Recent progress in the development of artificial intelligence technologies, aided by deep learning algorithms, has led to an unprecedented revolution in neuromorphic circuits, bringing us ever closer to brain-like computers. However, the vast majority of advanced algorithms still have to run on conventional computers. Thus, their capacities are limited by what is known as the von-Neumann bottleneck, where the central processing unit for data computation and the main memory for data storage are separated. Emerging forms of non-volatile random access memory, such as ferroelectric random access memory, phase-change random access memory, magnetic random access memory, and resistive random access memory, are widely considered to offer the best prospect of circumventing the von-Neumann bottleneck. This is due to their ability to merge storage and computational operations, such as Boolean logic. This paper reviews the most common kinds of non-volatile random access memory and their physical principles, together with their relative pros and cons when compared with conventional CMOS-based circuits (Complementary Metal Oxide Semiconductor). Their potential application to Boolean logic computation is then considered in terms of their working mechanism, circuit design and performance metrics. The paper concludes by envisaging the prospects offered by non-volatile devices for future brain-inspired and neuromorphic computation.

The electronic structure and the corresponding electrical conductive behavior of the Cu/Cr₂C/TiN stack were assessed according to a newly developed first-principle model based on density functional theory. Using an additional Cr₂C layer provides the metal-like characteristic of the Cu/Cr₂C/TiN stack with much larger electrical conduction coefficients (i.e., mobility, diffusivity, and electrical conductivity) than the conventional Ag/Ti₃C₂/Pt stack due to the lower activation energy. This device is therefore capable of offering faster switching speeds, lower programming voltage, and better stability and durability than the memristor device with conventional Ti₃C₂ MXene.

Unmanned aerial vehicle (UAV) remote-sensing images present unique challenges to the object-detection task due to uneven object densities, low resolution, and drastic scale variations. Downsampling is an important component of deep networks that expands the receptive field, reduces computational overhead, and aggregates features. However, object detectors using multi-layer downsampling result in varying degrees of texture feature loss for various scales in remote-sensing images, degrading the performance of multi-scale object detection. To alleviate this problem, we propose a lightweight texture reconstructive downsampling module called TRD. TRD models part of the texture features lost as residual information during downsampling. After modeling, cascading downsampling and upsampling operators provide residual feedback to guide the reconstruction of the desired feature map for each downsampling stage. TRD structurally optimizes the feature-extraction capability of downsampling to provide sufficiently discriminative features for subsequent vision tasks. We replace the downsampling module of the existing backbone network with the TRD module and conduct a large number of experiments and ablation studies on a variety of remote-sensing image datasets. Specifically, the proposed TRD module improves 3.1% AP over the baseline on the NWPU VHR-10 dataset. On the VisDrone-DET dataset, the TRD improves 3.2% AP over the baseline with little additional cost, especially the ${\mathrm{AP}}_{\mathrm{S}}$, ${\mathrm{AP}}_{\mathrm{M}}$, and ${\mathrm{AP}}_{\mathrm{L}}$ by 3.1%, 8.8%, and 13.9%, respectively. The results show that TRD enriches the feature information after downsampling and effectively improves the multi-scale object-detection accuracy of UAV remote-sensing images.

High-precision cloud detection is crucial for the interpretation of remote sensing images. Although deep learning-based methods have achieved remarkable results, challenges remain in the fine-grained segmentation of cloud boundaries and the detection of thin clouds over complex land surfaces. To this end, this paper proposes a dual-branch co-optimization network (DFC-Net), which achieves cross-scale feature fusion and detail enhancement through the dual-branch interaction mechanism of global context modeling and local detail sensing. In the network architecture, the Residual Fusion Module (ResFM) is designed to construct an enhanced global feature pyramid, which then fuses with the decoder feature to mitigate spatial information loss and strengthen global representation. In addition, the multilevel jump-connection fusion strategy adopted at the decoding end achieves efficient integration of low-level texture features, which significantly improves the model's ability to characterize the fine structure of cloud boundaries. Experimental results on the HRC_WHU and 95-Cloud datasets show that DFC-Net outperforms existing state-of-the-art methods in all performance metrics, and especially exhibits excellent performance advantages in cloud boundary segmentation and thin cloud detection tasks.

Cloud detection constitutes a pivotal task in remote sensing preprocessing, yet detecting cloud boundaries and identifying thin clouds under complex scenarios remain formidable challenges. In response to this challenge, we designed a network model, named NFCNet. The network comprises three submodules: the Hybrid Convolutional Attention Module (HCAM), the Spatial Pyramid Fusion Attention (SPFA) module, and the Dual-Stream Convolutional Aggregation (DCA) module. The HCAM extracts multi-scale features to enhance global representation while matching channel importance weights to focus on features that are more critical to the detection task. The SPFA module employs a novel adaptive feature aggregation method that simultaneously compensates for detailed information lost in the downsampling process and reinforces critical information in upsampling to achieve more accurate discrimination between cloud and non-cloud pixels. The DCA module integrates high-level features with low-level features to ensure that the network maintains its sensitivity to detailed information. Experimental results using the HRC_WHU, CHLandsat8, and 95-Cloud datasets demonstrate that the proposed algorithm surpasses existing optimal methods, achieving finer segmentation of cloud boundaries and more precise localization of subtle thin clouds.