Search

Node classification in structural networks is a longstanding important problem in many real-world applications. Recent studies have shown that network embedding can greatly facilitate node classification by employing embedding algorithms to learn feature representations of nodes. Despite of promising performance, existing network embedding based methods are hard to capture the actual category features of a node because of the linearly inseparable problem in low-dimensional space; meanwhile they cannot incorporate both network structure information and node labels information into the representations simultaneously. To address the above problems, this paper presents a novel Deep Kernel Supervised Hashing (DKSH) method to learn hashing representations of nodes for node classification. Specifically, a deep multiple kernel learning is first employed to map nodes into suitable Hilbert space to deal with linearly inseparable problem. Then, instead of only considering structural similarity between two nodes, a novel similarity matrix is designed to merge both network structure information and node labels information. Supervised by the similarity matrix, the learned hashing representations can preserve the two kinds of information simultaneously from the learned Hilbert space. Extensive experiments show that the proposed method significantly outperforms the state-of-the-art baselines over three real-world benchmark datasets. [ABSTRACT FROM AUTHOR]

Two new maytansinoids, N -methyltreflorine (1) and methyltrewiasine (2), were isolated from the dried fruits of Trewia nudiflora , together with three known congeners (3 – 5). Their structures were elucidated by spectroscopic methods, and the absolute configuration of 1 and 2 was determined by X-ray crystallographic analysis. Compounds 1 – 5 exhibited strong cytotoxicity against human tumor cell lines, including HeLa, MV-4 – 11, and MCF-7, with IC50 values ranging from 0.12 to 11 nM. Compounds 1 and 4 also showed inhibitory activity against the MCF-7/ADR cell line with IC50 values of 13 and 28 nM, respectively. Compounds 1 and 2 significantly inhibited tubulin polymerization in vitro with IC50 values of 3.6 and 3.2 µM, respectively. [ABSTRACT FROM AUTHOR]

This paper presents an efficient CMOS dc-dc converter with fast transient recovery. A fast-transient control operating in conjunction with a linearly scaled gate-driving technique is used to concurrently improve the transient response and light-load efficiency of a dc-dc converter. The controller operates under a pulse-width modulation mode during steady state and enables a saturation mode during transient to attain fast transient response. The linearly scaled gate-driving technique optimizes the gate-driving voltage with respect to the changing load leading to lower gate-driving loss and better light-load efficiency. A prototype chip was implemented using a commercial 0.35-μm CMOS process to validate the proposed techniques. The measurement result shows a 5% increase in light-load efficiency and achieves an overall maximum efficiency of 90%. Moreover, the transient recovery time of a 450 mA step load change is less than 9 μs. [ABSTRACT FROM PUBLISHER]