Combining Domain Knowledge and Deep Learning Makes NMT More Adaptive

2017 ◽  
pp. 90-101
Author(s):  
Liang Ding ◽  
Yanqing He ◽  
Lei Zhou ◽  
Qingmin Liu
Keyword(s):  
Deep Learning ◽  
Domain Knowledge

scholarly journals Domain knowledge integration into deep learning for typhoon intensity classification

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Maiki Higa ◽  
Shinya Tanahara ◽  
Yoshitaka Adachi ◽  
Natsumi Ishiki ◽  
Shin Nakama ◽  
...  
Keyword(s):  
Deep Learning ◽  
Domain Knowledge ◽  
Knowledge Integration ◽  
Cluster Formation ◽  
Satellite Image ◽  
Image Features ◽  
Feature Maps ◽  
Typhoon Intensity ◽  
Intensity Class ◽  
Intensity Classification

AbstractIn this report, we propose a deep learning technique for high-accuracy estimation of the intensity class of a typhoon from a single satellite image, by incorporating meteorological domain knowledge. By using the Visual Geometric Group’s model, VGG-16, with images preprocessed with fisheye distortion, which enhances a typhoon’s eye, eyewall, and cloud distribution, we achieved much higher classification accuracy than that of a previous study, even with sequential-split validation. Through comparison of t-distributed stochastic neighbor embedding (t-SNE) plots for the feature maps of VGG with the original satellite images, we also verified that the fisheye preprocessing facilitated cluster formation, suggesting that our model could successfully extract image features related to the typhoon intensity class. Moreover, gradient-weighted class activation mapping (Grad-CAM) was applied to highlight the eye and the cloud distributions surrounding the eye, which are important regions for intensity classification; the results suggest that our model qualitatively gained a viewpoint similar to that of domain experts. A series of analyses revealed that the data-driven approach using only deep learning has limitations, and the integration of domain knowledge could bring new breakthroughs.

Exploring Biases Between Human and Machine Generated Designs

2018 ◽  
Vol 141 (2) ◽  
Author(s):  
Christian E. Lopez ◽  
Scarlett R. Miller ◽  
Conrad S. Tucker
Keyword(s):  
Computer Simulation ◽  
Deep Learning ◽  
Domain Knowledge ◽  
Mechanical Turk ◽  
Amazon Mechanical Turk ◽  
Design Tools ◽  
Generative Design ◽  
Generate Functional ◽  
Intended Function ◽  
2D Sketches

The objective of this work is to explore the possible biases that individuals may have toward the perceived functionality of machine generated designs, compared to human created designs. Toward this end, 1187 participants were recruited via Amazon mechanical Turk (AMT) to analyze the perceived functional characteristics of both human created two-dimensional (2D) sketches and sketches generated by a deep learning generative model. In addition, a computer simulation was used to test the capability of the sketched ideas to perform their intended function and explore the validity of participants' responses. The results reveal that both participants and computer simulation evaluations were in agreement, indicating that sketches generated via the deep generative design model were more likely to perform their intended function, compared to human created sketches used to train the model. The results also reveal that participants were subject to biases while evaluating the sketches, and their age and domain knowledge were positively correlated with their perceived functionality of sketches. The results provide evidence that supports the capabilities of deep learning generative design tools to generate functional ideas and their potential to assist designers in creative tasks such as ideation.

scholarly journals Zebrafish Embryo Vessel Segmentation Using a Novel Dual ResUNet Model

2019 ◽  
Vol 2019 ◽  
pp. 1-14 ◽  
Author(s):  
Kun Zhang ◽  
Hongbin Zhang ◽  
Huiyu Zhou ◽  
Danny Crookes ◽  
Ling Li ◽  
...  
Keyword(s):  
Deep Learning ◽  
Domain Knowledge ◽  
Zebrafish Embryo ◽  
Fluorescent Protein ◽  
Vessel Segmentation ◽  
Dual Model ◽  
Shape Constraint ◽  
Learning Framework ◽  
Robust Segmentation ◽  
Characteristic Features

Zebrafish embryo fluorescent vessel analysis, which aims to automatically investigate the pathogenesis of diseases, has attracted much attention in medical imaging. Zebrafish vessel segmentation is a fairly challenging task, which requires distinguishing foreground and background vessels from the 3D projection images. Recently, there has been a trend to introduce domain knowledge to deep learning algorithms for handling complex environment segmentation problems with accurate achievements. In this paper, a novel dual deep learning framework called Dual ResUNet is developed to conduct zebrafish embryo fluorescent vessel segmentation. To avoid the loss of spatial and identity information, the U-Net model is extended to a dual model with a new residual unit. To achieve stable and robust segmentation performance, our proposed approach merges domain knowledge with a novel contour term and shape constraint. We compare our method qualitatively and quantitatively with several standard segmentation models. Our experimental results show that the proposed method achieves better results than the state-of-art segmentation methods. By investigating the quality of the vessel segmentation, we come to the conclusion that our Dual ResUNet model can learn the characteristic features in those cases where fluorescent protein is deficient or blood vessels are overlapped and achieves robust performance in complicated environments.

scholarly journals SCFont: Structure-Guided Chinese Font Generation via Deep Stacked Networks

2019 ◽  
Vol 33 ◽  
pp. 4015-4022 ◽  
Author(s):  
Yue Jiang ◽  
Zhouhui Lian ◽  
Yingmin Tang ◽  
Jianguo Xiao
Keyword(s):  
Deep Learning ◽  
Domain Knowledge ◽  
State Of The Art ◽  
Automatic Generation ◽  
Chinese Characters ◽  
Generation System ◽  
Model Learning ◽  
High Quality ◽  
Large Numbers ◽  
Quantitative Assessments

Automatic generation of Chinese fonts that consist of large numbers of glyphs with complicated structures is now still a challenging and ongoing problem in areas of AI and Computer Graphics (CG). Traditional CG-based methods typically rely heavily on manual interventions, while recentlypopularized deep learning-based end-to-end approaches often obtain synthesis results with incorrect structures and/or serious artifacts. To address those problems, this paper proposes a structure-guided Chinese font generation system, SCFont, by using deep stacked networks. The key idea is to integrate the domain knowledge of Chinese characters with deep generative networks to ensure that high-quality glyphs with correct structures can be synthesized. More specifically, we first apply a CNN model to learn how to transfer the writing trajectories with separated strokes in the reference font style into those in the target style. Then, we train another CNN model learning how to recover shape details on the contour for synthesized writing trajectories. Experimental results validate the superiority of the proposed SCFont compared to the state of the art in both visual and quantitative assessments.

scholarly journals Integrate multi-omics data with biological interaction networks using Multi-view Factorization AutoEncoder (MAE)

BMC Genomics ◽  
2019 ◽  
Vol 20 (S11) ◽  
Author(s):  
Tianle Ma ◽  
Aidong Zhang
Keyword(s):  
Machine Learning ◽  
Deep Learning ◽  
Domain Knowledge ◽  
Large Scale ◽  
Learning Model ◽  
Interaction Networks ◽  
Omics Data ◽  
Biological Interaction ◽  
Molecular Features ◽  
Training Objective

Abstract Background Comprehensive molecular profiling of various cancers and other diseases has generated vast amounts of multi-omics data. Each type of -omics data corresponds to one feature space, such as gene expression, miRNA expression, DNA methylation, etc. Integrating multi-omics data can link different layers of molecular feature spaces and is crucial to elucidate molecular pathways underlying various diseases. Machine learning approaches to mining multi-omics data hold great promises in uncovering intricate relationships among molecular features. However, due to the “big p, small n” problem (i.e., small sample sizes with high-dimensional features), training a large-scale generalizable deep learning model with multi-omics data alone is very challenging. Results We developed a method called Multi-view Factorization AutoEncoder (MAE) with network constraints that can seamlessly integrate multi-omics data and domain knowledge such as molecular interaction networks. Our method learns feature and patient embeddings simultaneously with deep representation learning. Both feature representations and patient representations are subject to certain constraints specified as regularization terms in the training objective. By incorporating domain knowledge into the training objective, we implicitly introduced a good inductive bias into the machine learning model, which helps improve model generalizability. We performed extensive experiments on the TCGA datasets and demonstrated the power of integrating multi-omics data and biological interaction networks using our proposed method for predicting target clinical variables. Conclusions To alleviate the overfitting problem in deep learning on multi-omics data with the “big p, small n” problem, it is helpful to incorporate biological domain knowledge into the model as inductive biases. It is very promising to design machine learning models that facilitate the seamless integration of large-scale multi-omics data and biomedical domain knowledge for uncovering intricate relationships among molecular features and clinical features.

scholarly journals Abstractive Text Summary of COVID-19 Documents based on LSTM Method and Word Embedding

Webology ◽  
2021 ◽  
Vol 18 (2) ◽  
pp. 1011-1022
Author(s):  
Saja Naeem Turky ◽  
Ahmed Sabah Ahmed AL-Jumaili ◽  
Rajaa K. Hasoun
Keyword(s):  
Deep Learning ◽  
Domain Knowledge ◽  
Input Sequence ◽  
Word Embedding ◽  
Suggested Model ◽  
Specific Domain ◽  
Scientific Texts ◽  
Proposed Model ◽  
Pass Through ◽  
Single Statement

An abstractive summary is a process of producing a brief and coherent summary that contains the original text's main concepts. In scientific texts, summarization has generally been restricted to extractive techniques. Abstractive methods that use deep learning have proven very effective in summarizing articles in public fields, like news documents. Because of the difficulty of the neural frameworks for learning specific domain- knowledge especially in NLP task, they haven't been more applied to documents that are related to a particular domain such as the medical domain. In this study, an abstractive summary is proposed. The proposed system is applied to the COVID-19 dataset which a collection of science documents linked to the coronavirus and associated illnesses, in this work 12000 samples from this dataset have been used. The suggested model is an abstractive summary model that can read abstracts of Covid-19 papers then create summaries in the style of a single-statement headline. A text summary model has been designed based on the LSTM method architecture. The proposed model includes using a glove model for word embedding which is converts input sequence to vector forms, then these vectors pass through LSTM layers to produce the summary. The results indicate that using an LSTM and glove model for word embedding together improves the summarization system's performance. This system was evaluated by rouge metrics and it achieved (43.6, 36.7, 43.6) for Rouge-1, Rouge-2, and Rouge-L respectively.

scholarly journals Marrying Medical Domain Knowledge With Deep Learning on Electronic Health Records: A Deep Visual Analytics Approach (Preprint)

2020 ◽  
Author(s):  
Rui Li ◽  
Changchang Yin ◽  
Samuel Yang ◽  
Buyue Qian ◽  
Ping Zhang
Keyword(s):  
Heart Failure ◽  
Deep Learning ◽  
Risk Prediction ◽  
Visual Analytics ◽  
Domain Knowledge ◽  
Knowledge Graph ◽  
Data Set ◽  
Medical Domain ◽  
Clinical Risks

BACKGROUND Deep learning models have attracted significant interest from health care researchers during the last few decades. There have been many studies that apply deep learning to medical applications and achieve promising results. However, there are three limitations to the existing models: (1) most clinicians are unable to interpret the results from the existing models, (2) existing models cannot incorporate complicated medical domain knowledge (eg, a disease causes another disease), and (3) most existing models lack visual exploration and interaction. Both the electronic health record (EHR) data set and the deep model results are complex and abstract, which impedes clinicians from exploring and communicating with the model directly. OBJECTIVE The objective of this study is to develop an interpretable and accurate risk prediction model as well as an interactive clinical prediction system to support EHR data exploration, knowledge graph demonstration, and model interpretation. METHODS A domain-knowledge–guided recurrent neural network (DG-RNN) model is proposed to predict clinical risks. The model takes medical event sequences as input and incorporates medical domain knowledge by attending to a subgraph of the whole medical knowledge graph. A global pooling operation and a fully connected layer are used to output the clinical outcomes. The middle results and the parameters of the fully connected layer are helpful in identifying which medical events cause clinical risks. DG-Viz is also designed to support EHR data exploration, knowledge graph demonstration, and model interpretation. RESULTS We conducted both risk prediction experiments and a case study on a real-world data set. A total of 554 patients with heart failure and 1662 control patients without heart failure were selected from the data set. The experimental results show that the proposed DG-RNN outperforms the state-of-the-art approaches by approximately 1.5%. The case study demonstrates how our medical physician collaborator can effectively explore the data and interpret the prediction results using DG-Viz. CONCLUSIONS In this study, we present DG-Viz, an interactive clinical prediction system, which brings together the power of deep learning (ie, a DG-RNN–based model) and visual analytics to predict clinical risks and visually interpret the EHR prediction results. Experimental results and a case study on heart failure risk prediction tasks demonstrate the effectiveness and usefulness of the DG-Viz system. This study will pave the way for interactive, interpretable, and accurate clinical risk predictions.

scholarly journals Deep Learning to Detect Skin Cancer using Google Colab

2019 ◽  
Vol 8 (6) ◽  
pp. 2176-2183 ◽  
Keyword(s):  
Artificial Intelligence ◽  
Machine Learning ◽  
Deep Learning ◽  
Skin Cancer ◽  
Domain Knowledge ◽  
Error Rates ◽  
Data Set ◽  
Training Models ◽  
Learning Machine ◽  
High Level

Different mathematical models, Artificial Intelligence approach and Past recorded data set is combined to formulate Machine Learning. Machine Learning uses different learning algorithms for different types of data and has been classified into three types. The advantage of this learning is that it uses Artificial Neural Network and based on the error rates, it adjusts the weights to improve itself in further epochs. But, Machine Learning works well only when the features are defined accurately. Deciding which feature to select needs good domain knowledge which makes Machine Learning developer dependable. The lack of domain knowledge affects the performance. This dependency inspired the invention of Deep Learning. Deep Learning can detect features through self-training models and is able to give better results compared to using Artificial Intelligence or Machine Learning. It uses different functions like ReLU, Gradient Descend and Optimizers, which makes it the best thing available so far. To efficiently apply such optimizers, one should have the knowledge of mathematical computations and convolutions running behind the layers. It also uses different pooling layers to get the features. But these Modern Approaches need high level of computation which requires CPU and GPUs. In case, if, such high computational power, if hardware is not available then one can use Google Colaboratory framework. The Deep Learning Approach is proven to improve the skin cancer detection as demonstrated in this paper. The paper also aims to provide the circumstantial knowledge to the reader of various practices mentioned above.

scholarly journals Making Radiomics More Reproducible across Scanner and Imaging Protocol Variations: A Review of Harmonization Methods

2021 ◽  
Vol 11 (9) ◽  
pp. 842
Author(s):  
Shruti Atul Mali ◽  
Abdalla Ibrahim ◽  
Henry C. Woodruff ◽  
Vincent Andrearczyk ◽  
Henning Müller ◽  
...  
Keyword(s):  
Deep Learning ◽  
Domain Knowledge ◽  
Data Augmentation ◽  
Image Data ◽  
Clinical Decision ◽  
Generative Adversarial Networks ◽  
Imaging Protocol ◽  
Image Domain ◽  
Style Transfer ◽  
Augmentation Techniques

Radiomics converts medical images into mineable data via a high-throughput extraction of quantitative features used for clinical decision support. However, these radiomic features are susceptible to variation across scanners, acquisition protocols, and reconstruction settings. Various investigations have assessed the reproducibility and validation of radiomic features across these discrepancies. In this narrative review, we combine systematic keyword searches with prior domain knowledge to discuss various harmonization solutions to make the radiomic features more reproducible across various scanners and protocol settings. Different harmonization solutions are discussed and divided into two main categories: image domain and feature domain. The image domain category comprises methods such as the standardization of image acquisition, post-processing of raw sensor-level image data, data augmentation techniques, and style transfer. The feature domain category consists of methods such as the identification of reproducible features and normalization techniques such as statistical normalization, intensity harmonization, ComBat and its derivatives, and normalization using deep learning. We also reflect upon the importance of deep learning solutions for addressing variability across multi-centric radiomic studies especially using generative adversarial networks (GANs), neural style transfer (NST) techniques, or a combination of both. We cover a broader range of methods especially GANs and NST methods in more detail than previous reviews.

Sign in / Sign up

Export Citation Format

Share Document