Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

169,291 papers · 148 categories

Trend · papers per month

3517021,0521,403 · Jun 202019922001200920182026
48 results for CTR modeling

Paper improves Native ads CTR prediction using event embeddings and recurrent networks.

problem Hard CTR prediction for Native ads due to lack of direct query intent.
method Proposes a large-scale event embedding scheme and a recurrent neural network model.
result Significantly outperforms baseline and variants in CTR prediction.

Proposes DTS framework to predict CTR by tracking user interest evolution over time.

problem Predicting CTR by ignoring dynamic user interest changes over time.
method Integrates time information using ODEs in a neural network to model interest evolution.
result Achieves superior CTR prediction performance compared to existing methods.

FwFMs improve CTR prediction with fewer parameters.

problem Predicting click-through rate in display advertising with multi-field categorical data.
method Field-weighted Factorization Machines (FwFMs) to model feature interactions efficiently.
result FwFMs achieve competitive performance with significantly fewer parameters than FFMs.

DIEN predicts CTR by evolving user interests from behavior data.

problem Capturing dynamic user interests for accurate CTR prediction.
method DIEN captures temporal and evolving user interests using interest extractor and evolving layers with attention mechanisms.
result DIEN significantly outperforms state-of-the-art solutions in CTR prediction.

Meta-Embedding improves CTR predictions for new ads, addressing cold-start and warm-up issues.

problem Improving CTR predictions for new ads with little logging data.
method Meta-learning approach to generate initial embeddings for new ad IDs.
result Meta-Embedding significantly improves CTR predictions for various models, including lightweight and deep learning.

A new embedding structure, res-embedding, improves deep CTR models by enhancing generalization performance.

problem Deep CTR models often suffer from poor generalization performance due to learning of embedding parameters.
method Developed a res-embedding structure that combines a central embedding vector from an item-based interest graph with a residual embedding vector.
result Empirical evaluation shows significant improvement in model performance using the res-embedding structure.

DeepLight accelerates CTR predictions in ad serving by 46X.

problem Significantly increased serving delay and high memory usage for ad serving.
method Explicitly searching feature interactions, pruning layers, promoting sparsity.
result Accelerates model inference by 46X on Criteo dataset.

Paper visualizes and analyzes deep neural networks for CTR prediction.

problem Interpreting deep neural networks for CTR prediction is challenging.
method Visualizing and analyzing deep neural networks at neuron level, measuring layer-wise performance, and calculating saliency scores.
result Insights into the model's performance and mechanism.

Proposes FEFM and DeepFEFM for CTR prediction, outperforming state-of-the-art models.

problem Click-through rate prediction in online applications.
method Field-Embedded Factorization Machine (FEFM) and its deep counterpart DeepFEFM, combining feature embeddings and deep neural networks.
result FEFM and DeepFEFM outperform existing models in CTR prediction tasks.

CAN approximates explicit feature interactions for CTR prediction.

problem Learning explicit feature interactions from sparse features.
method Co-Action Network approximates explicit pairwise feature interactions without introducing too many additional parameters.
result CAN outperforms state-of-the-art CTR models and the cartesian product method.

DeepFM combines deep learning and factorization machines for CTR prediction.

problem Maximizing CTR in recommender systems through sophisticated feature interactions.
method End-to-end learning framework combining DNN and PNN with shared feature input.
result DeepFM-D outperforms existing models, achieving over 10% improvement in CTR.

This paper tackles overfitting in CTR models by introducing Multi-Epoch learning with Data Augmentation.

problem Overfitting of the embedding layer in CTR models during multi-epoch training.
method Introduces Multi-Epoch learning with Data Augmentation (MEDA) framework to reduce overfitting and enhance performance.
result MEDA minimizes overfitting and achieves data augmentation through varied embedding spaces, improving performance without overfitting.

AutoFIS automatically selects important feature interactions for CTR prediction models.

problem Manual feature interaction design is inefficient and prone to noise.
method Two-stage algorithm: search stage relaxes feature interactions to continuous parameters, re-train stage refines model performance.
result AutoFIS significantly improves CTR and CVR of FM-based models.

LS-PLM predicts ad clicks from large scale sparse data.

problem Predicting click-through rates (CTR) from large scale nonlinear sparse data.
method Formulated with L1L_1 and L2,1L_{2,1} regularizers, solved with directional derivatives and quasi-Newton method, and distributed on hundreds of machines.
result Captures nonlinear patterns from massive sparse data, reducing feature engineering.

FiBiNET combines feature importance and bilinear interactions for CTR prediction.

problem Improving click-through rate prediction in advertising and feed ranking systems.
method FiBiNET dynamically learns feature importance via SENET and bilinear feature interactions.
result FiBiNET outperforms shallow and deep models on real-world datasets.

This paper tackles delayed feedback in continuous training for CTR prediction, improving model performance by 3%.

problem Delayed feedback in CTR prediction leads to inferior performance and user experience.
method Comparing 5 loss functions and models in offline and online settings.
result Proposed methods outperform previous state-of-the-art by 3% relative cross entropy (RCE).

Proposes a novel network for CTR prediction by learning modality-specific and modality-invariant representations.

problem Learning good representation of items from multimodal features in E-commerce is challenging due to redundant information across modalities.
method Introduces a Multimodal Adversarial Representation Network (MARN) that calculates modality-specific weights and learns modality-invariant representations.
result Consistently achieves remarkable improvements over state-of-the-art methods in CTR prediction.

KFAtt improves CTR prediction by modeling user behavior with Kalman filtering attention.

problem Improving CTR prediction in personalized e-commerce search engines.
method KFAtt combines Kalman filtering with attention mechanisms to model user behavior.
result KFAtt outperforms existing methods in CTR prediction, achieving better performance in both offline and online settings.

LorentzFM learns feature interactions in a hyperbolic space, reducing parameters and achieving state-of-the-art results.

problem Learning feature interactions for recommendation and CTR predictions using deep learning methods is computationally and memory inefficient.
method Proposes LorentzFM, a model that learns feature interactions in a hyperbolic space, reducing the number of parameters and achieving comparable results.
result LorentzFM achieves comparable and sometimes better results than deep learning methods in recommendation and CTR prediction tasks.

This paper investigates the impact of normalization on deep neural networks for click-through rate prediction.

problem The effect of normalization on deep neural network models for CTR estimation.
method Systematic study of various normalization approaches applied to feature embedding and MLP part of DNN models.
result Correct normalization significantly enhances model performance, as demonstrated by extensive experiments on real-world datasets.

Method detects interactions for better CTR prediction.

problem Predicting click-through rate with high-dimensional categorical features and time-varying interactions.
method Online Random Intersection Chains (ORIC) for detecting informative interactions.
result ORIC detects high-interpretability interactions that improve CTR prediction.

AdaEnsemble learns adaptive feature interactions for CTR prediction.

problem Learning feature interactions for CTR prediction in recommender systems and Ads ranking.
method AdaEnsemble is a Sparsely-Gated Mixture-of-Experts (SparseMoE) architecture that dynamically selects feature interaction depth.
result AdaEnsemble achieves better prediction accuracy and inference efficiency compared to state-of-the-art models.

The study creates user personas based on user tenure and behavior for VoD streaming.

problem Understanding evolving user behavior in streaming services without explicit user profiles.
method Construct user personas using tenure timelines and temporal behavioral features.
result Personas provide stable and interpretable insights into user behavior evolution.

A new method trains lightweight neural networks using a more complex 'booster' network.

problem Training high-performing neural networks for real-time tasks like CTR prediction.
method Rocket launching framework: a cumbersome net guides the training of a lightweight net.
result Lightweight models achieve performance previously only possible with complex models.

Collaborative filtering (CF) is a successful approach commonly used by many recommender systems. Conventional CF-based methods use the ratings given to items by users as the sole source of information for learning to make recommendation. However, the ratings are often very sparse in many applications, causing CF-based …

2014-09-10abs ↗pdf ↗

Develops a deep learning framework to predict future tech directions for high-tech companies.

problem Difficult task in predicting future R&D trends for high-tech companies due to complexity and variety of factors.
method Deep Technology Forecasting (DTF) framework with three components: PCR, CTR, and DTT neural network.
result DTF framework precisely predicts future tech emphasis of companies using hybrid factors.

xDeepInt learns both vector-wise and bit-wise feature interactions.

problem Learning feature interactions for CTR prediction and recommendation.
method Polynomial Interaction Network (PIN) architecture with subspace-crossing mechanism.
result xDeepInt outperforms state-of-the-art models in CTR prediction and recommendation.

DHEN improves CVR prediction for ads with multitask learning and auxiliary loss.

problem Predicting conversion rates in ad-recommendation systems.
method DHEN integrates multiple feature-crossing modules and uses a multitask learning framework, ablation studies, and self-supervised auxiliary loss.
result DHEN achieves state-of-the-art performance in CVR prediction.

Adaptive regularization prevents overfitting in large-scale sparse feature models.

problem Overfitting in models with large-scale sparse categorical features.
method Adaptive regularization of embedding layers' norm budget.
result Improves model performance within a single epoch and prevents multi-epoch performance degradation.

Proposes DHEB model for predicting online ad performance.

problem Sparse data at individual unit level in online advertising.
method Dynamic Hierarchical Empirical Bayesian (DHEB) model with data-driven hierarchy and shrinkage-based estimations.
result Proposed method outperforms other models in accuracy and efficiency.

DES training speeds up large-scale recommender systems convergence.

problem Training large-scale recommender systems with dynamic sparse features.
method Distributed Equivalent Substitution (DES) framework for fully synchronous training.
result DES achieves higher AUC and up to 68.7% communication savings.

Tensor factorization is a powerful tool to analyse multi-way data. Compared with traditional multi-linear methods, nonlinear tensor factorization models are capable of capturing more complex relationships in the data. However, they are computationally expensive and may suffer severe learning bias in case of extreme dat…

2016-04-27abs ↗pdf ↗

This paper improves online ad revenue by optimizing auction performance directly.

problem Disconnection between ad ranking and auction performance in online advertising.
method Proposes new loss functions and ranking functions to maximize revenue.
result Proposed methods outperform state-of-the-art in maximizing platform revenue.