Search

E-commerce marketing forces are taking advantage of microblogs to deliver their advertisements to promote product information. The success of product information diffusion in microblog depends greatly on user behaviors -- browsing, commenting and reposting. In this paper, we divide user behaviors of Sina Weibo into four types corresponding to four different colors, and propose a method to predict user behavior based on DCGAN (Deep Convolutional Generative Adversarial Nets). By analyzing a real Sina Weibo dataset, the experimental results show that the prediction accuracy of the four types of user behaviors reaches more than 80%, which proves that our method is feasible and effective, and also can help companies succeed in their product advertisements. &nbsp

Deep convolution neural networks have been widely studied and applied in many computer vision tasks. However, they are commonly treated as black-boxes and plagued by the inexplicability. In this paper, we propose a novel method to visually interpret the convolutional neural network in the field of image classification. Our method is capable of generating fine-grained and class discriminative heatmap that highlights the important input features contributing to specific predictions. Specifically, through the combination of the modified deconvolution and the pixel-wise Grad-CAM, the fine-grained heatmap and discriminative mask can be fused to achieve fine-grained deconvolution characteristics, and retain the class discriminativeness of the Grad-CAM, enhancing the interpretation effect of the heatmap. Both qualitative and quantitative experiments on ILSVRC 2012 dataset and PASCAL VOC 2012 dataset are conducted. The results indicate that the proposed method achieves a better visual effect with less noise in comparison to the previous methods, especially for visualising small objects in simple contexts. Furthermore, this method can realize a moderately effective performance on weakly supervised instance segmentation tasks, whereas the existing methods only work for weakly supervised object localisation.

The Internet of Things (IoT) forms a foundation for cyber-physical systems. We propose an efficient and secure authentication scheme for machine-to-machine (M2M) networks in IoT enabled cyber-physical systems. Smart objects and smart devices over CPS are capable of capturing a variety of multimedia contents; interact with each other and also with the physical world in a fully automatic manner without human interference. The proposed scheme allows any pair of entities in an M2M network to mutually authenticate each other and agree on a session key for communicating data in a secure and efficient way. The authentication process does not incorporate the M2M service provider, and hence eliminates the burden of managing the authentication of massive scale devices at the edge of the network. The burden of the authentication process is offloaded and distributed on the gateways under the authority of this M2M service provider. The proposed scheme requires the mobile user to hold only one secret key provided by the M2M service provider, by which, he can roam randomly in the M2M network and authenticate to any of the gateways in the domain. Then, this authenticated gateway allows the mobile user to authenticate with any sensor node in the domain. In the proposed scheme, the authentication process does not rely on any public key cryptographic operations. Authentication is achieved using very few hash invocations and symmetric key encryptions. Therefore, the scheme is suitable for environmental sensors which are limited in resources (computation, storage, and energy). We analyze the security of the proposed scheme using BAN logic, which is widely accepted as a framework for the assessment of authentication protocols and also using ProVerif. We assess the efficiency of the proposed scheme and compare with some recently proposed schemes.

Renewable energy power prediction is crucial to economic dispatch and reliable operation of power systems. This paper proposes a wind power forecasting approach based on the Extreme Gradient Boosting (XGBoost) algorithm. XGBoost is not only an effective feature selection method but also an accurate forecasting approach. In order to avoid excessive manual interventions for hyperparameter tuning, the Tree-Structured Parzen Estimator (TPE) model is presented to optimize the hyperparameters of XGBoost. This forecasting strategy has been tested in a real wind farm in Spain, compared with Persistence and Support Vector Regression (SVR). The results show that the XGBoost algorithm has higher accuracy and is a novel effective approach for very short-term wind power prediction.

With the continuous increase of economic growth and load demand, the contradiction between source and load has gradually intensified, and the energy storage application demand has become increasingly prominent. Based on the installed capacity of the energy storage power station, the optimization design of the series-parallel configuration of each energy storage unit in the power station has become a top priority. Currently, the failure cost is rarely considered during planning and analyzing on internal structure of energy storage power stations. This study deals with optimization design of the series and parallel configuration of internal energy storage units in energy storage power stations. Besides equipment cost and operation and maintenance cost, failure cost and commissioning cost is considered in the study. The impact of equipment failure cost on the total cost of different configurations is focused on once the energy storage unit is integrated to the power station. And energy storage unit arrangement of the station configuration is optimized with the minimum total cost as the goal. Finally, case study based on an energy storage station to be built in Kunshan, China is presented to verify the feasibility and effectiveness of proposed configuration optimization.

Person re-identification (Re-ID) based on deep learning has made great progress and achieved state-of-the-art performance in recent years. However, the end-to-end properties of deep neural networks allow us to directly feedback the output results based on its input, making the inner working mechanism of the deep person Re-ID model and its decision reasons lack of transparency and explainability. This further impedes improvements to pedestrian recognition performance. As feature visualization has been proven to be an effective method for characterizing the middle layer of a neural network, we propose a novel gradient-based visualization method to interpret the internal features learned by deep person Re-ID. Based on the idea of transfer learning, this model regards the pretrained ResNet-50 on the ImageNet dataset as a basic network for deep person Re-ID. First, the network is fine-tuned on the person Re-ID dataset to achieve pedestrian classification, and then, the gradient-based visualization of the trained network is performed to highlight important regions contributing to image similarity. Experiments conducted on the Market-1501 dataset verify that our model can not only enable the network to identify key features of an individual across different images, but also provide visual interpretation for the pedestrian classification results to improve the reliability of person Re-ID and foster trust from users regarding its decisions.

A series of new Zn-porphyrin dyes (LYD-1 similar to 5) with various substituted and anchoring groups have been designed, fabricated and fully characterized by nuclear magnetic resonance (NMR), ultraviolet visible (UV-vis), and High Resolution Mass spectroscopy (HRMS). The solar cells sensitized by new dyes have been systematically investigated by photoelectrochemical experiments of current density-voltage (J - V), incident monochromatic photon-to-electron conversion efficiency (IPCE) and electrochemical impedance spectra (EIS). The results show that the performance of DSC devices is deeply affected by the substituted and anchoring groups of dyes. The best solar cell sensitized by LYD-5 with alkoxyl donors and salicylic acid exhibits excellent photovoltaic conversion efficiency of 8.5% compared with 7.7% of YD2-o-C8-based solar cell under the same conditions. Their structural features and electron distribution of HOMOs and LUMOs are calculated computationally using a DFT method at the level of B3LYP/6-31G(d)/LANL2DZ.

Novel artificial photosynthesis systems are devised as cells of dye/Pd/NR-MOx (M = Ti, Zn)||Co-Pi/W:BiVO4 that convert efficiently CO2 to alcohols. The photocathodes are amino-functionalized, palladium deposited and in-situ sensitized nano-TiO2 or ZnO/FTO electrodes that are characterized by XPS, TEM, XRD, UV-vis spectra and determined by electrochemical techniques. The cell of dye/Pd/S-TiO2||Co-Pi/W:BiVO4 uniquely generates ethanol under irradiation of 200 mW/cm2, reaching 0.56 % light quantum efficiency (QE) at 0.56 V and 0.13 % QE without external electron supply. The cell of dye/Pd/N-ZnO||Co-Pi/W:BiVO4 solely produces methanol in a rate of 42.8 μM h-1 cm-2 at 0.56 V of a Si-solar cell, that is far less than the electrochemical voltage of water splitting (1.23 V). Its QE reaches to 0.38 % that is equal to plants. The isotopic labelling experiments confirm the carbon source and oxygen releasing. The selectivity for alcohols of multi-functionalized semiconductors is discussed.

New dithiafulvenyl-pi-phenothiazine dyes have been devised and prepared for dye-sensitized solar cells. Various pi-spacers have been successfully introduced into the skeleton of dithiafulvenyl and phenothiazine unit to generate novel D-pi-D-A dyes (DPP-1 similar to 4). All dyes have been characterized with NMR, HRMS, UV-vis and fluorescence spectra, and taken into cyclic voltammetry measurements. The devices of new dyes have been determined by photoelectrochemical experiments (IV, IPCE and EIS), in which, solar cell of DPP-4 with biphenyl ring pi-spacer enhances obviously its photoelectric conversion efficiency to 7.66% reaching 94% of N719-based standard cell and displays good long-term stability with quasi-solid-state electrolyte. Density functional theory (DFT) calculations of new dyes provide further insight into the molecular geometries and the impacts of the torsion angles on their photovoltaic performance. Large dihedral angles in DPP dyes induce good charge separation for efficient unidirectional flow of electron from donor to acceptor. (C) 2016 Elsevier B.V. All rights reserved.

Recently, there has been increasing demand in developing low-cost, effective structure health monitoring system to be embedded into 3D-woven composite wind turbine blades to determine structural integrity and presence of defects. With measuring the strain and temperature inside composites at both in-situ blade resin curing and in-service stages, we are developing a novel scheme to embed a resistive-strain-based thin-metal-film sensory into the blade spar-cap that is made of composite laminates to determine structural integrity and presence of defects. Thus, with fiberglass, epoxy, and a thinmetal- film sensing element, a three-part, low-cost, smart composite laminate is developed. Embedded strain sensory inside composite laminate prototype survived after laminate curing process. The internal strain reading from embedded strain sensor under three-point-bending test standard is comparable. It proves that our proposed method will provide another SHM alternative to reduce sensing costs during the renewable green energy generation.

A high-voltage (HV) transmitter integrated circuit for ultrasound medical imaging applications is implemented using 0.18-μm bipolar/CMOS/DMOS technology. The proposed HV transmitter achieves high integration by only employing standard CMOS transistors in a stacked configuration with dynamic gate biasing circuit while successfully driving the capacitive micromachined ultrasound transducer device immersed in an oil environment without breakdown reliability issues. The HV transmitter including the output driver and the voltage level shifters generates over 10-Vp-p pulses at 1.25-MHz frequency and occupies only 0.022 mm2 of core die area.

A transmitter front-end IC for ultrasound medical imaging applications is implemented using 30-V [email protected] Bipolar/CMOS/DMOS process. The proposed ultrasound transmitter front-end, consisting of voltage level-shifters and an output driver, achieves high integration and maintains good reliability while generating a 33ns output pulse signal in order to drive a large capacitive load, which represents the equivalent capacitance of the capacitive micromachined ultrasound transducer device. The proposed high-frequency transmitter achieves a delay of less than 20ns at 30 Vp-p output voltage while successfully driving a capacitive load of 44 pF and consumes an area of 0.34mm^2.

We present the design of Analog front end (AFE) low noise pre-amplifier implemented in a high voltage 0.18-µm CMOS technology for a three dimensional ultrasound bio microscope (3D UBM) application. The fabricated chip has 4X16 pre-amplifiers implemented to interface a 2-D array of high frequency capacitive micro-machined ultrasound transducers (CMUT). Core AFE cell consists of a high-voltage pulser in the transmit path, and a low-noise transimpedance amplifier in the receive path. Proposed system offers a high image resolution by the use of high frequency CMUTs with associated high performance imaging electronics integrated together. Performance requirements and the design methods of the high bandwidth transimpedance amplifier are described in the paper. A single cell of transimpedance (TIA) amplifier and the bias circuit occupies a silicon area of 250X380 µm2 and the full chip occupies a total silicon area of 10x6.8 mm²., {"references":["Woonggyu Jung, Jeehyun Kim,. Mansik Jeon, Eric J. Chaney, Charles N. Stewart, and Stephen A. Boppart, \"Handheld Optical Coherence Tomography Scanner for Primary care Diagnostics,\" IEEE Transaction on Biomedical engineering, Vol.58, No.3, March2011.","Ira 0. Wygant, Xuefeng Zhuang, David T. Yeh, Omar Oralkan, A. Sanli Ergun, Mustafa Karaman, Butrus T. Khuri-Yakub,\" Integration of 2D CMUT Arrays with Front-End Electronics for Volumetric Ultrasound Imaging,\" IEEE Transactions on Ultrasonics,Ferroelcteronics, and Frequency control, Vol. 55. No.2, February 2008.","X. Cheng, D. F. Lemmerhirt, 0. D. Kripfgans, M. Zhang, C. Yang, C.A. Rich, and J. B. Fowlkers,\"CMUT-IN_CMOS Ultrasonic Transducer Arrays with on-chip electronics,\" IEEE Solid-state Sensors, Actuators and Microsystems conference, 2009, pages 1222-1225.","Xiwei Huang, Hyouk-Kyu Cha, Dongning Zhao, Bin Guo, Minkyu Je, Hao Yu \"Transimpedance Amplifier for Integrated 3D Ultrasound Biomicroscope Applications\" World Academy of Science Engineering and technology 69 2012.","Dongning Zhao, Meng Tong Tan, Hyouk_Kyu Cha, Jinli Qu, Yan Mei, Hao Yu, Arindam Basu, Minkyu Je,\"High—Voltage pulser for ultrasound imaging,\" IEEE International symposium on Integrated Circuits(ISIC) 2011, Dec 2011.","Gokce Gurun, Jaime S. Zahorian, Alper Sisman, Mustafa Karman, Paul E. Hasler and F. Levent, \" IEEE Transaction on Biomedical circuits and systems, Vol. 6,No. 5, October2012.","Gokce Gurun, Paul E. Hasler and F. Levent,\" Front-End Receiver Electronics For High Frequency Monolithic CMUT-on-CMOS Imaging arrays,\" IEEE Transactions on Ultrasonics, Ferroelcteronics, and Frequency control, Vol. 58. No.8, August 2011.","Dipankar Nag, Jia Hao Cheong, and Je Minkyu,\" Low noise Low power Preamplifier for Integrated Ultrasound needle Intervention application,\" International Journal of Information and Electronics Engineering, Vol. 3, January 2013."]}

This paper presents the design and implementation of a fully integrated transimpedance amplifier (TIA) as the analog frontend receiver for Capacitive Micromachined Ultrasound Transducers (CMUTs) for ultrasound biomicroscope imaging application. The amplifier is designed to amplify the received signals from 17.5MHz to 52.5MHz with a center frequency of 35MHz. The TIA was fabricated in GF 0.18μm 1P6M 30V high voltage process. The measurement results show that the designed amplifier can reach a transimpedance gain of 61.08dBΩ and operating frequency from 17.5MHz to 100MHz with 1VP-P output voltage under 6V power supply.

We describe a smart sensor for EEG acquisition comprising a programmable gain low-noise amplifier followed by an integrated feature extraction and classification circuits. The feature extraction block comprises a bank of four band-pass filters followed by a wide dynamic range peak detector. The output of the peak detector is fed into a spiking neural network implementing the extreme learning machine (ELM) for classification. The advantage of ELM is that it has been shown to attain comparable performance to support vector machine (SVM) but with fewer computational nodes. We describe simulation results of each block designed in 0.35 um CMOS and demonstrate system level performance by using this to detect seizure onset in epileptic patients. The system can be reconfigured for other applications like speech classification. Accepted version

This paper presents the design and implementation of a fully integrated high-voltage (HV) front-end transducer for MEMS ultrasonic applications. Each of the front-end transducers in the array includes a HV transmitting driver and a 50MHz capacitive micro-machined ultrasound transducer (CMUT). The design adapts low voltage integrated circuit (IC) design techniques to create integrated high voltage interfaces. Effective methods to implement high voltage interfaces to integrated CMOS/DMOS circuits are presented. Using a 0.18μm CMOS/DMOS process, simulation results of this interface IC show that can generated a 30V high voltage pulse from a 1.8V input triggering signal to drive a 50MHz MEMS ultrasound transducer with the largest 44pF capacitive load.

We present a prototype lateral-PNP (LPNP) BJT interface IC for an SOI capacitive accelerometer with a measured resolution of 6.3mug/VHz and an output noise floor of -118dBVradicHz for a gain of 204mV/g at extremely low frequencies. The resolution is improved by further reducing the low-frequency Vf noise and offset of the LPNP input interface using a chopper stabilization technique. The interface is designed and fabricated in a 3V 0.6mum CMOS process and interfaced with the accelerometer device with wire-bonds, where the accelerometer is fabricated on an SOI substrate using a simple process that yields high sensitivity in a small die size. The power consumption of the IC is 3.75mW with external clocking.

With the development of Internet and databases application techniques, the demand that lots of databases in the Internet are permitted to remote query and access for authorized users becomes common, and the problem that how to protect the copyright of relational databases arises. This paper simply introduces the knowledge of cloud model firstly, includes cloud generators and similar cloud. And then combined with the property of the cloud, a method of protecting relational databases copyright with cloud watermark is proposed according to the idea of digital watermark and the property of relational databases. Meanwhile, the corresponding watermark algorithms such as cloud watermark embedding algorithm and detection algorithm are proposed. Then, some experiments are run and the results are analyzed to validate the correctness and feasibility of the watermark scheme. In the end, the foreground of watermarking relational database and its research direction are prospected.

With the development of the Internet and wide applications of databases, the databases providers are worrying about their ownership of the databases. Combined with the digital watermark technique and the characteristic of relational databases, a scheme of watermarking relational databases based on content Characteristic is proposed in this paper. The corresponding algorithms of watermark embedding and detection are proposed and analyzed by probability. Then some watermark embedding and detection experiments are conducted, and the results are analyzed and discussed. In additional, some attacked experiments are conducted to prove the feasibility and the robustness of the proposed algorithms.

We have self-assembled regimented arrays of vertical ∼100 nm diameter Ge Esaki tunnel diodes using nanosphere lithography. Measurements of the current-voltage characteristics of individual nanodiodes using conductive atomic force microscopy at room temperature reveal pronounced negative differential resistance under forward bias, with a peak to valley ratio of 2–4. These diode arrays could constitute a neuromorphic circuit architecture exhibiting collective computational activity.