National University of Defense Technology

About: National University of Defense Technology is a education organization based out in Changsha, China. It is known for research contribution in the topics: Computer science & Radar. The organization has 39430 authors who have published 40181 publications receiving 358979 citations. The organization is also known as: Guófáng Kēxuéjìshù Dàxué & NUDT.

Learning Linear Discriminant Projections for Dimensionality Reduction of Image Descriptors

Abstract: In this paper, we present Linear Discriminant Projections (LDP) for reducing dimensionality and improving discriminability of local image descriptors. We place LDP into the context of state-of-the-art discriminant projections and analyze its properties. LDP requires a large set of training data with point-to-point correspondence ground truth. We demonstrate that training data produced by a simulation of image transformations leads to nearly the same results as the real data with correspondence ground truth. This makes it possible to apply LDP as well as other discriminant projection approaches to the problems where the correspondence ground truth is not available, such as image categorization. We perform an extensive experimental evaluation on standard data sets in the context of image matching and categorization. We demonstrate that LDP enables significant dimensionality reduction of local descriptors and performance increases in different applications. The results improve upon the state-of-the-art recognition performance with simultaneous dimensionality reduction from 128 to 30.

Traffic Sign Recognition Using Kernel Extreme Learning Machines With Deep Perceptual Features

Abstract: Traffic sign recognition plays an important role in autonomous vehicles as well as advanced driver assistance systems. Although various methods have been developed, it is still difficult for the state-of-the-art algorithms to obtain high recognition precision with low computational costs. In this paper, based on the investigation on the influence that color spaces have on the representation learning of convolutional neural network, a novel traffic sign recognition approach called DP-KELM is proposed by using a kernel-based extreme learning machine (KELM) classifier with deep perceptual features. Unlike the previous approaches, the representation learning process in DP-KELM is implemented in the perceptual Lab color space. Based on the learned deep perceptual feature, a kernel-based ELM classifier is trained with high computational efficiency and generalization performance. Through the experiments on the German traffic sign recognition benchmark, the proposed method is demonstrated to have higher precision than most of the state-of-the-art approaches. In particular, when compared with the hinge loss stochastic gradient descent method which has the highest precision, the proposed method can achieve a comparable recognition rate with significantly fewer computational costs.

Video Scene Parsing with Predictive Feature Learning

Abstract: Video scene parsing is challenging due to the following two reasons: firstly, it is non-trivial to learn meaningful video representations for producing the temporally consistent labeling map; secondly, such a learning process becomes more difficult with insufficient labeled video training data. In this work, we propose a unified framework to address the above two problems, which is to our knowledge the first model to employ predictive feature learning in the video scene parsing. The predictive feature learning is carried out in two predictive tasks: frame prediction and predictive parsing. It is experimentally proved that the learned predictive features in our model are able to significantly enhance the video parsing performance by combining with the standard image parsing network. Interestingly, the performance gain brought by the predictive learning is almost costless as the features are learned from a large amount of unlabeled video data in an unsupervised way. Extensive experiments over two challenging datasets, Cityscapes and Camvid, have demonstrated the effectiveness of our model by showing remarkable improvement over well-established baselines.

Rumination and Default Mode Network Subsystems Connectivity in First-episode, Drug-Naive Young Patients with Major Depressive Disorder.

Abstract: Neuroimaging evidence implicates the association between rumination and default mode network (DMN) in major depressive disorder (MDD). However, the relationship between rumination and DMN subsystems remains incompletely understood, especially in patients with MDD. Thirty-three first-episode drug-naive patients with MDD and thirty-three healthy controls (HCs) were enrolled and underwent resting-sate fMRI scanning. Functional connectivity analysis was performed based on 11 pre-defined regions of interest (ROIs) for three DMN subsystems: the midline core, dorsal medial prefrontal cortex (dMPFC) and medial temporal lobe (MTL). Compared with HCs group, patients with MDD exhibited increased within-system connectivity in the dMPFC subsystem and inter-system connectivity between the dMPFC and MTL subsystems. Decreased inter-system connectivity was identified between the midline core and dMPFC subsystem in MDD patients. Depressive rumination was positively correlated with within-system connectivity in the dMPFC subsystem (dMPFC-TempP) and with inter-system connectivity between the dMPFC and MTL subsystems (LTC-PHC). Our results suggest MDD may be characterized by abnormal DMN subsystems connectivity, which may contribute to the pathophysiology of the maladaptive self-focus in MDD patients.

Constructing Differentially 4-Uniform Permutations Over ${\BBF}_{2^{2k}}$ via the Switching Method

Abstract: Many block ciphers use permutations defined on F(22k ) with low differential uniformity, high nonlinearity, and high algebraic degree as their S-boxes to provide confusion. It is well known that, for a function on F(2n), the lowest differential uniformity is 2 and the functions achieving this lower bound are called almost perfect nonlinear (APN) functions. However, due to the lack of knowledge on APN permutations on F(22k ), differentially 4-uniform permutations are usually chosen as S-boxes. For example, the currently endorsed Advanced Encryption Standard chooses one such function, the multiplicative inverse function, as its S-box. By a recent survey on differentially 4-uniform permutations over F(22k ), there are only five known infinite families of such functions, and most of them have small algebraic degrees. In this paper, we apply the powerful switching method to discover many CCZ-inequivalent infinite families of such functions on F(22k ) with optimal algebraic degree, where k is an arbitrary positive integer. This greatly expands the list of differentially 4-uniform permutations and hence provide more choices for the S-boxes. Furthermore, lower bounds for the nonlinearity of the functions obtained in this paper are presented and they imply that some infinite families have high nonlinearity.