It is expected that a mixture of autonomous and manual vehicles will persist as part the intelligent transportation system (ITS) for many decades. Thus, addressing safety issues arising from this mix before are entirely popularized crucial. As ITS has increased in complexity, exhibit problems such low intention recognition rate poor real-time performance when predicting driving direction; these seriously affect comfort mixed traffic systems. Therefore, ability to predict direction real time...

Breast cancer, the most common cancer in women, is receiving increasing attention. The lack of high-quality medical resources, especially highly skilled doctors, remote areas makes diagnosis breast inefficient and causes great harm to women. emergence e-health has improved situation a certain extent, but its capabilities are still hampered by technical limitations, which manifest two main aspects. First, due network bandwidth it difficult guarantee real-time transmission pathology images...

Speech emotion recognition (SER) is becoming the main human–computer interaction logic for autonomous vehicles in next generation of intelligent transportation systems (ITSs). It can improve not only safety but also personalized in-vehicle experience. However, current vehicle-mounted SER still suffer from two major shortcomings. One insufficient service capacity vehicle communication network, which unable to meet needs next-generation ITSs terms data transmission rate, power...

The main aim of this paper is to give two general asymptotic expansions for the gamma function, which include Gosper formula as their special cases.Furthermore, we present an inequality function.

Cross-modality knowledge transfer aims to apply learned in the source modality target modality. It is more challenging than general task because of aggravated shift problem due introducing heterogeneous data. This paper proposes a novel fast cross-modality method via contextual autoencoder transformation. In particular, encoder projects representations into Then bridge semantic shared among and modalities, decoder exerts an additional constraint reconstruct original We show that this...

In the era of big data, data annotation is integral to numerous applications. However, it widely acknowledged as a laborious and time-consuming process, significantly impeding scalability efficiency data-driven To reduce human cost, we demonstrate CORAL, collaborative automatic labeling system driven by large language models (LLMs), which achieves high-quality with least effort. Firstly, CORAL employs LLM automatically annotate vast datasets, generating coarse-grained labels. Subsequently,...

Abstract Application development often faces problems such as demand changes, long cycle, and low efficiency. This paper attempts to solve with automated generation technology. The article abstracts the design process of a typical application system; analyzes key issues; then, after designing overall architecture builder, it describes interface presentation layer, control layer (business logic layer) data model layer; finally this challenges encountered by builder improvements in practice.

Currently in the power industry, there has been a demand for intelligent computing and real-time feedback at edge using embedded devices. However, some scenarios, excessive computation still seriously affects practicality of Take transmission inspection drone aerial images as an example, requirement to identify small objects large resolution makes inference model can only be deployed on servers or desktops with strong support. In order realize intelligence devices object detection, this...

Abstract Deep learning has advanced intelligent vehicle development, improving communication and language comprehension. The Transformer model is currently the mainstream approach for text rewriting(coreference resolution) in dialogue systems, aiming to detect rephrase sentences that require modification. Nonetheless, model, consisting of a stack blocks, complex extensive, making it inefficient processing tasks. Accordingly, this paper proposes proficient approach, denoted as FM-Transformer,...

In recent years, as deep learning models represented by CNN have been successfully applied in many computer vision fields, defect detection methods based on also widely various industrial scenarios. To solve the problem of poor effect printed matter task learning, a progressive residual network LSTM was proposed. Firstly, is divided into multiple stages, and each stage passes through cycle before feature extraction. The input output result previous original for channel connection. After...

With the promotion of artificial intelligence technology in various industries China, deep learning driven by electric power data has begun to be applied on a large scale industry, but actual application we have also found some problems, for example, current process and tools, seemingly clear independent, are actually interdependent not flexible enough; common requirements such as retrieval, functions enough often re-developed. Accumulated algorithm exploration will cause same sample appear...

With the development of business, there always exists a large demand for application software in electric power industry. From technical point view, has been no revolutionary changes architecture such recent years. On other hand, softwares are required to meet industry standard like security standard, and pass tests. Theoretically, above two features make works various have high similarity. But practice, amount time personnel spent on does not show significant decrease. To simplify improve...