Filters








388 Hits in 4.9 sec

Affordable Uplift: Supervised Randomization in Controlled Experiments [article]

Johannes Haupt, Daniel Jacob, Robin M. Gubela, Stefan Lessmann
2019 arXiv   pre-print
Training and monitoring of uplift models require experimental data.  ...  To increase the cost-efficiency of experimentation and facilitate frequent data collection and model training, we introduce supervised randomization.  ...  We will discuss IPW as a method that is easily integrated into model building and evaluation and discuss the doubly robust estimator as a recent extension.  ... 
arXiv:1910.00393v1 fatcat:je2dkoj3mje3fb6u4a6tnsdvl4

Stabilized Doubly Robust Learning for Recommendation on Data Missing Not at Random [article]

Haoxuan Li, Chunyuan Zheng, Xiao-Hua Zhou, Peng Wu
2022 arXiv   pre-print
In recommender systems, users always choose the favorite items to rate, which leads to data missing not at random and poses a great challenge for unbiased evaluation and learning of prediction models.  ...  Currently, the doubly robust (DR) method and its variants have been widely studied and demonstrate superior performance.  ...  Doubly robust estimation in missing data and causal inference models.  ... 
arXiv:2205.04701v2 fatcat:r2vr7rg5ofcqvkdv3z67m7f2wy

Doubly Robust Collaborative Targeted Learning for Debiased Recommendations [article]

Peng Wu, Haoxuan Li, Yan Lyu, Chunyuan Zheng, Xiao-Hua Zhou
2022 arXiv   pre-print
To address selection bias and confounding bias, the doubly robust (DR) method and its variants show superior performance due to the double robustness property and smaller bias under inaccurate propensity  ...  In recommender systems, the collected data always contains various biases and leads to the challenge of accurate predictions.  ...  rate prediction [7, 45] , and uplift modeling [23, 25, 26] .  ... 
arXiv:2203.10258v2 fatcat:v2w5fch7lbdtpozuzmnrxtckdq

A general framework for causal classification [article]

Jiuyong Li, Weijia Zhang, Lin Liu, Kui Yu, Thuc Duy Le, Jixue Liu
2020 arXiv   pre-print
Experiments have shown two instantiations of the framework work for causal classification and for uplift (causal heterogeneity) modelling, and are competitive with the other uplift (causal heterogeneity  ...  We discuss the conditions when causal classification can be resolved by uplift (and causal heterogeneity) modelling methods.  ...  Theorem 2, improves other uplift (causal heterogeneity) modelling methods.  ... 
arXiv:2003.11940v3 fatcat:ftcaz7hhezeohency6ahdmirla

On the Opportunity of Causal Learning in Recommendation Systems: Foundation, Estimation, Prediction and Challenges [article]

Peng Wu, Haoxuan Li, Yuhao Deng, Wenjie Hu, Quanyu Dai, Zhenhua Dong, Jie Sun, Rui Zhang, Xiao-Hua Zhou
2022 arXiv   pre-print
Finally, we formalize many debiasing and prediction tasks in RS, and summarize the statistical and machine learning-based causal estimation methods, expecting to provide new research opportunities and  ...  Many causal-based prediction and debiasing studies rarely discuss the causal interpretation of various biases and the rationality of the corresponding causal assumptions.  ...  [Wang et al., 2019a] propose the doubly robust (DR) method and the joint learning optimization technique.  ... 
arXiv:2201.06716v3 fatcat:2mn5piulfraqbcfvn2d4rpgzby

Evaluation Methods and Measures for Causal Learning Algorithms [article]

Lu Cheng, Ruocheng Guo, Raha Moraffah, Paras Sheth, K. Selcuk Candan, Huan Liu
2022 arXiv   pre-print
The survey seeks to bring to the forefront the urgency of developing publicly available benchmarks and consensus-building standards for causal learning evaluation with observational data.  ...  ., based on statistical methods) to causal learning with big data (i.e., the intersection of causal inference and machine learning), in this survey, we review commonly-used datasets, evaluation methods  ...  The views, opinions, and/or findings expressed are those of the authors and should not be interpreted as representing the official views or policies of the Army Research Office or the U.S. Government.  ... 
arXiv:2202.02896v1 fatcat:ykvg7gfwxfawjgkenvmmkbzpxa

Offline A/B Testing for Recommender Systems

Alexandre Gilotte, Clément Calauzènes, Thomas Nedelec, Alexandre Abraham, Simon Dollé
2018 Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining - WSDM '18  
We focus on evaluation methods that compute an estimator of the potential uplift in revenue that could generate this new technology.  ...  Before A/B testing online a new version of a recommender system, it is usual to perform some offline evaluations on historical data.  ...  Doubly robust estimator. The easiest case is when we dispose of external knowledge like a reward model. We can use this as a control variate to improve our current estimator [6] .  ... 
doi:10.1145/3159652.3159687 dblp:conf/wsdm/GilotteCNAD18 fatcat:xynkxdlocbf5lmns56dmkksumm

CATE meets ML

Daniel Jacob
2021 Digital Finance  
As it turns out, machine learning methods are the tool for generalized prediction models.  ...  The presented toolbox of methods contains meta-learners, like the doubly-robust, R-, T- and X-learner, and methods that are specially designed to estimate the CATE like the causal BART and the generalized  ...  Acknowledgements Financial support of the European Union's Horizon 2020 research and innovation  ... 
doi:10.1007/s42521-021-00033-7 fatcat:koynjpnt5bhvdd3ohnlwn6uvbu

Religiosity and parental educational aspirations for children in Kenya

Martin Paul Jr. Tabe-Ojong, Emmanuel Nshakira-Rukundo
2021 World Development Perspectives  
) and elicit parental aspirations for children using vignettes.  ...  By employing inverse probability weighting with regression adjustment and multivalued treatment effects estimators on cross-sectional data, we show that membership in a religious institution and high levels  ...  In Section 4, the data and measurement of variables and the empirical methods in this paper.  ... 
doi:10.1016/j.wdp.2021.100349 fatcat:5wzchliqkndqbfuvrkdtw5n5be

Page-level Optimization of e-Commerce Item Recommendations [article]

Chieh Lo, Hongliang Yu, Xin Yin, Krutika Shetty, Changchen He, Kathy Hu, Justin Platz, Adam Ilardi, Sriganesh Madhvanath
2021 arXiv   pre-print
Item recommendation modules on the IDP are often curated and statically configured for all customers, ignoring opportunities for personalization.  ...  In our online A/B test, our framework improved click-through rate by 2.48% and purchase-through rate by 7.34% over a static configuration.  ...  To overcome this issue, we adopt both Doubly Robust (DR) Estimator and Direct Method (DM) which evaluate both metrics in an unbiased manner [14] .  ... 
arXiv:2108.05891v1 fatcat:pthcwahr5vfsbflvanmtwy6rxy

Individual Treatment Prescription Effect Estimation in a Low Compliance Setting

Thibaud Rahier, Amélie Héliou, Matthieu Martin, Christophe Renaudin, Eustache Diemert
2021 Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining  
of competition and ad blockers for instance).  ...  We propose a new approach for the estimation of the IPE that takes advantage of observed compliance information to prevent signal fading.  ...  In that specific low-dimensional case, there is therefore no need to implement more complex models (Figure 8 ) such as doubly-robust methods or tree/forest-based methods. • Oracle predicts the theoretical  ... 
doi:10.1145/3447548.3467343 fatcat:yg33lsoy75h3de3yobmzjbeg6q

CATE meets ML – The Conditional Average Treatment Effect and Machine Learning [article]

Daniel Jacob
2021 arXiv   pre-print
As it turns out, machine learning methods are the tool for generalized prediction models.  ...  The presented toolbox of methods contains meta-learners, like the Doubly-Robust, R-, T- and X-learner, and methods that are specially designed to estimate the CATE like the causal BART and the generalized  ...  Including more ML methods could improve the prediction accuracy depending on the data generating process. Using two-step sample splitting with cross-fitting further improves the prediction.  ... 
arXiv:2104.09935v2 fatcat:452b7na2hjdazmth2pzskeuhui

Individual Treatment Prescription Effect Estimation in a Low Compliance Setting [article]

Thibaud Rahier, Amélie Héliou, Matthieu Martin, Christophe Renaudin, Eustache Diemert
2020 arXiv   pre-print
of competition and ad blockers for instance).  ...  We propose a new approach for the estimation of the IPE that takes advantage of observed compliance information to prevent signal fading.  ...  In that specific low-dimensional case, there is therefore no need to implement more complex models (Figure 8 ) such as doubly-robust methods or tree/forest-based methods. • Oracle predicts the theoretical  ... 
arXiv:2008.03235v2 fatcat:sdtfacpxuzexvdk3izvk3ouk6q

Evaluating Treatment Prioritization Rules via Rank-Weighted Average Treatment Effects [article]

Steve Yadlowsky, Scott Fleming, Nigam Shah, Emma Brunskill, Stefan Wager
2021 arXiv   pre-print
Our definition of the RATE nests a number of existing metrics, including the Qini coefficient, and our analysis directly yields inference methods for these metrics.  ...  There are a number of available methods that can be used for choosing whom to prioritize treatment, including ones based on treatment effect estimation, risk scoring, and hand-crafted rules.  ...  The purpose of the data is to provide a benchmark for uplift modeling, and therefore, the results are not meant to be used in a particular application.  ... 
arXiv:2111.07966v1 fatcat:jtzfcqbpezgnvbv2egrlkfpov4

Reinforcement Learning in Practice: Opportunities and Challenges [article]

Yuxi Li
2022 arXiv   pre-print
Then we discuss challenges, in particular, 1) foundation, 2) representation, 3) reward, 4) exploration, 5) model, simulation, planning, and benchmarks, 6) off-policy/offline learning, 7) learning to learn  ...  , computer systems, and science and engineering.  ...  A predictive model is built on domain knowledge, real-world data, and high-fidelity simulators; a robust method accounts for worst-case scenarios and takes conservative actions, and an adaptive method  ... 
arXiv:2202.11296v2 fatcat:xdtsmme22rfpfn6rgfotcspnhy
« Previous Showing results 1 — 15 out of 388 results