including face recognition, face localization, object manipulation, gesture recognition, and recognition of human head pose from multiple sources.
Structural Matching through Reinforcement Learning
Eliminating Dither in RGB-based 3D Face Recognition with Deep Learning: A Unified Approach
Hierarchical Gaussian Process Models
Convolutional Residual Learning for 3D Human Pose Estimation in the Wild
Unsupervised Learning of Depth and Background Variation with Multi-scale ScalingThis paper presents a simple model-based approach for predicting future facial poses by combining a pair of convolutional-based deep Convolutional Neural Networks (CNNs). Our approach outperforms previous models that use only a single convolutional-bijection network to achieve accurate detection of facial pose. In addition, we show that it is possible to perform a CNN to predict future pose with small training samples. The proposed approach is applicable to different applications, including face recognition, face localization, object manipulation, gesture recognition, and recognition of human head pose from multiple sources.