Search

A huge number of display advertising campaigns are launched every day by advertisers in order to promote their products or services. The main objective of each advertiser is to display ads to specific groups of users, i.e. users who meet specific criteria or their interests are related to the promoted products or services. Audience expansion, also known as audience look-alike targeting, is one of the major display advertising techniques that helps advertisers to discover audiences with similar attributes to a target audience who is interested in advertisers' products or services. In this paper, we present different audience expansion schemes able to identify users with similar browsing interests to those of the seed users provided by the advertiser. The proposed audience expansion schemes are based on different unsupervised representation models that are able to capture the interests of the users according to their browsing history. We have conducted an extensive empirical study on a real data collected from an advertising platform to analyse the effectiveness of the proposed schemes to expand the audiences of five different advertisers.

Synchronization in the time-frequency domain is a key step when measurement on physical layer is performed in the fifth-generation (5G) new radio (NR) system. We measure received burst on physical layer to obtain synchronization information by improved correlation algorithms. Simulation results show that parameters can be acquired precisely from the view of measurement.

Predicting drug-drug interactions (DDIs) is crucial for patient safety and public health. The existing DDI prediction methods mainly fall into three categories: knowledge-based, similarity-based and network-based. Most recently, studies have demonstrated that integrating heterogeneous drug features is significantly important for developing high-accuracy prediction models, but it also brings many new challenges, i.e. heterogeneous properties, non-linear relations and incomplete data. In this paper, we propose a multi-modal deep auto-encoders based drug representation learning method for the DDI prediction, abbreviated as DDI-MDAE. The proposed method learns unified representations of drugs simultaneously from multiple drug feature networks using multi-modal deep auto-encoders. Then we adopt several operators on the learned drug embeddings to represent drug-drug pairs, and utilize the random forest to train models for the DDI prediction. Experimental results show that DDI-MDAE effectively learns the representations of drugs by fusing diverse information, and outperforms the other state-of-the-art benchmark methods. More importantly, DDI-MDAE works even for drugs without any known interaction.

In the cascaded inverter, the second-harmonic current (SHC) in the input of inverter penetrates into the front-end dc-dc converter, leading to a poor performance. In this paper, the pre-regulator+LLC converter, which shows superior performance in the applications with wide input voltage range, is employed as the front-end stage. Firstly, the boost converter is taken as the pre-regulator and the model of boost+LLC converter is proposed. Based on that, three approaches for SHC reduction are proposed from a point of output impedance, which are: 1) Reducing the filter capacitor of boost converter; 2) Introducing a notch filter in cascade with voltage regulator, 3) Adding a virtual impedance in series with the output rectifier of LLC resonant converter. Finally, a 6-kVA cascaded inverter is tested and experimental results are presented to verify the validity of the proposed methods.

In this paper, we propose an improved method to reconstruct object shape for shape sensors based on fiber bragg grating (FBG), where FBG array is embedded in silicone gel as protection shell and glued on rubber tape surface to distinguish bending direction. The calibration experiments of FBG are presented to ensure relative precise conversion between FBG wavelength offsets and bending curvature. In the improved method, spline interpolation technique instead of traditional linear technique is employed to characterize relationship between bending curvature and arc length, with Frenet-serret equation used in the procedure of object shape reconstruction. The experiment results show that spline interpolation technique induces lower error than linear interpolation technique, which verifies that an improvement in object shape reconstruction can be achieved by employing spline interpolation technique.

Underwater water optical communication (UWOC) requires high-power laser transmitter with good beam quality for high-rate data modulation. Since master oscillator power amplifier (MOPA) has the characteristic of preserving the property of seed laser (e.g. bandwidth and beam quality), it is expected to resolve the problem. In this paper, we theoretically simulate the large-signal modeling of the amplified laser light and the frequency response of a MOPA system and experimentally amplified 3-mW 1-um light to 145mW. The seed light came from 1064-nm TO-Packaged laser diode and was coupled into HI1060. To satisfy the requirements of the transmitters for UWOC, we used second harmonic generation and got 0.06mW 532-nm laser light. Finally, according to the theoretical analysis of a MOPA system and the electrical response of a laser driver, we chose Manchester Encoding to modulate the amplified 1064-nm laser light and achieved 100-Mbps transmission. The system was error free for 1◊104 bits.

Most previous studies investigated the impact of paternalistic leadership and challenge stressor on employee's in-role performance in the individual level. In this study, we focused on the impact in group level. Data were collected from 208 employees within 35 research and design work teams of real estate company and architecture design institute in 7 cities of mainland China. Team paternalistic leadership and team challenge stressor were rated by employees whereas in-role performance was assessed by their immediate supervisors. We found that team paternalistic leadership had a cross-level positive effect on employee's in-role performance which was mediated by team challenge stressor. Specifically, in terms of three dimensions of team paternalistic leadership, only team benevolence leadership and team moral leadership showed indirect effects on employee's in-role performance mediated by team challenge stressor.

In this paper, information leakage from electromagnetic radiation of differential transmission lines is analyzed. The model of differential transmission lines is built. And common mode noise caused by unbalance structure of the differential transmission lines and delay between differential drivers is simulated. These common mode noises have a corresponding relationship with the information transmitted over the differential lines. It can cause information leakage. Finally, the experiment presents the possibility of information leakage from differential transmission lines.

Total Knee Replacement (TKR) is an increasingly common surgery worldwide. A significant contributor to TKR success is post-surgical rehabilitation. In this work, we present Fun-Knee™, a novel sensor-equipped knee support complemented by mobile device-supported games, specifically designed for “gamified” TKR rehabilitation. Two inclinometers are used to measure the knee angle, which is used as the main input to the developed game. Human-Centered Design theory is applied throughout the game design to ensure a customized, dynamic gaming experience to maximize the pain distraction effect and to increase the exercise compliance and improve the rehabilitation outcome. Preliminary survey results collected from practicing physiotherapists show promising outcomes of the developed prototype, in terms of hardware and software characteristics, usability, clinical utility and overall effectiveness.

Electromagnetic radiation of computer contains video information which causes the information leakage. Electromagnetic radiation of PC contains video information, and the image can be reconstructed. In this paper, the way of computer video information transmission of DVI video interface be analyzed. The model of time and frequency domain corresponding to serial transmission and parallel transmission of video information is built. According to the principle of video information interception and reconstruction, the radiation leakage characteristics when different video signal is transmitted through the serial and parallel way is researched in this paper. Through the analysis and measurement of the electromagnetic information leakage, then we can get a conclusion that it is easy to be intercepted when the gray image is transmitted in serial way, black and white image is transmitted in parallel way, because it shows the characteristic of “false parallel and real serial”.

Urea-Selective catalytic reduction(Urea-SCR) method is an effective way to reduce the NO x emissions for diesel engine, and the closed-loop state feedback control strategy is widely studied by many researchers. In order to estimate the un-measurement state, this paper uses extended Kalman filter(EKF) to design a state observer for Urea-SCR system. Firstly, a two-cell Urea-SCR model is given. Secondly the two-cell Urea-SCR model is dispersed by Backward Euler method at each sampling instant. Thirdly, tailpipe NO x and ammonia concentration are chosen as the output of the Urea-SCR system and then used by EKF to estimate state variables of the upstream and downstream catalyst. At last, the effectiveness of the observer is validated by simulations under FTP75 driving cycles in enDYNA.

Electromagnetic radiation of PC contains video information, and the image can be reconstructed. It is well known after the Van Eck's paper. The video cable used to connect the display with the graphics controller turned out to be the primary source of the leaking signal. For digital video interface, not all kinds of color images can be intercepted and reconstructed. Firstly we analyzed which color image can be reconstructed based the digital interface standards. Then we propose a method that changes the original PGB signals to safety color, while keeping the quality of the image by using the characteristics of the human visual and color mixing. By implementing this method on FPGA board in an ordinary computer system, we confirmed the effectiveness of our method as a countermeasure for the electromagnetic video information leakage.

The environment of power system is complex, and the field to line coupling and coupling between lines are important factors lead to electromagnetic disturbance, so the electromagnetic compatibility problem in the system is serious. Electromagnetic compatibility on-site testing can distinguish disturbance sources and diagnose the failure of equipment. When testing the disturbance of secondary equipment in on-site environment, received signal contains not only the disturbance of secondary equipment, but also ambient noise, which influence the accuracy of testing. In this paper, spectral difference method based on threshold calculation is used. With wavelet theory and correlation analysis, ambient noise is eliminated effectively, and the accuracy of on-site testing is improved. Through the actual test, the validity of the method is verified.

An increasing number of safety features are being added to various engineering designs and processes of today and they have become the most important consideration in every design. Construction is a high-risk industry, hence the need for such systems is more pronounced. The cost of repairing damaged parts is significant due to the expensive price of construction machinery and related equipment. Based on the necessity for safety of the machinery and personnel on site, a suitable sensor system is designed and built to detect possible threats of collision with structures or people. Due to harsh operational environments, these sensors must be immune to a variety of interference and noise and must work in different weather conditions. A sensor system based on ultrasonic range-finding and infrared motion detection has been proposed which has undergone different tests to check feasibility of the design. The prototype worked as expected with some limitations which could be improved with better design and hence. Considering the low cost of the proposed system and its great economic and social benefits, the system would be a practically valuable option to be added onto existing machinery.

Aiming at the relationships of the high altitude meteorological parameters, such as longitude, latitude, atmospheric pressure, temperature and humidity, the methods of least squares fitting test data and error analysis by MATLAB are proposed in this paper. Temperature, humidity and air pressure change curve equations are established at the different longitude, latitude and altitude, and the influence of temperature and pressure on humidity detection accuracy is analyzed. The results show that the temperature fitting residual error is in −3 (°C) ∼ + 4 (°C), humidity fitting residual error is in −7% ∼ + 5%, the detection error precision is improved effectively, which can provide reliable compensation theory basis for signal detection.

Based on derivation of fluid inertia velocity distribution, the paper proposes an analytical model for pressure distribution of squeeze film damper (SFD) in long bearing assumption under impact loading. Compared with numerical simulation and test results, the model is reasonable. The model analysis results reflect the reversion behavior about SFD radial force. Therefore, a design rule of synchro-response for SFD is recommended, and then an expression of SFD parameters is given which is in accordance with the rule. These results are useful for design and for real-time active control of SFD-rotor-system, and are helpful for further theoretical investigation on oil film dynamic behavior also.

This paper presents an accelerated life testing method to solve the problem of in-pipe robot reliability in quantitative analysis. By established inverse power law probability density function model based on Weibull distribution, the results of accelerated life testing can be used to extrapolate reliability features of in-pipe robot under the normal condition, especially for single or rare prototype. The experimental results are in agreement with theoretical analysis, this forecast the mean time between the failures of the in-pipe robot and avoid its lost caused by online failure.

The commercial computer fluid dynamics software package FLUENT was used to simulate the flow field of the new toilet system. The mesh model of the flow field was built with the GAMBIT. With the numerical simulation of the flow field on air intakes and flushing system, the pressure and velocity vector cloud diagram have been given. The information would be very useful in experimental investigation. The study has offered the significant theoretic reference to optimize the toilet system.

In this paper, two high-accuracy low-cost current balancing circuits are proposed for LED backlight system by using capacitors and mini transformers. Compared with traditional active and passive balancing approaches, the proposed schemes greatly reduce cost and power losses, and achieve highly accurate current matching among all LED strings in wide dimming ranges. The current balancing principles of the proposed circuits are analyzed, and high accuracy designs are verified by simulation and experimental results.

Clusters are hidden in subspaces of high dimensional data, i.e., only a subset of features is relevant for each cluster. Subspace clustering is challenging since the search for the relevant features of each cluster and the detection of the final clusters are circular dependent and should be solved simultaneously. In this paper, we point out that feature correlation and distance divergence are important to subspace clustering, but both have not been considered in previous works. Feature correlation groups correlated features independently thus helps to reduce the search space for the relevant features search problem. Distance divergence distinguishes distances on different dimensions and helps to find the final clusters accurately. We tackle the two problems with the aid of a small amount domain knowledge in the form of must-links and cannot-links. We then devise a semi-supervised subspace clustering algorithm CDCDD. CDCDD integrates our solutions of the feature correlation and distance divergence problems, and uses an adaptive dimension voting scheme, which is derived from a previous unsupervised subspace clustering algorithm FINDIT. Experimental results on both synthetic data sets and real data sets show that the proposed CDCDD algorithm outperforms FINDIT in terms of accuracy, and outperforms the other constraint based algorithm SCMINER in terms of both accuracy and efficiency.

Communications system with more antennas carried in vehicle, plane and ship is regarded as antenna and scatter body structure. Wire antenna and scatter body structure with faceted triangle and quadrangle surfaces are taken into consideration. An partition approach consisting of three steps is presented. The first step is to partition all surfaces of the scatter body by recursively bisecting each triangle and quadrangle to get the required sizes of all elements, neglecting the presence of antennas. The second step is to partition the elements connected with the wire antenna into smaller areas. The third step is to partition the narrow elements to improve their shape quality. Engineering applications show that this algorithm is stable, reliable, fast.

To improve the intelligentability and usability of internet equipments and server management, effective and precise discovery of internet topology is becoming a very important part of internet management and its arithmetic and relizational techniques are one important element in evaluating the performance of internet management system. Most internet equipments nowadays support SNMP protocol, and this thesis designs an algorithm on internet topology automatic discovery system. This arithmetic can discover all the routers and assigned-internet active pc, and analyses its detailed realization of the key techniques.

An 80-km-reach WDM PON with 10-Gb/s downstream and 2.5-Gb/s upstream re-modulated signals is demonstrated. The dispersion tolerant Inverse-RZ-Duobinary is employed as the downstream format while upstream ASK re-modulation is performed by reflective SOA.

Within the software industry software piracy is a great concern. In this article, this issue is addressed through a prevention technique called software watermarking. The watermarking schema can be used for protecting the software written in jME API. A program is marked by hiding the copyright in a scene graph. According to a secret initial order of scene graph traversal, a secret message is embedded through circular-left shift operation and a numbering system. Furthermore, the embedding and extraction technique is discussed in detail, and evaluated against a variety of attacks. The algorithm is evaluated against a variety of attacks. The method does not change the size of scene graph, and the embedded watermarks are more robust because disturbing the watermark would affect the semantic and the functionality of the underlying software.

A novel OFSK transmitter using phase modulator-embedded optical loop mirror and carrier suppression modulation is proposed and experimentally demonstrated. Only one laser source is needed and its structure is simple for potential integration.

A WDM-PON with simultaneous delivery of 10-Gb/s point-to-point downstream and upstream data as well as 10-Gb/s multicast data is proposed and demonstrated. The multicast overlay control is realized by using optical ON/OFF switches at OLT.

Multicast overlay techniques [1–2] in WDM-PONs can support delivery of the same data or video service to a designated subset of subscribers, which can also be flexibly reconfigured at the optical line terminal (OLT). In this paper, we propose a novel polarization-assisted scheme for flexible multicast overlay in a WDM-PON, which supports both point-to-point (P2P) downstream and upstream data delivery. With our proposed wavelength assignment and routing schemes, the system performance is greatly improved. No additional dedicated light source for the multicast data is needed. We have experimentally demonstrated 10-Gbit/s transmissions for the P2P downstream and upstream data, as well as the multicast data in a WDM-PON.

In this paper, the video information camouflage processing method based on the pulse amplitude modulation was studied aim at the electromagnetic characteristic of computer video information. Camouflage information modulation function was proposed, the matching vision relation between character information and background color was determined, and the image fusion processing algorithm of camouflage modulation information and useful information was confirmed. Finally, the correctness and effectiveness of theoretical analysis was validated through simulation and experiments.

In this paper, on the basis of the spot pulse width corresponded with character video information vertical stroke, the safe bandwidth of video information is confirmed, filtering image quality appraisal model which takes difference threshold limit value as distinction parameter is built by Weber – Fechner law, the corresponding relation of filtering stop-band cut-off frequency and character is determined, the FIR digital filter is designed. In addition, aim at the periodic variation characteristic of the insertion loss which the digital filter inherent, a combinatorial filtering method combine digital filter and analog filter is presented. At last, the correctness and effectiveness of theoretical analysis are validated through simulations and experiments.

According to the character of the typical serial information leakage and the feature in the information leakage problem that multiple points (zone) converge and couple toward one point (zone), several key concepts, such as the emission zone, transmission zone and receiving zone, are defined in this paper. Based on the idea of electromagnetic topology, a completely new construction method of system topological graph is proposed. Then, this method is applied to analyze a simplified serial information leakage system, the generalized transmission coupling model is also established. What’s more, the modeling process of field-to-field coupling is discussed emphatically. Therefore a solid foundation is laid for the electromagnetic topology analysis of serial information leakage.

A method for tracking on motion of small target is presented that is based on edge detection of image sequences. This approach detects the edge of small target of image sequences first and decides the position of centroid. According to the position tracking on motion of small target can be completed by least-square fit. The trajectory of tracking and collecting of small target is compared. The result implies that the tracking on small target comparing to experiment has little error and can achieve satisfactory effect.

The TBD algorithm based on dynamic programming is an effective method for detecting dim target. But the DP process during the multi-frame integration causes the expansion of targets and strong noise which may produce a large amount of false targets. To reduce them, a new discrimination method which takes advantage of the feature of trajectory is proposed in this paper. The false alarm caused by noise reveals strong randomicity on its trajectory which is different from that caused by true target, and the new method based on the statistics of different level direction histogram can be used to reduce false targets caused by noise. Moreover a new concatenated discrimination scheme of SDH+LEV is proposed. The simulation indicates that the discrimination method of SDH is good for reducing false alarms caused by noise and preserving real targets at the same time. The concatenated discrimination scheme can delete false alarms caused by noise effectively.

High switching frequency, high resolution digital pulse-width modulator (DPWM) is one of major challenges in the implementation of digital-controlled power converters, especially in the voltage regulator (VR) application. This paper proposes three different digital duty cycle modulation schemes to improve the resolution. Almost 10 times improvement on resolution can be achieved in the voltage regulator (VR) apparition. Design difficulty of the DPWM can be greatly reduced by proposed modulation schemes. Experimental results verify those modulation concepts.