Click-through Rate Prediction Model Research Articles

Feature-Interaction-Enhanced Sequential Transformer for Click-Through Rate Prediction

Click-through rate (CTR) prediction plays a crucial role in online services and applications, such as online shopping and advertising. The performance of CTR prediction can have a direct impact on user experience and the revenue of the online platforms. For CTR prediction models, self-attention-based methods have been widely applied to this field. Recent works generally adopted the Transformer architecture, where the self-attention mechanism can capture the global dependencies of the user’s historical interactions and predict the next item. Despite the effectiveness of self-attention methods in modeling sequential user behaviors, most sequential recommenders hardly exploit feature interaction techniques to extract high-order feature combinations. In this paper, we propose a Feature-Interaction-Enhanced Sequence Model (FESeq), which integrates feature interaction and the sequential recommendation model in a cascading structure. Specifically, the interacting layer in FESeq is an automatic feature engineering step for the Transformer model. Then, we add a linear time interval embedding layer and a positional embedding layer to the Transformer in the sequence-refiner layer to learn both the time intervals and the position information in the user’s sequence behaviors. We also design an attention-based sequence pooling layer that can model the relevance of the user’s historical behaviors and the target item representation through scaled bilinear attention. Our experiments show that the proposed method beats all the baselines on both public and industrial datasets.

An ensemble learning framework for click-through rate prediction based on a reinforcement learning algorithm with parameterized actions

Click-through rate (CTR) prediction is essential for targeted advertising systems. Although there have been many studies on CTR prediction and forming some representative models, building a CTR prediction model that performs well under various scenarios remains a formidable challenge. Ensemble learning is a class of methods that improve the final performance by aggregating multiple base models. This paper proposes an ensemble learning framework that can predict the click probability for each ad impression by combining the results of one or more CTR prediction models. The framework utilizes an improved Reinforcement Learning (RL) algorithm that supports parametric operations to dynamically determine the models and their weights that participated in predicting CTR for each ad impression. Specifically, for each ad impression, the agent generates a discrete action and a continuous action vector based on predicted click-through rates by all base models. The discrete action determines the number of base models involved in prediction, while each continuous action decides the weight of each base model. We validate the ensemble learning framework on three benchmark datasets, including iPinYou, Criteo, and Avazu. The experimental results demonstrate that, in most cases, the ensemble learning framework outperforms the base models in terms of AUC and LogLoss metrics. Finally, we give two simplified variants of the ensemble learning framework and discuss their applicability.

Deep Time-Aware Attention Neural Network for Sequential Recommendation

For click-through rate (CTR) prediction tasks, a good prediction performance can be obtained by full explorations of both user behavior and item behavior. Since user’s interests have a great influence on user’s behaviors, it is very important to learn users’ intrinsic interests according to their behaviors. User interests are not only diverse but also in dynamic change. However, the dynamics of user interests’ change are not fully taken into account by the majority of current CTR models. The latest sequential recommendation algorithm ignores the subjectivity of users when it uses a two-layer recurrent neural network to model the item behavior from the perspective of the evolution of items. In this work, we propose a recurrent neural network model called DTIAN (Deep Time-Aware Interest Attention Network) to address these issues. By leveraging the user behaviors and the corresponding temporal information, DTIAN captures user interests and intent changes to the target item. Therefore, the users’ recent interests are enhanced compared to early interests with the attention mechanism. In addition, each module of the proposed model can be plugged into other mainstream models to improve the performance of current models. The experimental results show that the proposed DTIAN can outperform the current popular CTR prediction models slightly and significantly reduce the training time, which makes it possible to implement lightweight models.

CowClip: Reducing CTR Prediction Model Training Time from 12 Hours to 10 Minutes on 1 GPU

The click-through rate (CTR) prediction task is to predict whether a user will click on the recommended item. As mind-boggling amounts of data are produced online daily, accelerating CTR prediction model training is critical to ensuring an up-to-date model and reducing the training cost. One approach to increase the training speed is to apply large batch training. However, as shown in computer vision and natural language processing tasks, training with a large batch easily suffers from the loss of accuracy. Our experiments show that previous scaling rules fail in the training of CTR prediction neural networks. To tackle this problem, we first theoretically show that different frequencies of ids make it challenging to scale hyperparameters when scaling the batch size. To stabilize the training process in a large batch size setting, we develop the adaptive Column-wise Clipping (CowClip). It enables an easy and effective scaling rule for the embeddings, which keeps the learning rate unchanged and scales the L2 loss. We conduct extensive experiments with four CTR prediction networks on two real-world datasets and successfully scaled 128 times the original batch size without accuracy loss. In particular, for CTR prediction model DeepFM training on the Criteo dataset, our optimization framework enlarges the batch size from 1K to 128K with over 0.1% AUC improvement and reduces training time from 12 hours to 10 minutes on a single V100 GPU. Our code locates at github.com/bytedance/LargeBatchCTR.

FinalMLP: An Enhanced Two-Stream MLP Model for CTR Prediction

Click-through rate (CTR) prediction is one of the fundamental tasks in online advertising and recommendation. Multi-layer perceptron (MLP) serves as a core component in many deep CTR prediction models, but it has been widely shown that applying a vanilla MLP network alone is ineffective in learning complex feature interactions. As such, many two-stream models (e.g., Wide&Deep, DeepFM, and DCN) have recently been proposed, aiming to integrate two parallel sub-networks to learn feature interactions from two different views for enhanced CTR prediction. In addition to one MLP stream that learns feature interactions implicitly, most of the existing research focuses on designing another stream to complement the MLP stream with explicitly enhanced feature interactions. Instead, this paper presents a simple two-stream feature interaction model, namely FinalMLP, which employs only MLPs in both streams yet achieves surprisingly strong performance. In contrast to sophisticated network design in each stream, our work enhances CTR modeling through a feature selection module, which produces differentiated feature inputs to two streams, and a group-wise bilinear fusion module, which effectively captures stream-level interactions across two streams. We show that FinalMLP achieves competitive or even better performance against many existing two-stream CTR models on four open benchmark datasets and also brings significant CTR improvements during an online A/B test in our industrial news recommender system. We envision that the simple yet effective FinalMLP model could serve as a new strong baseline for future development of two-stream CTR models. Our source code will be available at MindSpore/models and FuxiCTR/model_zoo.

Feature Dual Supervision Model for the Searches of Online Advertising Audiences

Online advertising has become one of the most important strategies used by companies. They get the valuable results from Internet marketing and communication strategies. Therefore, it is necessary to study the click-through rate (CTR) model to search the potential audiences in online advertising. The advertisers desire to search for potential candidates through a large number of queries for audiences in programmatic advertising. Facing such a large corpus, the most common method is that using two-tower model to learn user’s queries and ad representations, and then the similarity function is applied to match the feature representation to get the potential audiences related to the ad. However, in the process of feature extraction, there is a lack of information interaction between the two towers, resulting in the loss of details in the representation. In order to alleviate the lack of information interaction between the networks in the two-tower model during feature extraction. In this paper, we propose a novel model named Feature Dual Supervision Model (FDSM), which integrates by Feature Expression Unit (FEU) and Feature Supervision Unit (FSU). The FEU is used to extract ads or users features, and FSU generates a weight vector to supervise the working process of the FEU. In addition, we propose a feature cross-layer with bridge connections in FDSM to achieve effective feature interaction between ad and user representations. Finally, we conduct experiments on the Tencent Lookalike and MovieLens datasets. The experimental results indicate that the FDSM model outperforms other state-of-the-art CTR prediction models in audience expansion.

Click-through Rate Prediction and Uncertainty Quantification Based on Bayesian Deep Learning.

Click-through rate (CTR) prediction is a research point for measuring recommendation systems and calculating AD traffic. Existing studies have proved that deep learning performs very well in prediction tasks, but most of the existing studies are based on deterministic models, and there is a big gap in capturing uncertainty. Modeling uncertainty is a major challenge when using machine learning solutions to solve real-world problems in various domains. In order to quantify the uncertainty of the model and achieve accurate and reliable prediction results. This paper designs a CTR prediction framework combining feature selection and feature interaction. In this framework, a CTR prediction model based on Bayesian deep learning is proposed to quantify the uncertainty in the prediction model. On the squeeze network and DNN parallel prediction model framework, the approximate posterior parameter distribution of the model is obtained using the Monte Carlo dropout, and obtains the integrated prediction results. Epistemic and aleatoric uncertainty are defined and adopt information entropy to calculate the sum of the two kinds of uncertainties. Epistemic uncertainty could be measured by mutual information. Experimental results show that the model proposed is superior to other models in terms of prediction performance and has the ability to quantify uncertainty.

An AI-enabled approach for improving advertising identification and promotion in social networks

The rapid development of the social economy has considerably impacted the traditional advertising industry. In terms of identifying and recommending advertisements, improving the accuracy of advertisement promotion and identification is an urgent need. Artificial intelligence (AI) methods are of practical application to facilitate digital transformation in the advertising industry. Among the many AI methods, the network model based on the genetic algorithm back propagation (GABP) neural network is the most compatible with applications in the advertising industry. In this work, the GABP neural network is applied in the construction of social networks to predict the click-through rate (CTR) of website advertising applications through the optimization of advertising promotion strategies. In the processes of optimizing and improving the system model using AI methods, the technological focus is the accuracy of advertising promotion and identification for the developing advertising industry. First, background information on the era of internet and AI development is analyzed. The new media technology is discussed through AI research and traditional advertising industry literature. In addition, a CTR prediction model is created for advertising applications based on the GABP network. In the neural network improvement process, the performance of GABP is enhanced through multiple iterations by optimizing the application strategy of advertising scenarios according to the topology of network connections. The study results demonstrate that different algorithms' recognition accuracy and precision show an increasing trend as the number of model iterations increases. The recognition accuracy of GABP increases from 49 % to 72 %, and the recognition precision of the algorithm increases from 69 % to 86 %. In addition, the area under the curve (AUC) value of the GABP network is only 0.6 before the number of neurons increases. When the number of neurons is 400, the AUC value of the algorithm reaches 0.82, and the comprehensive diagnostic value of the system dramatically improves. This research has significant reference value for reforming the traditional advertising work mode, enhancing intellectual development and promoting resource efficiency in the advertising industry.

MLIM: A CTR prediction model describing evolution law of user interest

With the advent of the digital economy era, business systems such as web advertising and recommendation system have put forward the demand for predicting the click through rate (CTR) of items. However, the current CTR prediction research is not enough to mine user behavior, resulting in the lack of accuracy of user interest representation. In this paper, we propose a CTR prediction model, called MLIM, which can deep mine the evolution law of user interest. Specifically, we first use BiGRU to obtain the low-level user interest representation in the interest extraction layer, and then continue to use attention mechanism, BiGRU and sliding time window multi-components collaborative modeling in the interest evolution layer to obtain multi-level user interest representation with richer information, which can improve the accuracy of CTR prediction to a certain extent. Comprehensive experiments on two real datasets show that the proposed model achieves better performance than the mainstream baselines integrating user behavior analysis.

Exploring different interaction among features for CTR prediction

Advertising Click-Through Rate (CTR) prediction is a relatively successful application in the field of recommendation system. Improving the accuracy of advertising click-through rate can not only make the user experience better, but also give more benefits to advertising platforms and advertisers. It can be seen from the research status that the interaction between learning features has become a very important part of the advertising CTR prediction model. Although the existing CTR prediction models based on deep learning have achieved good results, some models only consider a single interaction mode and lack the diversity of feature interaction. To resolve this problem, this paper proposes a CTR prediction model based on multi-feature interaction, called EDIF, which aims to enhance the diversity of feature interaction. Firstly, the model learns multiple different embedding vectors for each feature in the embedding layer, which reflects the correlation between features; secondly, in the interaction of high-order features, the embedding vectors of each feature are added and pooled to form the aggregation vector of the whole feature as the input, which reflects the integrity of the feature; finally, after the feature embedding operation, the model introduces two layers in parallel: compressed excitation network layer and explicit high-order interaction layer, which improves the ability of feature interaction. We have done a lot of experiments on two public data sets, Avazu and Criteo. The results show that the model effect of this paper has great advantages over the latest model.

Upper confidence bound integrated genetic algorithm‐optimized long short‐term memory network for click‐through rate prediction

AbstractData on online advertising is rising rapidly due to the fast development of science and technology. Click‐through rate (CTR) prediction has become a critical task regarding the digital advertising industry and a key element in increasing advertising profits and user experience. Therefore, this article describes the problem of CTR prediction as a function of sequence classification tasks. Then, we proposed a novel optimization strategy to solve the high‐dimensional problem and find a subset of relevant variables to ensure high performance of our model and maximize the number of clicks. Here, we introduced a feature selection and hyper‐parameter optimization approach using genetic algorithms (GA) and the upper confidence bound (UCB) model to optimize micro‐targeting technology, along with the long short‐term memory (LSTM) network‐based CTR prediction model. The efficiency of the proposed UCB‐LSTM‐GA model and two hybrid models, namely LSTM‐GA and LSTM‐PSO, is evaluated by comparing them to each other and to other machine‐learning‐based classification methods, including LSTM using a UCB algorithm (UCB‐LSTM), High‐order Attentive Factorization Machine (HoAFM), genetic algorithm‐artificial neural network (GA‐ANN), and a feature interaction graph neural network model (Fi‐GNN). Our solution achieved as high as 87%, 89%, and 92% for respectively accuracy, precision, and recall, using the popular python tools with real Avazu datasets.

DistilledCTR: Accurate and scalable CTR prediction model through model distillation

Accuracy and scalability are critical to the efficiency and effectiveness of real-time recommender systems. Recent deep learning-based click-through rate prediction models are improving in accuracy but at the expense of computational complexity. The purpose of this study is to propose an accurate and scalable click-through rate (CTR) prediction model for real-time recommendations. This study investigates the complexity, accuracy, and scalability aspects of various CTR models. This work ensembles top CTR models using a gated network and distill into a deep neural network (DNN) using a knowledge distillation framework. Distilled DNN model is more accurate and 20x scalable than any of the individual CTR models. The low latency of distilled model makes it scalable and fit for deployment in real-time recommender systems. The proposed distillation framework is extensible to integrate any CTR models to the ensemble and can be distilled to any neural architecture.

Research on prediction of user advertising recommendation system based on NFM model

Click-through rate (CTR) predictions can impact business revenue and improve user experience. In large-scale enterprise-level advertising systems and recommendation systems, CTR prediction is a very important link, and it has always been one of the key issues of academic research. In the case of extremely unbalanced positive and negative samples, this paper uses as few positive samples as possible to predict the user's advertising click behavior. To solve the above problems, on the basis of data processing, this paper proposes a CTR prediction model that combines feature engineering and machine learning. In terms of feature engineering, this model selects the relevant features of users and videos from the original data based on the feature selection method, selects important features based on random forest (RF), and forms the interaction features of users and videos by factoring interaction parameters. In terms of machine learning, this model is based on the factorization machine (FM) and the integrated learning model of the deep neural network (DNN), the neural factorization machine (NFM) to realize the click rate prediction model. Based on the NFM model, this paper generates a suitable CTR prediction model through the training of the train data set. After continuous optimization and improvement, the parameters with the best prediction effect are selected. The NFM model established in this paper has a high improvement in performance. Compared with FM, DeepFM, FNN, IFM and DCN models, the AUC value of NFM model is improved by about 0.03 on average. The NFM model can better predict the click behavior of users on advertisements, and can provide decision-making reference for advertisement placement, and can be applied to content recommendation of different advertising systems.

A joint learning model for click-through prediction in display advertising

Click-through rate (CTR) prediction is essential for targeted advertising and recommendation. At present, machine learning models are widely used to build CTR estimators, including logistic regression (LR), factorization machine (FM), and deep neural network (DNN). Unfortunately, these models adopt the single structure that only considers either low-order feature interactions (such as LR and FM) or high-order feature interactions (such as DNN), not sufficient for CTR prediction. Therefore, the joint learning models are proposed, such as Wide & Deep and DeepFM, which can exploit both high- and low-order feature interactions to predict CTR by combining two different models. In this paper, we first analyze the typical CTR prediction models’ structures and performance, and then summarize the general form and design rules of CTR estimators. Based on the general form, we further design a new joint learning model that combines two different residual networks to explore the feature interactions automatically. Compared with the widely adopted feed-forward neural network, the residual network is more capable of exploring complex feature interactions at different layers. Additionally, we introduce a neural attention network to learn the importance of each second-order interaction of features from various fields. Finally, we evaluate the prediction performance of the proposed model based on two real-world datasets (i.e., Criteo and Avazu) in terms of LogLoss and AUC metrics. The extensive experimental results demonstrate our model performs the best compared to the state-of-the-art baselines.

Deep Match to Rank Model for Personalized Click-Through Rate Prediction

Click-through rate (CTR) prediction is a core task in the field of recommender system and many other applications. For CTR prediction model, personalization is the key to improve the performance and enhance the user experience. Recently, several models are proposed to extract user interest from user behavior data which reflects user's personalized preference implicitly. However, existing works in the field of CTR prediction mainly focus on user representation and pay less attention on representing the relevance between user and item, which directly measures the intensity of user's preference on target item. Motivated by this, we propose a novel model named Deep Match to Rank (DMR) which combines the thought of collaborative filtering in matching methods for the ranking task in CTR prediction. In DMR, we design User-to-Item Network and Item-to-Item Network to represent the relevance in two forms. In User-to-Item Network, we represent the relevance between user and item by inner product of the corresponding representation in the embedding space. Meanwhile, an auxiliary match network is presented to supervise the training and push larger inner product to represent higher relevance. In Item-to-Item Network, we first calculate the item-to-item similarities between user interacted items and target item by attention mechanism, and then sum up the similarities to obtain another form of user-to-item relevance. We conduct extensive experiments on both public and industrial datasets to validate the effectiveness of our model, which outperforms the state-of-art models significantly.

A CTR prediction model based on user interest via attention mechanism

Recently, click-through rate (CTR) prediction is a challenge problem in the aspect of online advertising. Some researchers have proposed deep learning-based models that follow a similar embedding and MLP paradigm. However, the corresponding approaches generally ignore the importance of capturing the latent user interest behind user behaviour data. In this paper, we present a novel attentive deep interest-based network model called ADIN. Specifically, we capture the interest sequence in the interest extractor layer, and the auxiliary losses are employed to produce the interest state with the deep supervision. First, we model the dependency between behaviours by using a bidirectional gated recurrent unit (Bi-GRU). Next, we extract the interest evolving process that is related to the target and propose an interest evolving layer. At the same time, attention mechanism is embedded into the sequential structure. Then, the model learns highly non-linear interactions of features based on stack autoencoders. An experiment has been done using four real-world datasets, the proposed model achieves superior performance than the existing state-of-the-art models.

CTR Prediction Models Considering the Dynamics of User Interest

Click-through rate (CTR) prediction is one of the key areas in industrial bidding advertising. Recently, to improve prediction performance, researchers have proposed the interest-based deep models that learn the user's latent interest from historical click behaviors. However, the interest-based deep models lack the ability to trace the dynamics of interest from both the positive and negative samples, which leading to prediction errors. In particular, user's interest in candidate advertisements could be positive (i.e. interested) or negative (i.e. uninterested), and changing over time. To solve this problem, we propose a deep-based dynamic interest perception network (DIPN) that can trace both positive interest and negative interest. The proposed DIPN model introduces three new parts to the interest-based deep model: a gated recurrent unit (GRU) learns the implied interest vector from historical click sequences; an attention mechanism improves the expressiveness of interest vector with the weight vector; an interest degree feature scales the weight vector by the d-Softmax function to improve the expressiveness of interest vector. We evaluate the effectiveness of the DIPN model by conducting extensive comparative experiments using real datasets. The experimental results demonstrate that compared with state-of-the-art models, DIPN achieves the highest prediction performance.

STATISTICAL TESTING TECHNIQUE FOR COMPARISON MACHINE LEARNING MODELS PERFORMANCE

The statistical testing technique is considered to compare the metrics values of machine learning models on a test set. Since the values of metrics depend not only on the models, but also on the data, it may turn out that different models are the best on different test sets. For this reason, the traditional approach to comparing the values of metrics on a test set is often not enough. Sometimes a statistical comparison of the results obtained on the basis of cross-validation is used, but in this case it is impossible to guarantee the independence of the obtained measurements, which does not allow the use of the Student's t-test. There are criteria that do not require independent measurements, but they have less power. For additive metrics, a technique is proposed in this paper, when a test sample is divided into N parts, on each of which the values of the metrics are calculated. Since the value on each part is obtained as the sum of independent random variables, according to the central limit theorem, the obtained metrics values on each of the N parts are realizations of the normally distributed random variable. To estimate the required sample size, it is proposed to use normality tests and build quantile– quantile plots. You can then use a modification of the Student's t-test to conduct a statistical test comparing the mean values of the metrics. A simplified approach is also considered, in which confidence intervals are built for the base model. A model whose metric values do not fall into this interval works differently from the base model. This approach reduces the amount of computations needed, however, an experimental analysis of the binary cross-entropy metric for CTR (Click-Through Rate) prediction models showed that it is more rough than the first one.

Deep Interest Evolution Network for Click-Through Rate Prediction

Click-through rate (CTR) prediction, whose goal is to estimate the probability of a user clicking on the item, has become one of the core tasks in the advertising system. For CTR prediction model, it is necessary to capture the latent user interest behind the user behavior data. Besides, considering the changing of the external environment and the internal cognition, user interest evolves over time dynamically. There are several CTR prediction methods for interest modeling, while most of them regard the representation of behavior as the interest directly, and lack specially modeling for latent interest behind the concrete behavior. Moreover, little work considers the changing trend of the interest. In this paper, we propose a novel model, named Deep Interest Evolution Network (DIEN), for CTR prediction. Specifically, we design interest extractor layer to capture temporal interests from history behavior sequence. At this layer, we introduce an auxiliary loss to supervise interest extracting at each step. As user interests are diverse, especially in the e-commerce system, we propose interest evolving layer to capture interest evolving process that is relative to the target item. At interest evolving layer, attention mechanism is embedded into the sequential structure novelly, and the effects of relative interests are strengthened during interest evolution. In the experiments on both public and industrial datasets, DIEN significantly outperforms the state-of-the-art solutions. Notably, DIEN has been deployed in the display advertisement system of Taobao, and obtained 20.7% improvement on CTR.