Search

Aiming at the low recognition accuracy of the traditional machine learning algorithm which is susceptible to digital writing quality, inter-digital adhesion, random noise background and other factors in the process of adhesion handwritten digit recognition, an new method based on improved fast regional convolutional neural network(Faster RCNN) of adhesion handwritten digit recognition is proposed. Firstly, the NIST19 dataset is used as the basic dataset, and a mixed dataset is created by setting different hand-to-hand ratios with different degrees of overlap, and then randomly add salt and pepper noise and Gaussian noise in the experimental images. Secondly, aiming at the problem of a large number of overlapping objects in the handwritten digital images, a model based on improved Faster RCNN network is built and trained with the above data sets. Finally, the average accuracy of the model is evaluated. The experimental results show that the average detection accuracy of the proposed model is good. Compared with the original Faster RCNN and YOLO models, the improved model not only reduces the scale of parameters, but also ensures high recognition accuracy, and realizes the accurate and efficient recognition of handwritten adhesive digits.

Mode mixing is a limitation of the empirical mode decomposition (EMD) method appropriate for physiological signal analysis. In 2008, boundary condition map presented by Rilling and Flandrin provided the efficiency of separating the two components of a two-tone signal as a function of their amplitude and frequency ratios. Until 2019, their findings were still applied. However, their maps only give an uncertainty-like efficiency of mode mixing separation for two-tone signals. In this paper, we propose a criterion for mode mixing separation in EMD, which provides a binary judgment on mode mixing separation instead of the above-mentioned efficiency. By comparing the slopes of the two components, we found that the phenomenon of mode mixing occurs as the extrema of the high-tone component are suppressed by the low-tone component. Under this condition, the criterion shows the relation among their amplitude ratio, frequency ratio, and relative phase between the two components. Given with the values of the three parameters, one can affirm whether the two components are mixed according to the criterion. Accordingly, we derive a black/white three-dimensional (3D) map that plots the binary result of mode mixing in black or white as a function of the three parameters. Our map agrees with Rilling's map and the results obtained from our gait analysis. Among the 23 sets of center-of-mass trajectory signals, six sets encountered the mode mixing problem and their coordinates of the three parameters were found in the black region of the map, while the other 17 sets were in the white region.

A high performance 3D dipole antenna with metal thickness >100 μm for wide bandwidth and lateral radiation is realized on InFO package. 25 % wide fractional bandwidth, from 60 to 77 GHz, has been obtained. The beamforming capability of the antenna array system with 6 dBi gain is measured in a 40 nm CMOS RFIC co-designed system.

High Q-factor (quality factor) 3D solenoid inductor formed by RDL (redistribution layer) and copper via in the molding compound on InFO package was fabricated. The effect of the turns and cross sectional area on Q-factor is discussed. The 3D solenoid inductor with 3 turns achieves the Q-factor of 59 at 3GHz and 0.68× lower of resistance and 1.6× higher of Q-factor than 2D spiral inductor at the same inductance. The coupling between two closely placed inductors is also investigated and it is found that the coupling in orthogonal placement is 12 dB lower than that of in parallel placement. The sample was fabricated and measured. The measurement results agree with the simulation, indicating the feasibility of the 3D inductor on InFO package.

Deploying multiple sensors in the field of AIoTs is inevitable for improving human life to be more intelligent. Unfortunately, most of prior studies pay their attentions merely on "how a single sensor can operate properly in its application scenario" but not on "how multiple sensors can operate coordinately in the same scenario". The current study adopts the concept of Game Theory to propose a sensor interactive model and its respective algorithm design so that the sensors mounted in AIoTs can achieve communicational consensus.

Any MapReduce-based big data analytics clusters systems like Hadoop, Spark face the substantial challenge of the Long Tail problem: a small subset of straggling tasks significantly impede parallel jobs completion. Specially, in heterogeneous environments some tasks become stragglers because of poor performance of some computing nodes, data skew, etc. Therefore, stragglers are well recognized as a major bottleneck in big data processing and hence the early detection and accurate identification of stragglers can have important impacts on big data processing. In past years speculative execution strategies that have been proposed to meet their challenges such as misjudgment or delaying of straggling tasks, improper selection of backup nodes, etc., which result in inaccurate and inefficient performance of the speculative execution. Encouraged by recent successes in applying reinforcement learning (RL) techniques to solve complex online control problems, we study RL can be used for automatic choose right strategy to address stragglers without human-intervention. In this paper, we present Cheetah: a novel dynamic optimal approach for speculative strategy which identifies stragglers by reinforcement learning and automatic chooses the best strategy to launch speculative tasks on heterogeneous cluster. We implement Cheetah with popular reinforcement learning frameworks, and deploy it on a testbed of heterogeneous Spark cluster with 4 nodes. According to the experimental results, Compared to existing approaches, Cheetah reduces the job completion time in different type applications while achieving superior performance. For example, it demonstrates up to almost 16% reduction average job completion time and improve 25% on accuracy over existing solutions on the heterogeneous Spark cluster.

The 32-bit security code provides guarantee for ensuring quality of cigarette and maintaining market circulation order. The 32-bit security code is printed on the surface of outer packing of whole cigarette, and it may be disturbed by printing method, printing position and package background, which will make the 32-bit security code be identified hardly. The improved adaptive TV(total variation) denoising model and restoration algorithm are provided in this paper, which can realize the restoration of 32-bit security code in strong interference environment. The performance of this algorithm is analyzed and evaluated by experiments, and it can prove that the indexes of this algorithm are better than other existing ones.

An electromagnetic spring-based variable stiffness actuator is a new concept with the potential to change stiffness faster than mechanical springs can; however, its nonlinear elastic property is a challenge in actuator design. In this paper, the torque response of a custom-made electromagnetic spring was studied using a ramping force test and electromagnetic simulation. A two-zone linear region, from 0° to 2° and 2° to 6.5°, was observed and explained through magnetic flux simulation to provide insight into the fundamentals of the electromagnetic spring. An unusual impedance response was also noted from this regional linearity, appeared on a step gain in Bode plot of end-point impedance in a dynamic test.

In this paper, a compact and miniaturized coupler is designed for high-power applications with coupling of about - 53dB in the frequency band between 10 ∼ 100 MHz. The coupler is composed of a compact coaxial line and a series of equalizers. Such design exhibits high power handling capacity (up to 2 kW), flat coupling of −53 dB, and high directivity (more than 30 dB). And the coupling value can also be conveniently tuned by the equalizers.

To simultaneously achieve high mobility and good reliability of thin-film transistors (TFTs), we controlled oxygen vacancies via nitrogen incorporation in InGaZnO TFTs from vacuum- or solution-based methods. For display applications, IGZO and crystalline silicon TFTs were integrated together to build hybrid CMOS. Also, additive patterning method is developed for solution-processed oxide semiconductors, based on which wafer-scale TFT arrays and NMOS have been demonstrated.

The playback of flights' surveillance information plays an import role when flights encounter special circumstances, such as accidents. It can help analyze when, where and why the special circumstances happened intuitively. Different from traditional air traffic control automation system, we propose a playback system based on double-buffers method. A playing buffer and a fetching buffer is created, one is for storing the surveillance information that will be play back and the other one is used for storing the surveillance information fetched from the database. Thus, data fetching and data playback can be done concurrently, to play back the flights' surveillance information continuously. What's more a reliable UDP is proposed to improve the reliability of the data communication of the playback system.

Annealing effect on amorphous indium-zinc-tungsten-oxide (a-IZWO) thin-film transistors (TFTs) was investigated. As the annealing temperature increases from 150 □ to 400 □ for one hour, the IZWO films remain the amorphous state being conducive to the large area uniformity. The field effect mobility of the devices increases as a function of annealing temperature and reaches to 16.2 cm2/V·s at 300 °C, along with the on/off current ratio of and subthreshold swing value of 0.20 V/decade.