Design and Simulation of High-Temperature Micro-Hotplate for Synthesis of Graphene Using uCVD Method.

The uCVD (microchemical vapor deposition) graphene growth system is an improved CVD system that is suitable for scientific research and experimental needs, and it is characterized by its rapid, convenient, compact, and low-cost features. The micro-hotplate based on an SOI wafer is the core component of this system. To meet the requirements of the uCVD system for the micro-hotplate, we propose a suspended multi-cantilever heating platform composed of a heating chip, cantilevers, and bracket. In this article, using heat transfer theory and thermoelectric simulation, we demonstrate that the silicon resistivity, current input cross-sectional size, and the convective heat transfer coefficient have a huge impact on the performance of the micro-heating platform. Therefore, in the proposed solution, we adopt a selective doping process to achieve a differentiated configuration of silicon resistivity in the cantilevers and heating chip, ensuring that the heating chip meets the requirements for graphene synthesis while allowing the cantilevers to withstand high currents without damage. Additionally, by adding brackets, the surfaces of the micro-hotplate have the same convective heat transfer environment, reducing the surface temperature difference, and improving the cooling rate. The simulation results indicate that the temperature on the micro-hotplate surface can reach 1050.8 °C, and the maximum temperature difference at different points on the surface is less than 2 °C, which effectively meets the requirements for the CVD growth of graphene using Cu as the catalyst.

Automatic Sparse Connectivity Learning for Neural Networks.

Since sparse neural networks usually contain many zero weights, these unnecessary network connections can potentially be eliminated without degrading network performance. Therefore, well-designed sparse neural networks have the potential to significantly reduce the number of floating-point operations (FLOPs) and computational resources. In this work, we propose a new automatic pruning method-sparse connectivity learning (SCL). Specifically, a weight is reparameterized as an elementwise multiplication of a trainable weight variable and a binary mask. Thus, network connectivity is fully described by the binary mask, which is modulated by a unit step function. We theoretically prove the fundamental principle of using a straight-through estimator (STE) for network pruning. This principle is that the proxy gradients of STE should be positive, ensuring that mask variables converge at their minima. After finding Leaky ReLU, Softplus, and identity STEs can satisfy this principle, we propose to adopt identity STE in SCL for discrete mask relaxation. We find that mask gradients of different features are very unbalanced; hence, we propose to normalize mask gradients of each feature to optimize mask variable training. In order to automatically train sparse masks, we include the total number of network connections as a regularization term in our objective function. As SCL does not require pruning criteria or hyperparameters defined by designers for network layers, the network is explored in a larger hypothesis space to achieve optimized sparse connectivity for the best performance. SCL overcomes the limitations of existing automatic pruning methods. Experimental results demonstrate that SCL can automatically learn and select important network connections for various baseline network structures. Deep learning models trained by SCL outperform the state-of-the-art human-designed and automatic pruning methods in sparsity, accuracy, and FLOPs reduction.

TR Self-Adaptive Cancellation Based Pipeline Leakage Localization Method Using Piezoceramic Transducers.

In this paper, we propose a novel time reversal-based localization method for pipeline leakage. In the proposed method, a so-called TR self-adaptive cancellation is developed to improve the leak localization resolution. First of all, the proposed approach time reverses and back-propagates the captured signals. Secondly, the time reversed signals with the various coefficients are superposed. Due to the synchronous temporal and spatial focusing characteristic of time reversal, those time reversed signals will cancel each other out. Finally, the leakage location is distinguished by observing the energy distribution of the superposed signal. In this investigation, the proposed method was employed to monitor a 58 m PVC pipeline. Three manually controllable valves were utilized to simulate the leakages. Six piezoceramic sensors equipped on the pipeline, recorded the NWP signals generated by the three valves. The experimental results show that the leak positions can accurately revealed by using the proposed approach. Furthermore, the resolution of the proposed approach can be ten times that of the conventional TR localization method.

Neural Embedding Singular Value Decomposition for Collaborative Filtering.

Singular value decomposition (SVD) is one of the most effective algorithms in recommender systems (RSs). Due to the iterative nature of SVD algorithms, one big challenge is initialization that has a major impact on the convergence and performance of RSs. Unfortunately, existing SVD algorithms in the literature typically initialize the user and item features in a random manner; thus, data information is not fully utilized. This work addresses the challenge of developing an efficient initialization method for SVD algorithms. We propose a general neural embedding initialization framework, where a low-complexity probabilistic autoencoder neural network initializes the features of user and item. This framework supports explicit and implicit feedback data sets. The design details of our proposed framework are elaborated and discussed. Experimental results show that RSs based on our proposed initialization framework outperform the state-of-the-art methods in rating prediction. Moreover, regarding item ranking, our proposed framework shows an improvement of at least 2.20% ~5.74% than existing SVD algorithms and other matrix factorization methods in the literature.