Particle swarm optimization (PSO) is a well-known optimization algorithm that shows good performances in solving different optimization problems. However, the PSO usually suffers from slow convergence. In this article, a reinforcement-learning-based parameter adaptation method (RLAM) is developed to enhance the PSO convergence by designing a network to control the coefficients of the PSO. Moreover, based on the RLAM, a new reinforcement-learning-based PSO (RLPSO) algorithm is designed. To investigate the performance of the RLAM and RLPSO, experiments on 28 CEC 2013 benchmark functions were carried out to compare with other adaptation methods and PSO variants. The reported computational results showed that the proposed RLAM is efficient and effective and that the proposed RLPSO is superior to several state-of-the-art PSO variants.

Based on recent research in artificial neural networks, researchers have focused on topics from brain-like computing based on the Von Neumann architecture to brain-inspired computing based on the integration of storage and calculation due to the large energy dissipation. Inspired by biological neural networks, people use superconducting devices, which have nonlinear dynamic characteristics, to construct synapse and neuron circuits. However, it is challenging to build large-scale superconducting spiking neural networks due to fan-ins and fan-outs of superconducting devices. In this paper, we present a criterion to scale up the superconducting neural network. Then, we present a three-layer fully connected superconducting spiking neural network construction scheme. Inspired by the biological neural network, we present a new training method based on frequency coding. After only 100 training iterations with one grayscale image of a digit, the learned three-layer fully connected superconducting spiking neural network has a recognition accuracy rate of 86.1% on the digit class. Moreover, the power dissipation of one spiking event in a superconducting synapse is $23.2{\bm{\ zJ}}$ , which shows that the network has outstandingly high efficiency over existing biologically inspired neural networks.

The limited computing resources on edge devices such as Unmanned Aerial Vehicles (UAVs) mean that lightweight object detection algorithms based on convolution neural networks require significant development. However, lightweight models are challenged by small targets with few available features. In this paper, we propose an LC-YOLO model that uses detailed information about small targets in each layer to improve detection performance. The model is improved from the one-stage detector, and contains two optimization modules: Laplace Bottleneck (LB) and Cross-Layer Attention Upsampling (CLAU). The LB module is proposed to enhance shallow features by integrating prior information into the convolutional neural network and maximizing knowledge sharing within the network. CLAU is designed for the pixel-level fusion of deep features and shallow features. Under the combined action of these two modules, the LC-YOLO model achieves better detection performance on the small object detection task. The LC-YOLO model with a parameter quantity of 7.30M achieves an mAP of 94.96% on the remote sensing dataset UCAS-AOD, surpassing the YOLOv5l model with a parameter quantity of 46.61M. The tiny version of LC-YOLO with 1.83M parameters achieves 94.17% mAP, which is close to YOLOv5l. Therefore, the LC-YOLO model can replace many heavyweight networks to complete the small target high-precision detection task under limited computing resources, as in the case of mobile edge-end chips such as UAV onboard chips.

FPGA accelerators for lightweight convolutional neural networks (LWCNNs) have recently attracted significant attention. Most existing LWCNN accelerators focus on single-Computing-Engine (CE) architecture with local optimization. However, these designs typically suffer from high on-chip/off-chip memory overhead and low computational efficiency due to their layer-by-layer dataflow and unified resource mapping mechanisms. To tackle these issues, a novel multi-CE-based accelerator with balanced dataflow is proposed to efficiently accelerate LWCNN through memory-oriented and computing-oriented optimizations. Firstly, a streaming architecture with hybrid CEs is designed to minimize off-chip memory access while maintaining a low cost of on-chip buffer size. Secondly, a balanced dataflow strategy is introduced for streaming architectures to enhance computational efficiency by improving efficient resource mapping and mitigating data congestion. Furthermore, a resource-aware memory and parallelism allocation methodology is proposed, based on a performance model, to achieve better performance and scalability. The proposed accelerator is evaluated on Xilinx ZC706 platform using MobileNetV2 and ShuffleNetV2. Implementation results demonstrate that the proposed accelerator can save up to 68.3% of on-chip memory size with reduced off-chip memory access compared to the reference design. It achieves an impressive performance of up to 2092.4 FPS and a state-of-the-art MAC efficiency of up to 94.58%, while maintaining a high DSP utilization of 95%, thus significantly outperforming current LWCNN accelerators.

As an alternative to radio-frequency (RF) communications, optical wireless communications (OWC) can support high data rates and low power operations while providing good jamming resistance. Our focus in this paper is on deep ultraviolet (UV) outdoor communications (UVOC) where solar blind and non-line-of-sight operations are attractive. Light beams from UV LED arrays serve as information carriers. In an abstract sense, this is similar to directional transmissions in RF; however, the PHY layer characteristics significantly differ due to atmospheric scattering. First, we perform extensive experiments on a UV testbed towards understanding signal propagation and the impact of the PHY on medium access. We find that UV propagation supports (a) fully duplex communications and (b) multiple data rate transmissions. Next, we propose a novel contention-based media access control (UVOC-MAC) protocol that inherently accounts for the UV PHY layer and fully exploits multi-fold spatial reuse opportunities. Evaluations via both simulations and analysis show that UVOC-MAC effectively mitigates collisions and achieves high throughput. In particular, up to a 4-fold increase in throughput and 50% reduction in collision are possible compared to a MAC protocol agnostic to the UV PHY properties.

Blockwise reconstruction with adaptive rounding helps achieve acceptable 4-bit post-training quantization accuracy. However, adaptive rounding is time intensive, and the optimization space of weight elements is constrained to a binary set, thus limiting the performance of quantized models. The optimality of block-wise reconstruction requires that subsequent network blocks remain unquantized. To address this, we propose a two-stage post-training quantization scheme, AE-Qdrop, encompassing block-wise reconstruction and global fine-tuning. In the block-wise reconstruction stage, a progressive optimization strategy is introduced as a replacement for adaptive rounding, enhancing both quantization accuracy and efficiency. Additionally, the integration of randomly weighted quantized activation helps mitigate the risk of overfitting. In the global fine-tuning stage, the weights of each quantized network block are corrected simultaneously through logit matching and feature matching. Experiments in image classification and object detection tasks validate that AE-Qdrop achieves high precision and efficient quantization. For the 2-bit MobileNetV2, AE-Qdrop outperforms Qdrop in quantization accuracy by 6.26%, and its quantization efficiency is fivefold higher.