Beijing University of Technology

About: Beijing University of Technology is a education organization based out in Beijing, Beijing, China. It is known for research contribution in the topics: Microstructure & Laser. The organization has 31929 authors who have published 31987 publications receiving 352112 citations. The organization is also known as: Běijīng Gōngyè Dàxué & Beijing Polytechnic University.

Multiscale CNNs for Brain Tumor Segmentation and Diagnosis

Abstract: Early brain tumor detection and diagnosis are critical to clinics. Thus segmentation of focused tumor area needs to be accurate, efficient, and robust. In this paper, we propose an automatic brain tumor segmentation method based on Convolutional Neural Networks (CNNs). Traditional CNNs focus only on local features and ignore global region features, which are both important for pixel classification and recognition. Besides, brain tumor can appear in any place of the brain and be any size and shape in patients. We design a three-stream framework named as multiscale CNNs which could automatically detect the optimum top-three scales of the image sizes and combine information from different scales of the regions around that pixel. Datasets provided by Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) organized by MICCAI 2013 are utilized for both training and testing. The designed multiscale CNNs framework also combines multimodal features from T1, T1-enhanced, T2, and FLAIR MRI images. By comparison with traditional CNNs and the best two methods in BRATS 2012 and 2013, our framework shows advances in brain tumor segmentation accuracy and robustness.

Comment on quantum private comparison protocols with a semi-honest third party

Abstract: As an important branch of quantum cryptography, quantum private comparison (QPC) has recently received a lot of attention. In this paper we study the security of previous QPC protocols with a semi-honest third party (TP) from the viewpoint of secure multi-party computation and show that the assumption of a semi-honest TP is unreasonable. Without the unreasonable assumption of a semi-honest TP, one can easily find that the QPC protocol (Tseng et al. in Quantum Inf Process, 2011, doi: 10.1007/s11128-011-0251-0 ) has an obvious security flaw. Some suggestions about the design of QPC protocols are also given.

Descope of the ALIA mission

Abstract: The present work reports on a feasibility study commissioned by the Chinese Academy of Sciences of China to explore various possible mission options to detect gravitational waves in space alternative to that of the eLISA/LISA mission concept. Based on the relative merits assigned to science and technological viability, a few representative mission options descoped from the ALIA mission are considered. A semi-analytic Monte Carlo simulation is carried out to understand the cosmic black hole merger histories and the possible scientific merits of the mission options in probing the light seed black holes and their coevolution with galaxies in early Universe. The study indicates that, by choosing the armlength of the interferometer to be three million kilometers and shifting the sensitivity floor to around one-hundredth Hz, together with a very moderate improvement on the position noise budget, there are certain mission options capable of exploring light seed, intermediate mass black hole binaries at high redshift that are not readily accessible to eLISA/LISA, and yet the technological requirements seem to within reach in the next few decades for China.

Deep Neural Network Technique for High-Dimensional Microwave Modeling and Applications to Parameter Extraction of Microwave Filters

Abstract: This article introduces the deep neural network method into the field of high-dimensional microwave modeling. Deep learning is nowadays highly successful in solving complex and challenging pattern recognition and classification problems. This article investigates the use of deep neural networks to solve microwave modeling problems that are much more challenging than that solved by the previous shallow neural networks. The most commonly used activation function in the existing deep neural network is the rectified linear unit (ReLU), which is a piecewise hard switch function. However, such a ReLU is not suitable for microwave modeling where the input–output relationships are smooth and continuous. In this article, we propose a new deep neural network to perform high-dimensional microwave modeling. A smooth ReLU is proposed for the new deep neural network. The proposed deep neural network employs both the sigmoid function and the smooth ReLU as activation functions. The new deep neural network can represent the smooth input–output relationship that is required for microwave modeling. An advanced three-stage deep learning algorithm is proposed to train the new deep neural network model. This algorithm can determine the number of hidden layers with sigmoid functions and those with smooth ReLUs in the training process. It can also overcome the vanishing gradient problem for training the deep neural network. The proposed deep neural network technique can solve microwave modeling problems in a higher dimension than the previous neural network method, i.e., shallow neural network method. Two high-dimensional parameter-extraction modeling examples of microwave filters are presented to demonstrate the proposed deep neural network technique.