This paper describes the performance evaluation of an SIMD system on a commonly encountered Standard Linear Equation problem. We introduce a prototype SIMD system, called the Systola 1024, whose parallel architecture ...
详细信息
ISBN:
(纸本)9810474806
This paper describes the performance evaluation of an SIMD system on a commonly encountered Standard Linear Equation problem. We introduce a prototype SIMD system, called the Systola 1024, whose parallel architecture is based upon the Instruction Systolic Array concept. To evaluate the performance of this architecture we have implemented an in-cache problem to measure its MFLOP by solving the Gaussian Elimination problem to obtain the LU factorization. We also examine the effects of increasing the array processors by a certain factor.
Aiming at the problem of image emotion classification, this paper proposes an image emotion classifier based on gene expression programming algorithm. Firstly, we extract HSV spatial histogram and SIFT features from t...
ISBN:
(数字)9781728143903
ISBN:
(纸本)9781728143910
Aiming at the problem of image emotion classification, this paper proposes an image emotion classifier based on gene expression programming algorithm. Firstly, we extract HSV spatial histogram and SIFT features from the image, then integrate them with K-means clustering and spatial pyramid matching. Finally, we design chromosome and fitness function based on image emotional features, then implement the emotional classification algorithm based on GEP. The experimental results show that GEP has a good recognition effect, and the accuracy is improved by 6% compared with traditional SVM algorithm.
The capacity ratio of harmonic suppression device and reactive power compensation device cannot reach the optimal value in the comprehensive power quality control optimization of subway power supply. This situation wi...
详细信息
In order to realize real-time and high-speed distortion removal of camera images, a new improved high resolution and high frame rate image distortion based on FPGA is proposed. The traditional method uses four random ...
详细信息
In order to realize real-time and high-speed distortion removal of camera images, a new improved high resolution and high frame rate image distortion based on FPGA is proposed. The traditional method uses four random access memory (RAM) to buffer the distorted image respectively, and then remaps it with the double rate synchronous dynamic random access memory (DDR), which will cause the system to be unable to process high-resolution images at high frame rates. A new RAM buffer architecture and an interpolation module more suitable for Field Programmable Gate Array (FPGA) implementation are proposed, which can improve the system’s ability to process high-speed and high-resolution distorted images in real time within the acceptable range of resources. Compared with the traditional FPGA de-distortion algorithm, the proposed method can de-distortion with the frame rate of 100 fps and the resolution of 3072 x 2048 x 8bit.
The development of the Industrial Internet of Things (IIoT) is increasingly relying on the low-latency and low-jitter information exchange. Time-Sensitive networking (TSN) guarantees the deterministic transmission by ...
详细信息
According to the problem of over-segmentation using graph-based image segmentation(GBIS), combined with Mean Shift, the paper proposed an improved graph-based image segmentation algorithm(IGBIS) using L*a*b* color spa...
详细信息
Widespread applications of the power electronic-based loads continue to increase concerns over harmonic distortion. This paper therefore proposes a novel fuzzy logic variable hysteresis band current control technique ...
详细信息
Stress is an emotional state that affects our lives, so it is vital to detect it accurately. Numerous studies have shown that tissue oxygen saturation (StO2) is an effective physiological signal for identifying differ...
详细信息
Stress is an emotional state that affects our lives, so it is vital to detect it accurately. Numerous studies have shown that tissue oxygen saturation (StO2) is an effective physiological signal for identifying different stress states. However, previous StO2 feature extraction methods relied too heavily on baselines to be realized in practice. In this paper, a two-stream network combining shallow and deep features, called CRNet, is proposed for baseline-independent StO2 stress classification. In CRNet, shallow features are extracted using a CNN network and deep features are extracted using a ResNet network. For the extracted high-dimensional deep features, MLP module is employed for dimensionality reduction, which can effectively combine the two levels of features and improve the feature extraction capability of the network. Experimental results show that CRNet can significantly improve the performance of StO2 stress classification to the state-of-the-art (SOTA) even when using only raw imbalanced data.
This paper studies the performance of a new protection relay based on fault transient signal analysis in power systems with power electronic converters. The principle of the fault transient signal based protection tec...
详细信息
Nowadays, the construction and application of underwater acoustic (UWA) networks has received much focus in the research area of UWA communication. This paper constructed an underwater acoustic modem based on two type...
详细信息
暂无评论