No abstract
A convolutional neural network-based multiobject detection and tracking algorithm can be applied to vehicle detection and traffic flow statistics, thus enabling smart transportation. Aiming at the problems of the high computational complexity of multiobject detection and tracking algorithms, a large number of model parameters, and difficulty in achieving high throughput with a low power consumption in edge devices, we design and implement a low-power, low-latency, high-precision, and configurable vehicle detector based on a field programmable gate array (FPGA) with YOLOv3 (You-Only-Look-Once-version3), YOLOv3-tiny CNNs (Convolutional Neural Networks), and the Deepsort algorithm. First, we use a dynamic threshold structured pruning method based on a scaling factor to significantly compress the detection model size on the premise that the accuracy does not decrease. Second, a dynamic 16-bit fixed-point quantization algorithm is used to quantify the network parameters to reduce the memory occupation of the network model. Furthermore, we generate a reidentification (RE-ID) dataset from the UA-DETRAC dataset and train the appearance feature extraction network on the Deepsort algorithm to improve the vehicles’ tracking performance. Finally, we implement hardware optimization techniques such as memory interlayer multiplexing, parameter rearrangement, ping-pong buffering, multichannel transfer, pipelining, Im2col+GEMM, and Winograd algorithms to improve resource utilization and computational efficiency. The experimental results demonstrate that the compressed YOLOv3 and YOLOv3-tiny network models decrease in size by 85.7% and 98.2%, respectively. The dual-module parallel acceleration meets the demand of the 6-way parallel video stream vehicle detection with the peak throughput at 168.72 fps.
The interval temporal logic (ITL) model checking (MC) technique enhances the power of intrusion detection systems (IDSs) to detect concurrent attacks due to the strong expressive power of ITL. However, an ITL formula suffers from difficulty in the description of the time constraints between different actions in the same attack. To address this problem, we formalize a novel real-time interval temporal logic-real-time attack signature logic (RASL). Based on such a new logic, we put forward a RASL model checking algorithm. Furthermore, we use RASL formulas to describe attack signatures and employ discrete timed automata to create an audit log. As a result, RASL model checking algorithm can be used to automatically verify whether the automata satisfy the formulas, that is, whether the audit log coincides with the attack signatures. The simulation experiments show that the new approach effectively enhances the detection power of the MC-based intrusion detection methods for a number of telnet attacks, p-trace attacks, and the other sixteen types of attacks. And these experiments indicate that the new algorithm can find several types of real-time attacks, whereas the existing MC-based intrusion detection approaches cannot do that.
The design and analysis of authenticated key exchange protocol is an important problem in information security area. At present, extended Canetti-Krawczyk (eCK) model provides the strongest definition of security for two party key agreement protocol, however most of the current secure protocols can not be prove to secure without Gap assumption. To avoid this phenomenon, by using twinning key technology we propose a new two party key agreement protocol TUP which is obtained by modifying the UP protocol, then in conjunction with the trapdoor test, we prove strictly that the new protocol is secure in eCK model. Compared with previous protocols, the security assumption of new proposal is more standard and weaker, and it also solves an open problem in ProvSec'09
With the rapid development of the Internet of Things (IoT), malicious or affected IoT devices have imposed enormous threats on the IoT environment. To address this issue, trust has been introduced as an important security tool for discovering or identifying abnormal devices in IoT networks. However, evaluating trust for IoT devices is challenging because trust is a degree of belief with regard to various types of trust properties and is difficult to measure. Thus, a machine learning empowered trust evaluation method is proposed in this paper. With this method, the trust properties of network QoS (Quality of Service) are aggregated with a deep learning algorithm to build a behavioral model for a given IoT device, and the timedependent features of network behaviors are fully considered. Trust is also quantified as continuous numerical values by calculating the similarity between real network behaviors and network behaviors predicted by this behavioral model. Trust values can indicate the trust status of a device and are used for decision making. Finally, the proposed method is verified with experiments, and its effectiveness is described.
Accurate prediction of protein binding residues (PBRs) from sequence is important for the understanding of cellular activity and helpful for the design of novel drug. However, experimental methods are time-consuming and expensive. In recent years, a lot of computational predictors based on machine learning and deep learning models are proposed to reduce such consumption. But those methods often use MSA tools such as PSI-BLAST or NetSurfP to generate some statistical features and enter them into predictive models as necessary supplementary input. The input generation process normally takes long time, and there is no standard to specify which and how many statistic results should be provided to a prediction model. In addition, prediction of PBRs relies on residue local context, but the most appropriate scale is undetermined. Most works pre-selected certain residue features as input and a scale size based on expertise for certain type of PBRs. In this study, we propose a general tool-free end-to-end framework that can be applied to all types of PBRs, Multiscale Protein Binding Residues Prediction using language model (MsPBRsP). We adopt a pre-trained language model ProtTrans to save the large consumption caused by MSA tools, and use protein sequence alone as input to our model. To ease scale size uncertainty, we construct multi-size windows in attention layer and multi-size kernels in convolutional layer. We test our framework on various benchmark datasets including PBRs from protein-protein, protein-nucleotide, protein-small ligand, heterodimer, homodimer and antibody-antigen interactions. Compared with existing state-of-the-art methods, MsPBRsP achieves superior performance with less running time and higher prediction rates on every PBRs prediction task. Specifically, we boost F1 score by 27.1% and AUPRC score by 7.6% on NSP448 dataset and decrease running time from over 10 minutes to under 0.1s on average. The source code and datasets are available at https://github.com/biolushuai/MsPBRsP-for-multiple-PBRsprediction.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.