Neural architecture search (NAS) has significant progress in improving the accuracy of image classification. Recently, some works attempt to extend NAS to image segmentation which shows preliminary feasibility. However, all of them focus on searching architecture for semantic segmentation in natural scenes. In this paper, we design three types of primitive operation set on search space to automatically find two cell architecture DownSC and UpSC for semantic image segmentation especially medical image segmentation. Inspired by the U-net architecture and its variants successfully applied to various medical image segmentation, we propose NAS-Unet which is stacked by the same number of DownSC and UpSC on a U-like backbone network. The architectures of DownSC and UpSC updated simultaneously by a differential architecture strategy during the search stage. We demonstrate the good segmentation results of the proposed method on Promise12, Chaos, and ultrasound nerve datasets, which collected by magnetic resonance imaging, computed tomography, and ultrasound, respectively. Without any pretraining, our architecture searched on PASCAL VOC2012, attains better performances and much fewer parameters (about 0.8M) than U-net and one of its variants when evaluated on the above three types of medical image datasets.

Deep learning (DL) is a hot topic in current pattern recognition and machine learning. DL has unprecedented potential to solve many complex machine learning problems and is clearly attractive in the framework of mobile devices. The availability of powerful pattern recognition tools creates tremendous opportunities for next-generation smart applications. A convolutional neural network (CNN) enables data-driven learning and extraction of highly representative, hierarchical image features from appropriate training data. However, for some data sets, the CNN classification method needs adjustments in its structure and parameters. Mobile computing has certain requirements for running time and network weight of the neural network. In this paper, we first design an image processing module for a mobile device based on the characteristics of a CNN. Then, we describe how to use the mobile to collect data, process the data, and construct the data set. Finally, considering the computing environment and data characteristics of mobile devices, we propose a lightweight network structure for optical character recognition (OCR) on specific data sets. The proposed method using a CNN has been validated by comparison with the results of existing methods, used for optical character recognition.

The recent advances in convolutional neural networks (CNNs) have used for image classification to achieve remarkable results. Different fields of image datasets will need different CNN architectures to achieve exceptional performance. However, designing a good CNN architecture is a computationally expensive task and requires expert knowledge. In this paper, we propose an effective framework to solve different image classification tasks using a convolutional neural architecture search (CNAS). The framework is inspired by current research on NAS, which automatically learns the best architecture for a specific training dataset, such as MNIST and CIFAR-10. Many search algorithms have been proposed for implementing NAS; however, insufficient attention has been paid to the selection of primitive operations (POs) in the search space. We propose a more efficient search space for learning the CNN architecture. Our search algorithm is based on Darts (a differential architecture search method), but it considers different numbers of intermediate nodes and replaces some unused POs by channel shuffle operation and squeeze-and-excitation operation. We achieve a better performance than Darts on both the CIFAR10/CIFA100 and Tiny-ImageNet datasets. We retain the none operation in deriving the architecture. The performance of the model has slightly decreased, but the number of architecture parameters has been reduced by approximately 40%. To balance the performance and the number of architecture parameters, the framework can learn a dense architecture for high-performance machines, such as servers, but a sparse architecture for resource-constrained devices, such as embedded systems or mobile devices.

Intelligent transportation systems (ITSs) are important applications of 5G mobile communication technology. Vehicular ad hoc networks (VANETs) are key part of intelligent transportation systems. Although 5G technology and infrastructure can significantly promote the development of VANETs, in some scenarios, for example, if 5G service is not available or the vehicle speed is too fast, device to device (D2D) will be an important means of communication. Therefore, combining 5G and D2D VANET communication is crucial to provide better connectivity. However, due to high-speed mobility, uneven distribution, and unpredictable vehicle behavior, the VANET network topology changes rapidly, and network connectivity is unstable. This phenomenon represent a considerable challenge to the routing protocol; at the same time, the continuous establishment of new routes needs to send a large number of routing requests (RREQ) and routing reply (RREP) packets, and it therefore consumes considerable energy. How to choose a more stable route is the key to implementing the routing protocol. In this paper, we take the most representative features of vehicles and roadways into consideration and based on the hypothesis of uneven vehicle distribution, we use a nonhomogeneous Poisson process to analyze the network connectivity and take the network connectivity as one of the routing decision metrics. Therefore, using fuzzy logic to make routing decisions under multiple selection criteria, we developed a new routing protocol named LENC (low-latency and energy-efficient routing based on network connectivity). Simulations have been performed comparing classical AODV and LENC. The results indicate that LENC represents a great improvement in routing stability and is superior to AODV in several aspects.

The demand for edge device models equipped with multilingual visual capabilities is rapidly increasing in complex IoT application scenarios. While many studies have endowed models with strong visual sensory and language analysis capabilities, these models are often large and require substantial amounts of data. Moreover, multilingual parallel corpora are extremely scarce. Although large parameter sizes can enhance a model’s visual-language processing capabilities, the high training and inference costs make them unsuitable for edge devices and result in suboptimal performance in multilingual contexts. To address these challenges, this article proposes an generative visual language model, that is, cross-lingual, lightweight, data-efficient, easy to train, and easy to infer. We map both English and non-English features into the same space and align them with a visually distilled model while leveraging the inherent similarity information of languages to increase the supervision coverage of the dataset. Through extensive experiments, we demonstrate that our model achieves state-of-the-art performance across three downstream tasks: 1) image captioning; 2) machine translation; and 3) visual question answering, surpassing existing methods.