Explore the vast range of titles available.

Polymer memristors with light weight and mechanical flexibility are preeminent candidates for low-power edge computing paradigms. However, the structural inhomogeneity of most polymers usually leads to random resistive switching characteristics, which lowers the production yield and reliability of nanoscale devices. In this contribution, we report that by adopting the two-dimensional conjugation strategy, a record high 90% production yield of polymer memristors has been achieved with miniaturization and low power potentials. By constructing coplanar macromolecules with 2D conjugated thiophene derivatives to enhance the π – π stacking and crystallinity of the thin film, homogeneous switching takes place across the entire polymer layer, with fast responses in 32 ns, D2D variation down to 3.16% ~ 8.29%, production yield approaching 90%, and scalability into 100 nm scale with tiny power consumption of ~ 10 −15  J/bit. The polymer memristor array is capable of acting as both the arithmetic-logic element and multiply-accumulate accelerator for neuromorphic computing tasks. Though polymer memristors are promising for low‐power flexible edge computing applications, realizing efficient nanometer‐scale arrays remains a challenge. Here, the authors report a record high 90% production yield in nm‐scale 2D conjugated polymer memristors with homogeneous resistive switching.

The intelligent monitoring of cutting tools used in the manufacturing industry is steadily becoming more convenient. To accurately predict the state of tools and tool breakages, this study proposes a tool wear prediction technique based on multi-sensor information fusion. First, the vibrational, current, and cutting force signals transmitted during the machining process were collected, and the features were extracted. Next, the Kalman filtering algorithm was used for feature fusion, and a predictive model for tool wear was constructed by combining the ResNet and long short-term memory (LSTM) models (called ResNet-LSTM). Experimental data for thin-walled parts obtained under various machining conditions were utilized to monitor the changes in tool conditions. A comparison between the ResNet and LSTM tool wear prediction models indicated that the proposed ResNet-LSTM model significantly improved the prediction accuracy compared to the individual LSTM and ResNet models. Moreover, ResNet-LSTM exhibited adaptive noise reduction capabilities at the front end of the network for signal feature extraction, thereby enhancing the signal feature extraction capability. The ResNet-LSTM model yielded an average prediction error of 0.0085 mm and a tool wear prediction accuracy of 98.25%. These results validate the feasibility of the tool wear prediction method proposed in this study.

Background subtraction is an essential task in computer vision, and is often used as a pre-processing step for many advanced tasks. In this work, we propose a novel multi-scale feature fusion attention mechanism network to tackle cross-scene background subtraction. The cross-fusion of feature maps at different stages of the encoder makes the features input into the decoder contain low-level and high-level information. The spatial–channel attention based on the weight matrix makes the model focus on processing information related to foreground extraction. We evaluate the proposed model on the CDnet-2014 dataset with two scene-independent evaluation strategies and obtain competitive F-Measure. In addition, to evaluate the generalization ability of the model, we perform a cross-dataset evaluation scheme on the LASIESTA and SBI2015 datasets. The overall F-Measure of the model is 0.89 and 0.93, respectively. Experimental results demonstrate that the model performs well compared to the current state-of-the-art methods.

The image outpainting based on deep learning shows good performance and has a wide range of applications in many fields. The previous image outpainting methods mostly used a single image as input. In this paper, we use the left and right images as input images, and expand the unknown region in the middle to generate an image that connects the left and right images to form a complete semantically smooth image. A cascaded refinement residual attention model for image outpainting is proposed. The Residual Channel-Spatial Attention (RCSA) module is designed to effectively learn image information about known regions. The Cascaded Dilated-conv (CDC) module is used to capture deep features, and more semantic information is obtained through dilated convolutions of different rates. The Refine on Features Aggregation (RFA) module connects the encoder and decoder to refine the result image for generating it clearer and smoother. Experimental results show that the proposed model is able to hallucinate more meaningful structures and vivid textures and achieve satisfactory results.

Convolutional neural networks (CNNs) have achieved significant success in the field of single image dehazing. However, most existing deep dehazing models are based on atmospheric scattering model, which have high accumulate errors. Thus, Cascaded Deep Residual Learning Network for Single Image Dehazing (CDRLN) with encoder-decoder structure is proposed, which can directly restore the clean image from hazy image. The proposed algorithm consists of a primary network which predicts a residual map based on the entire image, and a sub-network which restores the haze-free image based on the residual image and the original hazy image. The encoder part of CDRLN embeds a context feature extraction module to fuse information effectively. In addition, the two-stage cascaded strategy can avoid feature dilution and restore detailed information, which reduces the color distortion in the dehazing process and generates a more natural, more real and less artifacts dehazed image. Experimental results demonstrate that the CDRLN surpasses previous state-of-the-art single image dehazing methods by a large margin on the synthetic datasets as well as real-world hazy images, and the visual effect of dehazed image is better.

Background subtraction is a challenging and fundamental task in computer vision, which aims at segmenting moving objects from the background. Recently, the attention mechanism has become a hot topic in the neural network. The algorithms based on encoder-decoder and multi-scale type network perform impressive results in the domain of background subtraction. In this paper, we propose a multi-scale inputs and labels (MSIL) model which is based on the encoder-decoder type network and the channel attention. The multi-scale fusion encoding (MSFE) module aims to utilize multi-scale inputs effectively, which can fuse the high-level and low-level features details. The channel attention (CA) module is introduced to connect the encoder and decoder to model channel-wise attentions. The multi-label supervision decoding (MLSD) module helps to learn richer hierarchical features and achieves better performance by the new multi-label supervision. The proposed model is also evaluated on the CDnet-2014 dataset and the LASIESTA dataset, which demonstrate the effectiveness and superiority of the proposed model by an average F-Measure of 0.9851 and 0.9633, respectively. In addition, scene independent evaluation experiments on the CDnet-2014 dataset demonstrate the effectiveness of the model on unseen videos.

Image inpainting methods based on deep convolutional neural networks (DCNN), especially generative adversarial networks (GAN), have made tremendous progress, due to their forceful representation capabilities. These methods can generate visually reasonable contents and textures; however, the existing deep models based on a single receptive field type usually not only cause image artifacts and content mismatches but also ignore the correlation between the hole region and long-distance spatial locations in the image. To address the above problems, in this paper, we propose a new generative model based on GAN, which is composed of a two-stage encoder–decoder with a Multi-Scale Encoder Network (MSE-Net) and a new Contextual Attention Model based on the Absolute Value (CAM-AV). The former utilizes different-size convolution kernels to encode features, which improves the ability of abstract feature characterization. The latter uses a new search algorithm to enhance the matching of features in the network. Our network is a fully convolutional network that can complete holes of arbitrary size, number, and spatial location in the image. Experiments with regular and irregular inpainting on different datasets including CelebA and Places2 demonstrate that the proposed method achieves higher quality inpainting results with reasonable contents than the most existing state-of-the-art methods.

Since conventional CMOS technology has met its development bottleneck, an alternative technology, quantum-dot cellular automata (QCA), attracted researchers’ attention and was studied extensively. The manufacturing process of QCA, however, is immature for commercial production because of the high defect rate. Seeking for designs that display excellent performance shows significant potentials for practical realizations. In the paper we propose a 5 × 5 module, which not only can implement three-input majority gate but also can realize five-input majority gate by adding another two inputs. A comprehensive analysis is made in terms of area, number of cells, energy dissipation and fault tolerance against single-cell omission defects. In order to testify the superiority of the proposed designs, preexisting related designs are tested and compared. Weighing up above four kinds of factors and technical feasibility, proposed majority gates perform fairly well. Further, we take full adders and multi-bit adders as illustrations to display the practical application of proposed majority gates. The detailed comparisons with previous adders reveal that proposed 5 × 5 module behaves well in circuits, especially the high degree of fault tolerance and the relatively small area, complexity and QCA cost, thereby making it more suitable for practical realizations in large circuit designs.

Due to its ultrasmall size and extremely low power consumption, quantum-dot cellular automata (QCA) technology represents a promising alternative to semiconductor transistors at the nanoscale. Nevertheless, the design of QCA circuits is limited by their high defect rate during fabrication, making fault-tolerant QCA structures a popular research topic. The aim of this work is to design a new fault-tolerant QCA majority gate based on a 3 × 5 tile. The majority gate guarantees good fault tolerance under single cell and double cell missing defects compared with several previous structures. The functional results when adopting the polarization and kink energy of the proposed majority gate under such single cell and double cell deposition defects are fully investigated. Besides, a series of new fault-tolerant adders are implemented based on the fault-tolerant majority gates. To evaluate the performance of the proposed adders, a thorough comparison versus previous adders with respect to fault tolerance, cell number, area, and delay is carried out. The results indicate that the proposed design can reach a high level of fault tolerance, and perform rather well in terms of other properties. For simulation analysis, the QCADesigner tool is used to check the functionality of all circuits.

This paper presents a fast transient response low-dropout (LDO) regulator with all-NPN push–pull buffer in 0.6-μm bipolar process. In order to improve the transient response, an all-NPN push–pull buffer is proposed. Based on single Miller capacitance (SMC), the use of the all-NPN push–pull buffer overcomes the shortcomings of the equivalent series resistance (ESR) that requires strict output capacitor types. Besides, the proposed merging structure of bandgap reference and error amplifier not only improves the transient response, but also simplifies the circuit and reduces the output noise. Implemented and fabricated in a 0.6-μm bipolar process, the proposed LDO regulator occupies an active area of 1.6 mm 2 . The measured maximum load current is 200 mA, and the circuit can work at the load current of 300 mA. Moreover, the measured line regulation and load regulation are 0.8 mV/V and 0.09 mV/mA, respectively.