Explore the vast range of titles available.

Currently, with the rapid development of deep learning, deep neural networks (DNNs) have been widely applied in various computer vision tasks. However, in the pursuit of performance, advanced DNN models have become more complex, which has led to a large memory footprint and high computation demands. As a result, the models are difficult to apply in real time. To address these issues, model compression has become a focus of research. Furthermore, model compression techniques play an important role in deploying models on edge devices. This study analyzed various model compression methods to assist researchers in reducing device storage space, speeding up model inference, reducing model complexity and training costs, and improving model deployment. Hence, this paper summarized the state-of-the-art techniques for model compression, including model pruning, parameter quantization, low-rank decomposition, knowledge distillation, and lightweight model design. In addition, this paper discusses research challenges and directions for future work.

Research on digital cultural heritage is concerned with the implementation of projection mapping (PJM) technologies, projection viewing, and interactive programs at cultural heritage sites. As PJM technology has come to play an increasingly important role in attracting visitors to museums, heritage pavilions and heritage sites, the topics of digital cultural heritage and digital cultural tourism have become widely discussed in professional and academic circles. However, questions have begun to emerge over the past decade about the relevance of the content presented using PJM at heritage sites and tourist attractions to the sites’ cultural value, and various researchers have attempted to evaluate the effectiveness of PJM on the visitor experience and generate proposals for improvement. Unfortunately, the usefulness of these attempts has been limited by several methodological shortcomings. Therefore, this study proposes an original system for evaluating visitor’s cultural experiences. By evaluating the effectiveness of PJM on visitor’s cultural experiences, a methodology and a set of guidelines for applying PJM that promotes cultural understanding were proposed, and further to achieve an integrated understanding of visitor’s tendency to recall PJM information. Furthermore, a trial run of the system was conducted by the authors in a study of a digital media campaign in October 2021 and the data derived from this investigation are presented in this article as a reference point for comparable cultural heritage and tourism sites.

Deep neural networks (DNNs) have made significant achievements in a wide variety of domains. For the deep learning tasks, multiple excellent hardware platforms provide efficient solutions, including graphics processing units (GPUs), central processing units (CPUs), field programmable gate arrays (FPGAs), and application-specific integrated circuit (ASIC). Nonetheless, CPUs outperform other solutions including GPUs in many cases for the inference workload of DNNs with the support of various techniques, such as the high-performance libraries being the basic building blocks for DNNs. Thus, CPUs have been a preferred choice for DNN inference applications, particularly in the low-latency demand scenarios. However, the DNN inference efficiency remains a critical issue, especially when low latency is required under conditions with limited hardware resources, such as embedded systems. At the same time, the hardware features have not been fully exploited for DNNs and there is much room for improvement. To this end, this paper conducts a series of experiments to make a thorough study for the inference workload of prominent state-of-the-art DNN architectures on a single-instruction-multiple-data (SIMD) CPU platform, as well as with widely applicable scopes for multiple hardware platforms. The study goes into depth in DNNs: the CPU kernel-instruction level performance characteristics of DNNs including branches, branch prediction misses, cache misses, etc, and the underlying convolutional computing mechanism at the SIMD level; The thorough layer-wise time consumption details with potential time-cost bottlenecks; And the exhaustive dynamic activation sparsity with exact details on the redundancy of DNNs. The research provides researchers with comprehensive and insightful details, as well as crucial target areas for optimising and improving the efficiency of DNNs at both the hardware and software levels.

The hypothalamus is the coordination center of the sheep ( Ovis aries ) endocrine system and plays an important role in the reproductive processes of sheep. However, the specific mechanism by which the hypothalamus affects sheep reproductive performance remains unclear. In this study, the hypothalamus tissues of high-reproduction small-tailed Han sheep and low-reproduction Wadi sheep were collected, and full-length transcriptome sequencing by Oxford Nanopore Technologies (ONT) was performed to explore the key functional genes associated with sheep fecundity. The differentially expressed genes (DEGs) were screened and enriched using DESeq2 software through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). Approximately 41.75 million clean reads were obtained from the hypothalamus tissues of high- and low-reproduction sheep, after quality control, 32,194,872 high-quality full-length sequences and 2,114 DEGs were obtained, including 1,247 upregulated genes and 867 downregulated genes ( P adjust < 0.05, |log2FC|>1). Some DEGs were enriched in oocyte meiosis, progesterone-mediated oocyte maturation, estrogen signaling pathway, GnRH signaling pathway and other development-related signaling pathways. The constructed protein-protein interaction (PPI) networks identified the reproduction-related genes, such as GSK3B , PPP2R1B , and PPP2CB . The results of this study will enrich and supplement the genomic information available for small-tailed Han sheep and Wadi sheep, as well as expand the understanding of the molecular mechanisms underlying the regulation of animal reproduction by the hypothalamus, and they also provided reference data for further investigations on the mechanism of high reproduction in sheep.

Lower limb exoskeleton robots have shown significant research value due to their capabilities of providing assistance to wearers and improving physical motion functions. As a type of robotic technology, wearable robots are directly in contact with the wearer’s limbs during operation, necessitating a high level of human–robot collaboration to ensure safety and efficacy. Furthermore, gait prediction for the wearer, which helps to compensate for sensor delays and provide references for controller design, is crucial for improving the the human–robot collaboration capability. For gait prediction, the plantar force intrinsically reflects crucial gait patterns regardless of individual differences. To be exact, the plantar force encompasses a doubled three-axis force, which varies over time concerning the two feet, which also reflects the gait patterns indistinctly. In this paper, we developed a transformer-based neural network (TFSformer) comprising convolution and variational mode decomposition (VMD) to predict bilateral hip and knee joint angles utilizing the plantar pressure. Given the distinct information contained in the temporal and the force-space dimensions of plantar pressure, the encoder uses 1D convolution to obtain the integrated features in the two dimensions. As for the decoder, it utilizes a multi-channel attention mechanism to simultaneously focus on both dimensions and a deep multi-channel attention structure to reduce the computational and memory consumption. Furthermore, VMD is applied to networks to better distinguish the trends and changes in data. The model is trained and tested on a self-constructed dataset that consists of data from 35 volunteers. The experimental results show that FTSformer reduces the mean absolute error (MAE) up to 10.83%, 15.04% and 8.05% and the mean squared error (MSE) by 20.40%, 29.90% and 12.60% compared to the CNN model, the transformer model and the CNN transformer model, respectively.

Cities play a vital role in social development, which contribute to more than 70% of global carbon emission. Low-carbon city construction and decarbonization of the energy sector are the critical strategies to cope with the increasingly serious climate change problems, and low-carbon technologies have attracted extensive attention. However, the potential of such technologies to reduce carbon emissions is constrained by various factors, such as space, operational environment, and safety concerns. As an essential territorial natural resource, underground space can provide large-scale and stable space support for existing low-carbon technologies. Integrating underground space and low-carbon technologies could be a promising approach towards carbon neutrality, and hence, warrants further exploration. First, a comprehensive review of the existing low-carbon technologies including the technical bottlenecks is presented. Second, the features of underground space and its low carbon potential are summarized. Moreover, a framework for the underground space based integrated energy system is proposed, including system configuration, operational mechanisms, and the resulting benefits. Finally, the research prospect and key challenges required to be settled are highlighted.

Due to the workforce shortage caused by the declining birth rate and aging population, robotics is one of the solutions to replace humans and overcome this urgent problem. This paper introduces a deep learning-based object detection algorithm for empty-dish recycling robots to automatically recycle dishes in restaurants and canteens, etc. In detail, a lightweight object detection model YOLO-GD (Ghost Net and Depthwise convolution) is proposed for detecting dishes in images such as cups, chopsticks, bowls, towels, etc., and an image processing-based catch point calculation is designed for extracting the catch point coordinates of the different-type dishes. The coordinates are used to recycle the target dishes by controlling the robot arm. Jetson Nano is equipped on the robot as a computer module, and the YOLO-GD model is also quantized by TensorRT for improving the performance. The experimental results demonstrate that the YOLO-GD model is only 1/5 size of the state-of-the-art model YOLOv4, and the mAP of YOLO-GD achieves 97.38%, 3.41% higher than YOLOv4. After quantization, the YOLO-GD model decreases the inference time per image from 207.92 ms to 32.75 ms, and the mAP is 97.42%, which is slightly higher than the model without quantization. Through the proposed image processing method, the catch points of various types of dishes are effectively extracted. The functions of empty-dish recycling are realized and will lead to further development toward practical use.

Currently, lithium-ion batteries have an increasingly urgent need for high-performance electrolytes, and additives are highly valued for their convenience and cost-effectiveness features. In this work, the feasibilities of fullerenes and fluorinated fullerenes as typical bis(fluorosulfonyl)imide/1,2-dimethoxymethane (LiFSI/DME) electrolyte additives are rationally evaluated based on density functional theory calculations and molecular dynamic simulations. Interestingly, electronic structures of C60, C60F2, C60F4, C60F6, 1-C60F8, and 2-C60F8 are found to be compatible with the properties required as additives. It is noted that that different numbers and positions of F atoms lead to changes in the deformation and electronic properties of fullerenes. The F atoms not only show strong covalent interactions with C cages, but also affect the C-C covalent interaction in C cages. In addition, molecular dynamic simulations unravel that the addition of trace amounts of C60F4, C60F6, and 2-C60F8 can effectively enhance the Li+ mobility in LiFSI/DME electrolytes. The results expand the range of applications for fullerenes and their derivatives and shed light on the research into novel additives for high-performance electrolytes.

Spiking neural networks (SNNs) are known as third generation neural networks due to their energy efficient and low power consumption. SNNs have received a lot of attention due to their biological plausibility. SNNs are closer to the way biological neural systems work by simulating the transmission of information through discrete spiking signals between neurons. Influenced by the great potential shown by the attention mechanism in convolutional neural networks, Therefore, we propose a Spiking Attention Neural Network (Sa-SNN). The network includes a novel Spiking-Efficient Channel Attention (SECA) module that adopts a local cross-channel interaction strategy without dimensionality reduction, which can be achieved by one-dimensional convolution. It is implemented by convolution, which involves a small number of model parameters but provides a significant performance improvement for the network. The design of local inter-channel interactions through adaptive convolutional kernel sizes, rather than global dependencies, allows the network to focus more on the selection of important features, reduces the impact of redundant features, and improves the network’s recognition and generalisation capabilities. To investigate the effect of this structure on the network, we conducted a series of experiments. Experimental results show that Sa-SNN can perform image classification tasks more accurately. Our network achieved 99.61%, 99.61%, 94.13%, and 99.63% on the MNIST, Fashion-MNIST, N-MNIST datasets, respectively, and Sa-SNN performed well in terms of accuracy compared with mainstream SNNs.

Amid escalating climate challenges, China’s carbon neutrality objectives necessitate energy electrification as a pivotal strategy. As a critical load hub, East China demonstrates significant trends toward cleaner energy—marked by growing renewable energy penetration and accelerated cross-regional direct current (DC) transmission deployment. Ensuring stable and efficient grid operation requires rigorous assessment of the impacts of ultra-high voltage DC (UHVDC) integration on grid stability. This study introduces the transient multi-feed-in short circuit ratio (TMSCR), a novel metric for evaluating new DC transmission systems’ influence on grid performance. We systematically investigate UHVDC integration within the East China power grid, emphasizing strategic DC landing point placement. Using TMSCR, the effects of diverse DC incorporation methods are analyzed. Furthermore, this research examines impacts of new DC connections on local and main grids, proposing targeted mitigation measures to enhance grid resilience. This comprehensive UHVDC impact analysis addresses a critical literature gap, providing actionable insights for East China power grid planning and establishing a foundation for subsequent grid planning and DC project feasibility studies during the ‘15th Five-Year Plan’ period.