A CTR Prediction Approach for Text Advertising Based on the SAE-LR Deep Neural Network

Abstract

For the autoencoder (AE) implemented as a construction component, this paper uses the method of greedy layer-by-layer pre-training without supervision to construct the stacked autoencoder (SAE) to extract the abstract features of the original input data, which is regarded as the input of the logistic regression (LR) model, after which the click-through rate (CTR) of the user to the advertisement under the contextual environment can be obtained. These experiments show that, compared with the usual logistic regression model and support vector regression model used in the field of predicting the advertising CTR in the industry, the SAE-LR model has a relatively large promotion in the AUC value. Based on the improvement of accuracy of advertising CTR prediction, the enterprises can accurately understand and have cognition for the needs of their customers, which promotes the multi-path development with high efficiency and low cost under the condition of internet finance.

Versatile Detection of Diverse Selective Sweeps with Flex-Sweep

“…To investigate the influence and impact of the different statistics on the model predictions, we compared the classification results of the network using only the new statistics with only the previously published statistics, the order of the statistics in the feature vector, and explored how each statistic contributes to Flex-sweep classification using Deep SHAP ( Lundberg and Lee 2017 ). The combination of all 11 statistics has substantially better power than either group alone ( supplementary fig.…”

“…Statistics vary in their importance in contributing to the model's classification depending on training and testing data, based on SHAP analyses ( Lundberg and Lee 2017 ). Both highfreq and hapDAF -s, calculated for the middle as well as the shoulders of the locus at various window sizes, make up most of the top 10 most important features for models trained and tested with simulations generated under the human demographic model, equilibrium demographic model, and demographic model with a population decline.…”

“…To explore how each statistic contributes to Flex-sweep classification, we used Deep SHAP (SHapley Additive exPlanations) to interpret the contribution of each feature (a statistic calculated at a region in the locus over a specific window size) ( Lundberg and Lee 2017 ). See supplementary methods, Supplementary Material online, for more details.…”

“…See supplementary methods, Supplementary Material online, for more details. We calculated SHAP values ( Lundberg and Lee 2017 ) for the equilibrium-trained, YRI demography-trained, and decline-trained models and interpreted the results in their corresponding testing data following two approaches: 1) clustering features by similarity and 2) averaging the importance across each statistic at small (50–100 kb), medium (200–500 kb), and large (1 Mb) window sizes in the middle (the seven windows in the center of the locus) and shoulders (the leftmost and rightmost seven windows of the locus) of each region.…”

Multi-modal recommendation algorithm fusing visual and textual features

“…Another example is that it is reasonable to recommend a game console to a 9-year-old boy, where <gender = male, age = 9, product category = game console>is an important set of third-order feature combinations, and these important feature combinations will bring a great improvement to the prediction performance. Traditional CTR prediction mainly relies on manually crafted features and uses shallow models for prediction, such as Logistic Regression (LR) [2][3][4]. However, manual feature engineering greatly relies on domain experts and makes it difficult to find all useful combinations of features.…”