Classifying medical relations in clinical text via convolutional neural networks

He, Bin; Guan, Yi; Dai, Rui

doi:10.1016/j.artmed.2018.05.001

Cited by 63 publications

(31 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Convolutional neural network is a deep feed-forward artificial neural network and one of the representative algorithms of deep learning [40] . It can automatically learn multi-layer features directly from images and has very good representation capabilities.…”

Section: B Basic Structure Of Convolutional Neural Networkmentioning

confidence: 99%

Image Target Recognition Model of Multi- Channel Structure Convolutional Neural Network Training Automatic Encoder

et al. 2020

View full text Add to dashboard Cite

The self-encoder is a typical unsupervised deep learning algorithm. In the field of unsupervised learning, it is very popular with researchers. Therefore, in view of the shortage of labeled training samples, the convolution kernel of a typical convolutional neural network is set by experience, and the network structure is fixed and it is difficult to re-learn later. This paper combines the convolutional neural network and the automatic encoder, and proposes a multi-based the method of integrated network structure to extract the features of the image for recognition. First, the SAE pre-trained CNN model convolution kernel is used to pre-train based on the classic CNN structure. Secondly, input and process image data of different scales to extract image space and spectral features respectively. Then, construct multiple channels, and use different scale filters and sampling intervals for different channels. Finally, after one layer of down sampling, the feature maps obtained from multiple channels are input into the fully connected layer, and after a hidden layer, the features finally used for classification are obtained. Experimental results show that the proposed method uses sparse automatic coding for pre-training time efficiency increased by 50%, and can further improve the recognition accuracy, the highest recognition rate reached 0.985. INDEX TERMS Convolutional neural network; multi-channel; training; automatic encoder; recognition.

show abstract

Section: B Basic Structure Of Convolutional Neural Networkmentioning

confidence: 99%

Image Target Recognition Model of Multi- Channel Structure Convolutional Neural Network Training Automatic Encoder

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Based on the word representation, CNN is used to fuse all local features for predicting topic relations from the global perspective. After convolution, the most useful features of each convolution kernel are extracted by max pooling [57]. The process can be described as follows:…”

Section: A: Relation Feature Modeling Layermentioning

confidence: 99%

A Topic Learning Pipeline for Curating Brain Cognitive Researches

Sheng

Chen

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Cognition is the most basic but complex process of human beings. Benefit from noninvasive neuroimaging technologies, a series of important brain projects have been carried out to model cognition from different aspects and levels. Because modeling such a complex phenomenon requires characterizations of numerous entities and cannot only depend on the efforts of one or more laboratories within a project cycle, a lot of neuroimaging text mining researches have focused on curating neuroimaging-based brain cognitive raw data, derived data and result data, to collect multi-aspect information about brain cognitive researches for comprehensively and objectively characterizing key entities of brain cognition. However, the data-centric perspective leads to the shortcomings of poor topic semantics and topic independent results. This paper proposes a brand-new perspective of big data sharing in neuroimaging, that is, curating brain cognitive researches. A new task definition of neuroimaging text mining and a topic learning pipeline integrating the heterogeneous deep neural networks and density clustering of topic relations are designed to realize this new perspective. The experimental results on actual data sets show that the proposed method can obtain more accurate and complete research topics for effectively characterizing brain cognition and its researches.

show abstract

“…Extracting and classifying relations is crucial for many NLP applications such as question answering and knowledge base completion. He, Guan and Dai propose a convolutional neural network (CNN) architecture with a multi-pooling operation for medical relation classification on clinical records and explore a loss function with a category-level constraint matrix [2]. Experiments using the 2010 i2b2/VA relation corpus are reported.…”

Section: Summary Of the Papersmentioning

confidence: 99%