Search Results for « Wide & Deep Learning »

Generalized linear models with nonlinear feature transformations are widely used for large-scale regression and classification problems with sparse inputs. Memorization of feature interactions through a wide set of cross-product feature transformations are effective and interpretable, while generalization requires more feature engineering effort. With less feature engineering, deep neural networks can generalize better to unseen feature combinations through low-dimensional dense embeddings learned for the sparse features. However, deep neural networks with embeddings can over-generalize and recommend less relevant items when the user-item interactions are sparse and high-rank. In this paper, we present Wide & Deep learning---jointly trained wide linear models and deep neural networks---to combine the benefits of memorization and generalization for recommender systems. We productionized and evaluated the system on Google Play, a commercial mobile app store with over one billion active users and over one million apps. Online experiment results show that Wide & Deep significantly increased app acquisitions compared with wide-only and deep-only models. We have also open-sourced our implementation in TensorFlow.

Wide & Deep, a simple yet effective learning architecture for recommendation systems developed by Google, has had a significant impact in both academia and industry due to its combination of the memorization ability of generalized linear models and the generalization ability of deep models. Graph convolutional networks (GCNs) remain dominant in node classification tasks; however, recent studies have highlighted issues such as heterophily and expressiveness, which focus on graph structure while seemingly neglecting the potential role of node features. In this paper, we propose a flexible framework GCNIII, which leverages the Wide & Deep architecture and incorporates three techniques: Intersect memory, Initial residual and Identity mapping. We provide comprehensive empirical evidence showing that GCNIII can more effectively balance the trade-off between over-fitting and over-generalization on various semi- and full- supervised tasks. Additionally, we explore the use of large language models (LLMs) for node feature engineering to enhance the performance of GCNIII in cross-domain node classification tasks. Our implementation is available at https://github.com/CYCUCAS/GCNIII.

Learning sophisticated feature interactions behind user behaviors is critical in maximizing CTR for recommender systems. Despite great progress, existing methods have a strong bias towards low- or high-order interactions, or rely on expertise feature engineering. In this paper, we show that it is possible to derive an end-to-end learning model that emphasizes both low- and high-order feature interactions. The proposed framework, DeepFM, combines the power of factorization machines for recommendation and deep learning for feature learning in a new neural network architecture. Compared to the latest Wide & Deep model from Google, DeepFM has a shared raw feature input to both its "wide" and "deep" components, with no need of feature engineering besides raw features. DeepFM, as a general learning framework, can incorporate various network architectures in its deep component. In this paper, we study two instances of DeepFM where its "deep" component is DNN and PNN respectively, for which we denote as DeepFM-D and DeepFM-P. Comprehensive experiments are conducted to demonstrate the effectiveness of DeepFM-D and DeepFM-P over the existing models for CTR prediction, on both benchmark data and commercial data. We conduct online A/B test in Huawei App Market, which reveals that DeepFM-D leads to more than 10% improvement of click-through rate in the production environment, compared to a well-engineered LR model. We also covered related practice in deploying our framework in Huawei App Market.

Android malware has exploded in popularity in recent years, due to the platform’s dominance of the mobile market. With the advancement of deep learning technology, numerous deep learning-based works have been proposed for the classification of Android malware. Deep learning technology is designed to handle a large amount of raw and continuous data, such as image content data. However, it is incompatible with discrete features, i.e., features gathered from multiple sources. Furthermore, if the feature set is already well-extracted and sparsely distributed, this technology is less effective than traditional machine learning. On the other hand, a wide learning model can expand the feature set to enhance the classification accuracy. To maximize the benefits of both methods, this study proposes combining the components of deep learning based on multi-branch CNNs (Convolutional Network Neural) with wide learning method. The feature set is evaluated and dynamically partitioned according to its meaning and generalizability to subsets when used as input to the model’s wide or deep component. The proposed model, partition, and feature set quality are all evaluated using the K-fold cross validation method on a composite dataset with three types of features: API, permission, and raw image. The accuracy with Wide and Deep CNN (WDCNN) model is 98.64%, improved by 1.38% compared to RNN (Recurrent Neural Network) model.

Objectives: This study aims to suggest a depression prediction model using variables indirectly related to depression, which helps determine whether the target is depressed. Methods: This study using Korea Institute for Health and Social Affairs (KIHSA)’s 2011, 2014, 2017 National Survey of Older Koreans, and the total number of subjects of the study is 30,571 elderly people 65 years of age or older. Results: The analysis way is deep learning especially Wide & Deep Learning model that separately learns simple data and complex data. The dependent variable is depression, and the independent variable is a total of 30 variables related to personal attributes, economic status, human relationships, health status, health behavior, and life satisfaction. As a result of the main study, all the independent variables used were found to be correlated with depression, performance of Wide & Deep model is 77.9% accuracy, 65.7% recall, 73.8% precision, 85.5% specificity, 69.5% F1-score, 85.3% AUROC, 90.0% AUPRC. Conclusions: It is expected that it will be helpful for the welfare of the elderly by searching for depressed patients at a lower cost and searching for potential depressed patients who want to hide their depression.

The flood of e-commerce platform spam comments affects consumers' purchasing decisions, which greatly damages the interests of consumers.In the process of spam comment recognition, the explicit discrete features of spam comments were usually used as the input of the model.This paper combines the implicit semantic features of spam comments and the explicit discrete features of spam comments to identify the spam comment.First, SMOTE oversampling method is used to balance positive and negative sample sets.Then, wide & deep model, a recommendation system model proposed by Google, is improved and applied to one of the public datasets of spam comment recognition and the commodity datasets collected from one of the biggest e-commerce platform in China.The experimental results show that the improved algorithm can achieve good results in both the gold-standard opinion spam datasets and the commodity datasets.

Aiming at the long-term prediction problem of wireless communication network traffic, a base station network traffic prediction method was proposed based on Wide & Deep learning. Firstly, S-H-ESD (seasonal hybrid extreme studentized deviate test) algorithm and data smoothing method were used to preprocess the network traffic data, and to reduce the impact of noise data on the prediction. Then, the network flow was input into the deep part (neural network) of the Wide & Deep model, the radio resource control (RRC) and physical resource block (PRB) were input into the wide part (linear model) of the Wide & Deep model, and the deep and wide parts were combined to predict the network traffic. The method established one prediction model for the network traffic of all base stations. The root mean squared logarithmic error (RMSLE) of prediction results was 0.985, which was significantly better than that of the traditional seasonal autoregressive integrated moving average model (RMSLE was 2.095) and that of the long short-term memory network model (RMSLE was 3.281). Experimental results showed that the Wide & Deep model could better solve the problem of long-term prediction of wireless network traffic via combining the memory ability of the linear model and the generalization ability of the depth model.

Abstract The high death toll and economic impact of the COVID-19 pandemic emphasize the need for effective population screening technologies. The high cost, limited availability, and slow nature of CT scans and PCR-based tests renders them impractical for frequent use among the general public. As chest X-ray (CXR) imaging is fast and economical, a high accuracy CXR-based test would be well-suited for such screenings. Deep learning algorithms are widely used to aid medical image diagnoses. We use a collection of state-of-the-art pre-trained deep neural network models with additional layers to detect the COVID-19 cases from a sample of healthy, COVID-19, and pneumonia patients. We observed models trained with concatenated features of multiple pre-trained deep learning architectures outperform the individual and ensemble models. Our final model obtained a recall of greater than 98% and a precision of greater than 95% on two separate datasets. The wide architecture of xCovNet may contribute to its robust behavior, providing a systematic approach to construct a reliable deep learning model for emerging datasets.

Accurate prediction of Adverse Drug Reactions (ADRs) holds immense importance in the field of clinical medicine and drug development. The requirement of accurate prediction spans various stages, ranging from drug design and clinical trials to marketing monitoring. The traditional ADR forecasting method has the disadvantage that it requires a lot of computing resources and is not suitable for large-scale forecasting. To address this issue, this study introduces the Wide & Deep model. This model combines the abilities of memorization and generalization to enhance the accuracy of ADR predictions. Additionally, we identify a shortcoming in the wide component of the traditional Wide & Deep model – the lack of nonlinear transformation. Therefore, we propose the inclusion of POLY2 in the Wide & Deep model to rectify this shortcoming. By incorporating POLY2, our aim is to retain the model’s memorization and generalization abilities, leverage the nonlinear relationship between features, and capture the interaction effect between drug chemical substructures for better model performance. To validate our proposed method, we conduct experiments on two datasets: the FDA Adverse Event Reporting System (FAERS) and PubChem. The evaluation metric utilized is the Area Under the Curve (AUC) score, which demonstrates that our method outperforms the original model. The results indicate that by combining POLY2 feature crosses with the Wide & Deep model, we have achieved significant improvements in the prediction of ADRs.