A Convolutional Neural Network for Modelling Sentences

Kalchbrenner, Nal; Grefenstette, Edward; Blunsom, Phil

doi:10.48550/arxiv.1404.2188

Cited by 204 publications

(259 citation statements)

References 0 publications

Supporting

Mentioning

259

Contrasting

Order By: Relevance

“…In order to make optimal use of the complexity and (dis)fluency features, which are sequential in nature, we built convolutional neural network (CNN) models. Originally proposed in computer vision, CNNs have been successfully adapted to various NLP tasks [24] and sentence classification tasks [25][26] [27]. The CNN model has the advantage over models that rely on aggregated features, e.g.…”

Section: Cnn Complexity + (Dis)fluency Modelsmentioning

confidence: 99%

Alzheimer's Disease Detection from Spontaneous Speech through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models

Qiao¹,

Yin²,

Wiechmann³

et al. 2021

Preprint

View full text Add to dashboard Cite

In this paper, we combined linguistic complexity and (dis)fluency features with pretrained language models for the task of Alzheimer's disease detection of the 2021 ADReSSo (Alzheimer's Dementia Recognition through Spontaneous Speech) challenge. An accuracy of 83.1% was achieved on the test set, which amounts to an improvement of 4.23% over the baseline model. Our best-performing model that integrated component models using a stacking ensemble technique performed equally well on cross-validation and test data, indicating that it is robust against overfitting.

show abstract

Section: Cnn Complexity + (Dis)fluency Modelsmentioning

confidence: 99%

Alzheimer's Disease Detection from Spontaneous Speech through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models

Qiao¹,

Yin²,

Wiechmann³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Renowned for their expressiveness and generality, artificial neural networks have become a standard part of nonlinear modeling with widespread use throughout areas as varied as medical [66][67][68] , hyperspectral 69 , and general 70 image classification, regression 71,72 , sentence modeling 73 , social media bot detection 74,75 , and hurricane trajectory and intensity modeling 76,77 . On the dynamical systems front, neural networks have been utilized for the accurate approximation of functions and their derivatives 78 , system approximation 79 , identification and control 36 , and modeling 38 .…”

Section: B Neural Network Based Integratormentioning

confidence: 99%

Global and Local Reduced Models for Interacting, Heterogeneous Agents

Thiem,

Kemeth,

Bertalan

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…The commonly used neural networks can be roughly divided into spatial-sensitive networks and relational-sensitive networks. Convolutional Neural Networks (CNN) [16,18,32] and Recurrent Neural Networks (RNN) [24,31,35] are two classic spatial-sensitive networks which aim to capture spatial correlation among pixels or words. Relationalsensitive networks like Graph Convolutional Networks (GCN) [12,20,37,40] and Self-Attention [7,34] are designed to extract similarity of node pairs where nodes are highly correlated.…”

Section: Introductionmentioning

confidence: 99%

TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data

Du,

Gao,

Chen

et al. 2021

Preprint

View full text Add to dashboard Cite

Tabular data are ubiquitous for the widespread applications of tables and hence have attracted the attention of researchers to extract underlying information. One of the critical problems in mining tabular data is how to understand their inherent semantic structures automatically. Existing studies typically adopt Convolutional Neural Network (CNN) to model the spatial information of tabular structures yet ignore more diverse relational information between cells, such as the hierarchical and paratactic relationships. To simultaneously extract spatial and relational information from tables, we propose a novel neural network architecture, TabularNet. The spatial encoder of TabularNet utilizes the row/column-level Pooling and the Bidirectional Gated Recurrent Unit (Bi-GRU) to capture statistical information and local positional correlation, respectively. For relational information, we design a new graph construction method based on the WordNet tree and adopt a Graph Convolutional Network (GCN) based encoder that focuses on the hierarchical and paratactic relationships between cells. Our neural network architecture can be a unified neural backbone for different understanding tasks and utilized in a multitask scenario. We conduct extensive experiments on three classification tasks with two real-world spreadsheet data sets, and the results demonstrate the effectiveness of our proposed TabularNet over state-of-the-art baselines.

show abstract

A Convolutional Neural Network for Modelling Sentences

Cited by 204 publications

References 0 publications

Alzheimer's Disease Detection from Spontaneous Speech through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models

Alzheimer's Disease Detection from Spontaneous Speech through Combining Linguistic Complexity and (Dis)Fluency Features with Pretrained Language Models

Global and Local Reduced Models for Interacting, Heterogeneous Agents

TabularNet: A Neural Network Architecture for Understanding Semantic Structures of Tabular Data

Contact Info

Product

Resources

About