We present a generic framework for data augmentation via dependency subtree swapping that is applicable to machine translation. We extract corresponding subtrees from the dependency parse trees of the source and targe...
详细信息
We introduce HunSum-1: a dataset for Hungarian abstractive summarization, consisting of 1.14M news articles. The dataset is built by collecting, cleaning and deduplicating data from 9 major Hungarian news sites throug...
详细信息
Due to the broad use of deep learning and its need for big data, annotated and available databases for different tasks are constantly appearing. Nevertheless, they often remain unexploited due to the difficulty of eff...
Due to the broad use of deep learning and its need for big data, annotated and available databases for different tasks are constantly appearing. Nevertheless, they often remain unexploited due to the difficulty of effectively performing transfer learning between different databases. In medical imaging, the task of transfer learning is challenging due to: the variety of image modalities, organ/cell shapes, etc., and the lack of available and annotated data. In this paper, we propose an automated pipeline for predicting the similarity values of new database compared to known annotated databases. The system consists of an autoencoder trained on a comprehensive loss function that considers image reconstruction, style features, and dataset membership. A similarity measure is defined based on the resulting 2D latent space, which is demonstrated to have a correlation with the pre-training results on not annotated databases. Hence, our similarity measure could be used to select the most suitable known database for transfer learning or domain adaptation.
Interoperability in the medical field is very important because it allows different healthcare providers, systems, and devices to exchange, interpret, and use patient health information efficiently and accurately. Int...
详细信息
ISBN:
(数字)9798350329520
ISBN:
(纸本)9798350329537
Interoperability in the medical field is very important because it allows different healthcare providers, systems, and devices to exchange, interpret, and use patient health information efficiently and accurately. Interoperability between stomatology and general practitioners (GPs) information systems is still a work in progress but there are a lot of researchers that work to develop applications in this domain. This paper describes the implementation of interoperability between a GP's information system and a stomatology medical system, achieved using the Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) standard. Communication between the two systems was done using Google Firebase services. Using this type of system can significantly decrease medical errors, avoiding administration of substances to which the patient is allergic, during dental procedures. Additionally, it can help GPs prevent exacerbating or triggering other health issues when patients have certain dental problems.
In the last years of this decade, the recognition of human activity has become important to a wide range of researchers in pattern recognition and human-computer interaction as a result of its wide range of real-world...
In the last years of this decade, the recognition of human activity has become important to a wide range of researchers in pattern recognition and human-computer interaction as a result of its wide range of real-world applications, such as gesture recognition, biometric user identification, surveillance by authorities, behavior analysis and health monitoring of the elderly. Human Activity Recognition (HAR) has become a significant topic in mobile and ubiquitous computing as a result of the widespread use of wearable sensor devices and the Internet of Things (loT). Deep Learning (DL) is one of the most commonly used problem-solving techniques in the HAR system. Nevertheless, there are major challenges in applying HAR to problems in recognizing various human activities. In this paper, presented and showed the activities of implementing a new combination of DL methods for multi-class user activity identification to HAR. Using DL methods can be extracting discriminative features automatically from raw sensor data. Specifically, in this work, we proposed a hybrid architecture that features a combination of Bidirectional Long Short-Term Memory (BILSTM) networks and support vector Machines (SVM) for the HAR task. The UCI HAR dataset is used to test the model, it consists of accelerometer and gyroscope data obtained from smartphones. The dataset is split into 30 % for testing and 70% for training. The results for the (BILSTM-SVM) model, showed that the highest accuracy for all users was 98.74 %, higher than all previous models using the same dataset.
We studied the nonlinear properties of some of the most promising nonlinear media for microelectronic applications - AlN and GaN. The nonlinear refractive index n2 and the multiphoton absorption β of the media are me...
详细信息
Agriculture is a vital sector for many countries' economies, but it faces numerous challenges, with weeds being a significant issue. In this study, a deep learning model called YOLOv8-n was implemented on a Raspbe...
详细信息
ISBN:
(数字)9798331505974
ISBN:
(纸本)9798331505981
Agriculture is a vital sector for many countries' economies, but it faces numerous challenges, with weeds being a significant issue. In this study, a deep learning model called YOLOv8-n was implemented on a Raspberry Pi 3B+ equipped with a 1080p resolution camera to detect weeds. The model's capability to identify weeds in real-time was demonstrated using a recorded video. Furthermore, tests conducted with photos taken from a cell phone revealed that the trained model achieved a mean Average Precision at 50 % Intersection over Union (mAP@50) of 0.92, indicating high accuracy. This research underscores the potential of combining deep learning models with cost-effective hardware to enhance agricultural efficiency.
This paper introduces a novel distributed optimization framework for large-scale AC Optimal Power Flow (OPF) problems, offering both theoretical convergence guarantees and rapid convergence in practice. By integrating...
详细信息
The paper describes the structure of an information system for predicting the profile of hot-rolled steel. To predict the transverse profile, an integrated model is used, including models of roll bending, roll profile...
The paper describes the structure of an information system for predicting the profile of hot-rolled steel. To predict the transverse profile, an integrated model is used, including models of roll bending, roll profile wear, roll thermal expansion, and roll flattening. Methods for calculating individual elements of the integrated model are described. A comparison of the predicted profile with the real one is given. Suggested criteria for assessing the quality of the transverse profile are also given. The order of the stages required for the software implementation of the hot rolling mill operator's information system is given.
暂无评论