Along with the progression of Internet of Things(IoT)technology,network terminals are becoming continuously more *** has been widely applied in various scenarios,including urban infrastructure,transportation,industry,...
详细信息
Along with the progression of Internet of Things(IoT)technology,network terminals are becoming continuously more *** has been widely applied in various scenarios,including urban infrastructure,transportation,industry,personal life,and other socio-economic *** introduction of deep learning has brought new security challenges,like an increment in abnormal traffic,which threatens network *** feature extraction leads to less accurate classification *** abnormal traffic detection,the data of network traffic is high-dimensional and *** data not only increases the computational burden of model training but also makes information extraction more *** address these issues,this paper proposes an MD-MRD-ResNeXt model for abnormal network traffic *** fully utilize the multi-scale information in network traffic,a Multi-scale Dilated feature extraction(MD)block is *** module can effectively understand and process information at various scales and uses dilated convolution technology to significantly broaden the model’s receptive *** proposed Max-feature-map Residual with Dual-channel pooling(MRD)block integrates the maximum feature map with the residual *** module ensures the model focuses on key information,thereby optimizing computational efficiency and reducing unnecessary information *** results show that compared to the latest methods,the proposed abnormal traffic detection model improves accuracy by about 2%.
Large models have recently played a dominant role in natural language processing and multimodal vision-language learning. However, their effectiveness in text-related visual tasks remains relatively unexplored. In thi...
详细信息
Large models have recently played a dominant role in natural language processing and multimodal vision-language learning. However, their effectiveness in text-related visual tasks remains relatively unexplored. In this paper, we conducted a comprehensive evaluation of large multimodal models, such as GPT4V and Gemini, in various text-related visual tasks including text recognition, scene text-centric visual question answering(VQA), document-oriented VQA, key information extraction(KIE), and handwritten mathematical expression recognition(HMER). To facilitate the assessment of optical character recognition(OCR) capabilities in large multimodal models, we propose OCRBench, a comprehensive evaluation benchmark. OCRBench contains 29 datasets, making it the most comprehensive OCR evaluation benchmark available. Furthermore, our study reveals both the strengths and weaknesses of these models, particularly in handling multilingual text, handwritten text, non-semantic text, and mathematical expression *** importantly, the baseline results presented in this study could provide a foundational framework for the conception and assessment of innovative strategies targeted at enhancing zero-shot multimodal *** evaluation pipeline and benchmark are available at https://***/Yuliang-Liu/Multimodal OCR.
With the development of the Internet of Things (IoT), people's lives have become more intelligent. However, the proliferation of numerous small devices has also introduced serious risks to network security. Networ...
详细信息
Generative Artificial Intelligence (GAI) possesses the capabilities of generating realistic data and facilitating advanced decision-making. By integrating GAI into modern Internet of Things (IoT), Generative Internet ...
详细信息
Foam concrete has the advantages of energy and shock absorption which can be used as an impact resistant material. This article designed a three-layer concrete gradient structure and analyzed the failure mechanism of ...
详细信息
Conventional silicate cement is difficult to use in extremely cold regions. But alkali-activated slag concrete (AASC) can use alkali-activated chemistry to accelerate the production of C-A-S-H cementitious materials, ...
详细信息
With the increasing requirements for supply chain security, traceability and efficiency in the healthcare industry, existing centralized or decentralized management systems show obvious limitations in dynamic environm...
详细信息
The compression and energy absorption properties of foam geopolymers increase stress wave attenuation under explosion impacts,reducing the vibration effect on the *** tests were conducted using several composite struc...
详细信息
The compression and energy absorption properties of foam geopolymers increase stress wave attenuation under explosion impacts,reducing the vibration effect on the *** tests were conducted using several composite structure models,including a concrete lining structure(CLS)without foam geopolymer and six foam geopolymer composite structures(FGCS)with different backfill parameters,to study the dynamic response and wave dissipation mechanisms of FGCS under explosive ***,strain,and vibration responses at different locations were synchronously *** damage modes and dynamic responses of different models were compared,and how wave elimination and energy absorption efficiencies were affected by foam geopolymer backfill parameters was *** results showed that the foam geopolymer absorbed and dissipated the impact energy through continuous compressive deformation under high strain rates and dynamic loading,reducing the strain in the liner structure by 52%and increasing the pressure attenuation rate by 28%.Additionally,the foam geopolymer backfill reduced structural vibration and liner deformation,with the FGCS structure showing 35%less displacement and 70%less acceleration compared to the *** FGCS model with thicker,less dense foam geopolymer backfill,having more pores and higher porosity,demonstrated better compression and energy absorption under dynamic impact,increasing stress wave attenuation *** analyzing the stress wave propagation and the compression characteristics of the porous medium,it was concluded that the stress transfer ratio of FGCS-ρ-579 was 77%lower than that of CLS,and the transmitted wave energy was 90%*** results of this study provide a scientific basis for optimizing underground composite structure interlayer parameters.
In the present work,a multi-element nanoglass(m-NG)of FeCoCrMoCBY is obtained first time by the laser ablation combined with inert gas condensation(laser-IGC)*** with the conventional rapid-quenched metallic glass(MG)...
详细信息
In the present work,a multi-element nanoglass(m-NG)of FeCoCrMoCBY is obtained first time by the laser ablation combined with inert gas condensation(laser-IGC)*** with the conventional rapid-quenched metallic glass(MG)with identical composition,the Fe-based m-NG demonstrates a superior performance as a self-supported electrocatalyst for hydrogen evolution reaction(HER)in acidic *** enhanced HER activity of m-NG is proposed to be closely related to its high en-ergy states,which is originated from the unique inhomogeneous nanostructures with a high density of low-coordinated ***,the Fe-based m-NG exhibits an outstanding comprehensive catalytic performance even beyond the commercial Pt/C catalyst in long-term test due to its self-optimization *** work not only opens the way to the preparation of m-NGs by the novel laser-IGC technique,but also makes a great contribution to developing low-cost,high-efficient,and super-durable HER electrocat-alysts in acidic environment.
To reduce key disagreement rate and increase key generation rate, this paper proposes a lightweight and robust shared secret key extraction scheme from atmospheric optical wireless channel. A conception of grouping sa...
详细信息
暂无评论