Institution
Beihang University
Education•Beijing, China•
About: Beihang University is a education organization based out in Beijing, China. It is known for research contribution in the topics: Control theory & Microstructure. The organization has 67002 authors who have published 73507 publications receiving 975691 citations. The organization is also known as: Beijing University of Aeronautics and Astronautics.
Topics: Control theory, Microstructure, Nonlinear system, Artificial neural network, Feature extraction
Papers published on a yearly basis
Papers
More filters
••
TL;DR: The promising adsorption performance of S(x)-LDH composites for uranyl ions from a variety of aqueous solutions including seawater shows superior selectivity for UO2(2+), over previously reported sorbents.
Abstract: There is a need to develop highly selective and efficient materials for capturing uranium (normally as UO22+) from nuclear waste and from seawater. We demonstrate the promising adsorption performance of Sx-LDH composites (LDH is Mg/Al layered double hydroxide, [Sx]2– is polysulfide with x = 2, 4) for uranyl ions from a variety of aqueous solutions including seawater. We report high removal capacities (qm = 330 mg/g), large KdU values (104–106 mL/g at 1–300 ppm U concentration), and high % removals (>95% at 1–100 ppm, or ∼80% for ppb level seawater) for UO22+ species. The Sx-LDHs are exceptionally efficient for selectively and rapidly capturing UO22+ both at high (ppm) and trace (ppb) quantities from the U-containing water including seawater. The maximum adsorption coeffcient value KdU of 3.4 × 106 mL/g (using a V/m ratio of 1000 mL/g) observed is among the highest reported for U adsorbents. In the presence of very high concentrations of competitive ions such as Ca2+/Na+, Sx-LDH exhibits superior selectivi...
365 citations
••
01 Oct 2019TL;DR: Differentiable soft quantization (DSQ) as mentioned in this paper is proposed to bridge the gap between the full-precision and low-bit networks, which can automatically evolve during training to gradually approximate the standard quantization.
Abstract: Hardware-friendly network quantization (e.g., binary/uniform quantization) can efficiently accelerate the inference and meanwhile reduce memory consumption of the deep neural networks, which is crucial for model deployment on resource-limited devices like mobile phones. However, due to the discreteness of low-bit quantization, existing quantization methods often face the unstable training process and severe performance degradation. To address this problem, in this paper we propose Differentiable Soft Quantization (DSQ) to bridge the gap between the full-precision and low-bit networks. DSQ can automatically evolve during training to gradually approximate the standard quantization. Owing to its differentiable property, DSQ can help pursue the accurate gradients in backward propagation, and reduce the quantization loss in forward process with an appropriate clipping range. Extensive experiments over several popular network structures show that training low-bit neural networks with DSQ can consistently outperform state-of-the-art quantization methods. Besides, our first efficient implementation for deploying 2 to 4-bit DSQ on devices with ARM architecture achieves up to 1.7× speed up, compared with the open-source 8-bit high-performance inference framework NCNN [31].
363 citations
••
04 May 2014
TL;DR: In this article, the authors used multiple instance learning (MIL) framework in classification training with deep learning features and found that automatic feature learning outperformed manual feature learning and achieved performance that's close to fully supervised approach.
Abstract: This paper studies the effectiveness of accomplishing high-level tasks with a minimum of manual annotation and good feature representations for medical images. In medical image analysis, objects like cells are characterized by significant clinical features. Previously developed features like SIFT and HARR are unable to comprehensively represent such objects. Therefore, feature representation is especially important. In this paper, we study automatic extraction of feature representation through deep learning (DNN). Furthermore, detailed annotation of objects is often an ambiguous and challenging task. We use multiple instance learning (MIL) framework in classification training with deep learning features. Several interesting conclusions can be drawn from our work: (1) automatic feature learning outperforms manual feature; (2) the unsupervised approach can achieve performance that's close to fully supervised approach (93.56%) vs. (94.52%); and (3) the MIL performance of coarse label (96.30%) outweighs the supervised performance of fine label (95.40%) in supervised deep learning features.
363 citations
••
TL;DR: This paper develops an energy-efficient, cost-effective, scaled-up corrosion engineering method for transforming inexpensive iron substrates into highly active and ultrastable electrodes for oxygen evolution reaction, and prepares active water-splitting electrocatalysts via corrosion engineering that are stable for thousands of hours.
Abstract: Although a number of nonprecious materials can exhibit catalytic activity approaching (sometimes even outperforming) that of iridium oxide catalysts for the oxygen evolution reaction, their catalytic lifetimes rarely exceed more than several hundred hours under operating conditions. Here we develop an energy-efficient, cost-effective, scaled-up corrosion engineering method for transforming inexpensive iron substrates (e.g., iron plate and iron foam) into highly active and ultrastable electrodes for oxygen evolution reaction. This synthetic method is achieved via a desired corrosion reaction of iron substrates with oxygen in aqueous solutions containing divalent cations (e.g., nickel) at ambient temperature. This process results in the growth on iron substrates of thin film nanosheet arrays that consist of iron-containing layered double hydroxides, instead of rust. This inexpensive and simple manufacturing technique affords iron-substrate-derived electrodes possessing excellent catalytic activities and activity retention for over 6000 hours at 1000 mA cm-2 current densities.
362 citations
••
TL;DR: This brief investigates the finite-time control problem associated with attitude stabilization of a rigid spacecraft subject to external disturbance, actuator faults, and input saturation and develops a novel fixed-time sliding mode surface, and the settling time of the defined surface is shown to be independent of the initial conditions of the system.
Abstract: This brief investigates the finite-time control problem associated with attitude stabilization of a rigid spacecraft subject to external disturbance, actuator faults, and input saturation. More specifically, a novel fixed-time sliding mode surface is developed, and the settling time of the defined surface is shown to be independent of the initial conditions of the system. Then, a finite-time controller is derived to guarantee that the closed-loop system is stable in the sense of the fixed-time concept. The actuator-magnitude constraints are rigorously enforced and the attitude of the rigid spacecraft converges to the equilibrium in a finite time even in the presence of external disturbances and actuator faults. Numerical simulations illustrate the spacecraft performance obtained using the proposed controller.
361 citations
Authors
Showing all 67500 results
Name | H-index | Papers | Citations |
---|---|---|---|
Yi Chen | 217 | 4342 | 293080 |
H. S. Chen | 179 | 2401 | 178529 |
Alan J. Heeger | 171 | 913 | 147492 |
Lei Jiang | 170 | 2244 | 135205 |
Wei Li | 158 | 1855 | 124748 |
Shu-Hong Yu | 144 | 799 | 70853 |
Jian Zhou | 128 | 3007 | 91402 |
Chao Zhang | 127 | 3119 | 84711 |
Igor Katkov | 125 | 972 | 71845 |
Tao Zhang | 123 | 2772 | 83866 |
Nicholas A. Kotov | 123 | 574 | 55210 |
Shi Xue Dou | 122 | 2028 | 74031 |
Li Yuan | 121 | 948 | 67074 |
Robert O. Ritchie | 120 | 659 | 54692 |
Haiyan Wang | 119 | 1674 | 86091 |