A5062326984- Topic Modeling
- Advanced Text Analysis Techniques
- Music and Audio Processing
- Healthcare Systems and Reforms
- Climate Change and Health Impacts
- Speech and Audio Processing
- Natural Language Processing Techniques
- Global Health Care Issues
- Air Quality and Health Impacts
- Health disparities and outcomes
- Stock Market Forecasting Methods
- Animal Vocal Communication and Behavior
- Chronic Disease Management Strategies
Xiangyang Central Hospital
2024
Hubei University of Arts and Science
2024
Wuhan Institute of Technology
2021
To protect endangered birds, a bird song classification model combining bi-directional long and short term memory neural network (Bi-LSTM) dense convolutional (DenseNet) is proposed. The main operations are as follows: Step 1: Classify, filter extract various features such Mel frequency cepstrum coefficients dump them into TFRecord input data for the model. 2: Build Bi-LSTM-DenseNet network; use cross-entropy loss function to tune structure; softmax classifier classify 20 species, save...
Background Multimorbidity has become a major public health problem among Chinese middle-aged and older adults, the most costly to care system. However, previous population-based studies of multimorbidity have focused on limited number chronic diseases, diagnosis was based participants’ self-report, which may oversimplify problem. At same time, there were few reports relationship between patterns costs. This study analyzed changes people in China over past decade, their association with...
Abstract Event subject extraction was to extract subjects of specific event types. For the traditional BiLSTM network, threshold is complicated, required parameters are many, and time cost high. This paper oriented financial field proposes a method introducing multi-head attention mechanism based on BIGRU network subjects. First, text vectorized, then word vector obtained input into learn context features, introduce depth feature values text. Finally, comparative experiment conducted data...
To address the problems that traditional neural Seq2Seq model is difficult to replicate factual details, cannot handle words out of word list, and generated content prone self-duplication, a generative summary based on Transformer Encoder-PGN with cross-layer parameter sharing proposed for Chinese news texts. The contextual embedding vector first obtained using Encoder pre-trained across layers shared parameters combined location features. Words are then copied from original text copying...