A Deep Multi-Scale Learning Approach for Person Re-Identification with Image Context – In this paper, we propose a novel deep convolutional neural network (CNN) architecture for annotating images with human-like appearance. The architecture consists of a convolutional layer, which trains to infer human-like appearance, a CNN classifier and an image-to-image fusion model, and layers which train to classify images to images with human-like appearance. As the output of the CNN layer is highly biased, it requires more knowledge of features and pose changes. Thus, we propose to jointly learn more features from the CNN layers and the feature model for annotating images with the human-like appearance. Experimental results show significant improvements in performance over the state-of-the-art, while the human-like appearance annotation has little impact on the annotation accuracy.

This paper presents an end-to-end approach for the semantic parsing of human language using machine learning based tools. The proposed approach is composed of two steps: (i) it learns semantic relationships by learning the representations learned by the human experts and (ii) it provides a natural way for semantic parsers to interact with the human experts. The proposed end-to-end approach is evaluated and compared with a previous state-of-the-art baseline based on Semantic Pattern Recognition (SPR) tasks. The results show that our approach outperforms the other baseline models compared to human experts without sacrificing performance by a large degree.

Learning to Generate Compositional Color Words and Phrases from Speech

The Multidimensional Scaling Solution Revisited: Algorithm and Algorithm Improvement for Graphical Models

A Deep Multi-Scale Learning Approach for Person Re-Identification with Image Context

Deep Learning of Sentences with Low Dimensionality

Generating Semantic Representations using Greedy MethodsThis paper presents an end-to-end approach for the semantic parsing of human language using machine learning based tools. The proposed approach is composed of two steps: (i) it learns semantic relationships by learning the representations learned by the human experts and (ii) it provides a natural way for semantic parsers to interact with the human experts. The proposed end-to-end approach is evaluated and compared with a previous state-of-the-art baseline based on Semantic Pattern Recognition (SPR) tasks. The results show that our approach outperforms the other baseline models compared to human experts without sacrificing performance by a large degree.