Xuan Bi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Xuan Bi has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Mochen Yang
Carlson School of Management, University of Minnesota, Minneapolis, Minnesota 55455
3 shared publications - Annie Qu
Department of Statistics and Applied Probability, University of California, Santa Barbara, CA, USA
2 shared publications - Gediminas Adomavicius
Carlson School of Management, University of Minnesota, Minneapolis, Minnesota 55455
2 shared publications - Xiaotong Shen
School of Statistics, University of Minnesota, Minneapolis, USA
2 shared publications - Alok Gupta
Carlson School of Management, University of Minnesota, Minneapolis, Minnesota 55455
1 shared publication Annie Qu
Department of Statistics, University of California, Irvine, Irvine, California 92697
1 shared publication
Latest Publications
- Open Access
Recommending Composite Items Using Multi-Level Preference Information: A Joint Interaction Modeling Approach
With the advancement of machine learning and artificial intelligence technologies, recommender systems have been increasingly used across a vast variety of platforms to efficiently and effectively match users with items. As application contexts become more diverse and complex, there is a growing need for more sophisticated recommendation techniques. One example is the composite item (for example, fashion outfit) recommendation where multiple levels of user preference information might be available and relevant. In this study, we propose JIMA, a joint interaction modeling approach that uses a single model to take advantage of all data from different levels of granularity and incorporate interactions to learn the complex relationships among lower-order (atomic item) and higher-order (composite item) user preferences as well as domain expertise (e.g., on the stylistic fit). We comprehensively evaluate the proposed method and compare it with advanced baselines through multiple simulation studies as well as with real data in both offline and online settings. The results consistently demonstrate the superior performance of the proposed approach.
Open AccessRecommending Composite Items Using Multi-Level Preference Information: A Joint Interaction Modeling Approach
With the advancement of machine learning and artificial intelligence technologies, recommender systems have been increasingly used across a vast variety of platforms to efficiently and effectively match users with items. As application contexts become more diverse and complex, there is a growing need for more sophisticated recommendation techniques. One example is the composite item (for example, fashion outfit) recommendation where multiple levels of user preference information might be available and relevant. In this study, we propose JIMA, a joint interaction modeling approach that uses a single model to take advantage of all data from different levels of granularity and incorporate interactions to learn the complex relationships among lower-order (atomic item) and higher-order (composite item) user preferences as well as domain expertise (e.g., on the stylistic fit). We comprehensively evaluate the proposed method and compare it with advanced baselines through multiple simulation studies as well as with real data in both offline and online settings. The results consistently demonstrate the superior performance of the proposed approach.
- Open Access
- 25 August 2025
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 27 (1) , e76941
Physician Use of Large Language Models: A Quantitative Study Based on Large-Scale Query-Level Data
Background: Generative artificial intelligence (GenAI) has rapidly emerged as a promising tool in health care. Despite its growing adoption, how physicians make use of it in medical practice has not been qualitatively studied. Existing literature has largely focused on theoretical applications or experimental validations, with limited insight into real-world physician engagement with GenAI technologies. Objective: The aim of this study was to leverage a fine-grained dataset at the query level to quantitatively examine how physicians incorporate GenAI into their clinical and research workflows. The primary objective was to analyze usage patterns over time and across physician demographics. A secondary goal was to assess potential risks to patient privacy arising from physicians’ interactions with GenAI platforms. Methods: This study collected 106,942 query-and-answer pairs by 989 physicians between August 29, 2023, and April 16, 2024. We performed topic classification to identify the most prevalent use cases, examining how these use cases evolved over time and across demographics. We also developed sensitivity classifiers to detect personally identifiable information in physicians’ queries to explore the potential privacy breach risks around physicians’ use of GenAI. Results: Approximately 40% (396/989) of the enrolled physicians were female, 45.9% (454/989) were younger than 25 years, and 54.1% (535/989) were between 25 and 56 years of age. The majority of them worked in clinical departments (680/989, 68.8%) or medical technology departments (127/989, 12.8%). Our classification-based quantitative analyses suggest the following. First, physicians use GenAI predominantly for medical research (64,379/106,942, 60.2%) rather than clinical practice (13,100/106,942, 12.25%). Second, physicians focus more on health care–related questions (rising from 64,165/106,942, 60% to 83,415/106,942, 78%) within the first 15% (16,041/106,942) of their query sequence. Third, the use of GenAI differed across physician demographics and features. Specifically, female physicians asked a larger proportion of clinical questions (female: 0.154 vs male: 0.108; P<.001) and administration questions (female: 0.027 vs male: 0.018; P<.001) than male physicians; younger physicians posed more clinical questions (age ≤25: 0.146 vs age ∈ (25, 40]: 0.115 vs age >40: 0.103; P<.001) but fewer research questions (age ≤25: 0.580 vs age ∈ (25, 40]: 0.607 vs age >40: 0.664; P<.001) than senior physicians; and physicians accessing GenAI via computers asked more research questions (computer: 0.637 vs mobile: 0.296; P<.001), whereas physicians using mobile devices asked more clinical questions (computer: 0.107 vs mobile: 0.264; P<.001). Fourth, only 2.68% (2866/106,942) of physician queries contained sensitive information, the majority of which were primarily derived from writing and editing. Conclusions: Physicians are actively integrating GenAI into their professional routines, primarily leveraging it for research but also increasingly for clinical support. Usage patterns vary significantly across demographic lines, including gender, age, and device preference. Despite the presence of sensitive information in some queries, the risk of privacy breaches appears to be low.
- 25 August 2025
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 27 (1) , e76941
Open AccessPhysician Use of Large Language Models: A Quantitative Study Based on Large-Scale Query-Level Data
Background: Generative artificial intelligence (GenAI) has rapidly emerged as a promising tool in health care. Despite its growing adoption, how physicians make use of it in medical practice has not been qualitatively studied. Existing literature has largely focused on theoretical applications or experimental validations, with limited insight into real-world physician engagement with GenAI technologies. Objective: The aim of this study was to leverage a fine-grained dataset at the query level to quantitatively examine how physicians incorporate GenAI into their clinical and research workflows. The primary objective was to analyze usage patterns over time and across physician demographics. A secondary goal was to assess potential risks to patient privacy arising from physicians’ interactions with GenAI platforms. Methods: This study collected 106,942 query-and-answer pairs by 989 physicians between August 29, 2023, and April 16, 2024. We performed topic classification to identify the most prevalent use cases, examining how these use cases evolved over time and across demographics. We also developed sensitivity classifiers to detect personally identifiable information in physicians’ queries to explore the potential privacy breach risks around physicians’ use of GenAI. Results: Approximately 40% (396/989) of the enrolled physicians were female, 45.9% (454/989) were younger than 25 years, and 54.1% (535/989) were between 25 and 56 years of age. The majority of them worked in clinical departments (680/989, 68.8%) or medical technology departments (127/989, 12.8%). Our classification-based quantitative analyses suggest the following. First, physicians use GenAI predominantly for medical research (64,379/106,942, 60.2%) rather than clinical practice (13,100/106,942, 12.25%). Second, physicians focus more on health care–related questions (rising from 64,165/106,942, 60% to 83,415/106,942, 78%) within the first 15% (16,041/106,942) of their query sequence. Third, the use of GenAI differed across physician demographics and features. Specifically, female physicians asked a larger proportion of clinical questions (female: 0.154 vs male: 0.108; P<.001) and administration questions (female: 0.027 vs male: 0.018; P<.001) than male physicians; younger physicians posed more clinical questions (age ≤25: 0.146 vs age ∈ (25, 40]: 0.115 vs age >40: 0.103; P<.001) but fewer research questions (age ≤25: 0.580 vs age ∈ (25, 40]: 0.607 vs age >40: 0.664; P<.001) than senior physicians; and physicians accessing GenAI via computers asked more research questions (computer: 0.637 vs mobile: 0.296; P<.001), whereas physicians using mobile devices asked more clinical questions (computer: 0.107 vs mobile: 0.264; P<.001). Fourth, only 2.68% (2866/106,942) of physician queries contained sensitive information, the majority of which were primarily derived from writing and editing. Conclusions: Physicians are actively integrating GenAI into their professional routines, primarily leveraging it for research but also increasingly for clinical support. Usage patterns vary significantly across demographic lines, including gender, age, and device preference. Despite the presence of sensitive information in some queries, the risk of privacy breaches appears to be low.
- 1 April 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Data Science
- Vol. 4 (2) , 101-113
Cost-Aware Calibration of Classifiers
Most classification techniques in machine learning are able to produce probability predictions in addition to class predictions. However, these predicted probabilities are often not well calibrated in that they deviate from the actual outcome rates (i.e., the proportion of data instances that actually belong to a certain class). A lack of calibration can jeopardize downstream decision tasks that rely on accurate probability predictions. Although several post hoc calibration methods have been proposed, they generally do not consider the potentially asymmetric costs associated with overprediction versus underprediction. In this research, we formally define the problem of cost-aware calibration and propose a metric to quantify the cost of miscalibration for a given classifier. Next, we propose three approaches to achieve cost-aware calibration, two of which are cost-aware adaptations of existing calibration algorithms; the third one (named MetaCal) is a Bayes optimal learning algorithm inspired by prior work on cost-aware classification. We carry out systematic empirical evaluations on multiple public data sets to demonstrate the effectiveness of the proposed approaches in reducing the cost of miscalibration. Finally, we generalize the definition and metric as well as solution algorithms of cost-aware calibration to account for nonlinear cost structures that may arise in real-world decision tasks. History: David Martens served as the senior editor for this article. Data Ethics & Reproducibility Note: There are no data ethics considerations. The code capsule is available on Code Ocean at https://doi.org/10.24433/CO.8552538.v1 and in the e-Companion to this article (available at https://doi.org/10.1287/ijds.2024.0038 ).
- 1 April 2025
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Data Science
- Vol. 4 (2) , 101-113
Cost-Aware Calibration of Classifiers
Most classification techniques in machine learning are able to produce probability predictions in addition to class predictions. However, these predicted probabilities are often not well calibrated in that they deviate from the actual outcome rates (i.e., the proportion of data instances that actually belong to a certain class). A lack of calibration can jeopardize downstream decision tasks that rely on accurate probability predictions. Although several post hoc calibration methods have been proposed, they generally do not consider the potentially asymmetric costs associated with overprediction versus underprediction. In this research, we formally define the problem of cost-aware calibration and propose a metric to quantify the cost of miscalibration for a given classifier. Next, we propose three approaches to achieve cost-aware calibration, two of which are cost-aware adaptations of existing calibration algorithms; the third one (named MetaCal) is a Bayes optimal learning algorithm inspired by prior work on cost-aware classification. We carry out systematic empirical evaluations on multiple public data sets to demonstrate the effectiveness of the proposed approaches in reducing the cost of miscalibration. Finally, we generalize the definition and metric as well as solution algorithms of cost-aware calibration to account for nonlinear cost structures that may arise in real-world decision tasks. History: David Martens served as the senior editor for this article. Data Ethics & Reproducibility Note: There are no data ethics considerations. The code capsule is available on Code Ocean at https://doi.org/10.24433/CO.8552538.v1 and in the e-Companion to this article (available at https://doi.org/10.1287/ijds.2024.0038 ).
- 1 December 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 35 (4) , 1890-1908
Background Music Recommendation on Short Video Sharing Platforms
On short video sharing platforms, users often choose background music for their videos. In this paper, we study the problem of background music recommendation for short videos on short video sharing platforms. In our recommendation setting, the item (music) is not recommended directly to the user, but to the video created by the user. When making music recommendations for videos, we consider three important players: users, videos, and music. We define a unique background music recommendation problem and design a novel background music recommendation model to address the problem. We propose a model based on the deep learning framework to effectively address the distinctive three-way relationships among users, videos, and music. Our model considers not only of the conventional user–music alignment, but also the alignment between videos and music. To evaluate our model, we conduct comprehensive experiments on real-world data collected from one of the most popular short video sharing platforms. Our proposed model significantly outperforms other existing models in recommendation performance. The superiority of our proposed model remains consistent across various scenarios, including cold-start recommendations, data sets with varying density levels, and data sets spanning diverse video categories.
- 1 December 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 35 (4) , 1890-1908
Background Music Recommendation on Short Video Sharing Platforms
On short video sharing platforms, users often choose background music for their videos. In this paper, we study the problem of background music recommendation for short videos on short video sharing platforms. In our recommendation setting, the item (music) is not recommended directly to the user, but to the video created by the user. When making music recommendations for videos, we consider three important players: users, videos, and music. We define a unique background music recommendation problem and design a novel background music recommendation model to address the problem. We propose a model based on the deep learning framework to effectively address the distinctive three-way relationships among users, videos, and music. Our model considers not only of the conventional user–music alignment, but also the alignment between videos and music. To evaluate our model, we conduct comprehensive experiments on real-world data collected from one of the most popular short video sharing platforms. Our proposed model significantly outperforms other existing models in recommendation performance. The superiority of our proposed model remains consistent across various scenarios, including cold-start recommendations, data sets with varying density levels, and data sets spanning diverse video categories.
- 1 August 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Management Science
- Vol. 70 (8) , 4974-4994
Understanding Partnership Formation and Repeated Contributions in Federated Learning: An Analytical Investigation
Limited access to large-scale data is a key obstacle to building machine learning (ML) applications in practice, partly due to a reluctance of information exchange among data owners out of privacy and data security concerns. To address this “information silo” problem, federated learning (FL) techniques have been proposed to enable decentralized model training via an orchestrating central server and have received increasing attention in several industries (including healthcare and finance). Despite its superior privacy protection property, adoption of FL is limited by a lack of systematic understanding of its underlying economics. In this paper, we take an analytical approach to answer two questions: (1) when do data owners prefer to form a FL partnership over building ML models by themselves and (2) how can different contractual mechanisms be used to promote repeated contributions to FL (the cooperative outcome that benefits all participants). We formulate an iterated prisoner’s dilemma (IPD) model that accounts for unique FL characteristics, including the specification of the payoff matrix and the involvement of a central server to sanction noncooperation. We find that partnership formation requires participants to be not too forward-looking in temporal preferences, which is contrary to the conventional wisdom in IPD. Furthermore, to promote repeated contributions, it is insufficient to only rely on penalties imposed by the central server or by participants for noncooperation, but a combination of both is enough. Our work advances theoretical understanding of the economics of FL and provides prescriptive insights that can inform FL participant selection and contract design. This paper was accepted by D. J. Wu, information systems. Funding: This work was partially supported by Cisco Research. Supplemental Material: The online appendices are available at https://doi.org/10.1287/mnsc.2023.00611 .
- 1 August 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Management Science
- Vol. 70 (8) , 4974-4994
Understanding Partnership Formation and Repeated Contributions in Federated Learning: An Analytical Investigation
Limited access to large-scale data is a key obstacle to building machine learning (ML) applications in practice, partly due to a reluctance of information exchange among data owners out of privacy and data security concerns. To address this “information silo” problem, federated learning (FL) techniques have been proposed to enable decentralized model training via an orchestrating central server and have received increasing attention in several industries (including healthcare and finance). Despite its superior privacy protection property, adoption of FL is limited by a lack of systematic understanding of its underlying economics. In this paper, we take an analytical approach to answer two questions: (1) when do data owners prefer to form a FL partnership over building ML models by themselves and (2) how can different contractual mechanisms be used to promote repeated contributions to FL (the cooperative outcome that benefits all participants). We formulate an iterated prisoner’s dilemma (IPD) model that accounts for unique FL characteristics, including the specification of the payoff matrix and the involvement of a central server to sanction noncooperation. We find that partnership formation requires participants to be not too forward-looking in temporal preferences, which is contrary to the conventional wisdom in IPD. Furthermore, to promote repeated contributions, it is insufficient to only rely on penalties imposed by the central server or by participants for noncooperation, but a combination of both is enough. Our work advances theoretical understanding of the economics of FL and provides prescriptive insights that can inform FL participant selection and contract design. This paper was accepted by D. J. Wu, information systems. Funding: This work was partially supported by Cisco Research. Supplemental Material: The online appendices are available at https://doi.org/10.1287/mnsc.2023.00611 .
- 1 March 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 35 (1) , 339-362
Consumer Acquisition for Recommender Systems: A Theoretical Framework and Empirical Evaluations
How to acquire the most valuable consumers to grow your recommender system? We propose a dynamic consumer acquisition model to enable value-driven acquisition decisions. We build a model of consumer acquisition that takes into account the value that a consumer contributes to the recommender system, the cost of their participation (e.g., privacy loss), and the value of their participation to other consumers (via network externality). We also propose data-driven procedures to estimate this model to enable informed, value-driven acquisition decisions. On three different data sets, we perform comprehensive simulation-based evaluations to demonstrate the performance of this dynamic consumer acquisition model. We find nuanced relationships between the firm’s choice of incentive strategies and acquisition outcomes. Neither a constant pricing strategy nor a greedy pricing strategy may be optimal. Instead, under a moderately greedy strategy, where the firm only partially extracts the network externality from consumers, the dynamic acquisition sequence can outperform random acquisition sequences on firm utility, recommender system performance, and consumer surplus simultaneously. Our work contributes a novel theoretical framework, practical insights, and design artifacts to facilitate effective consumer acquisition in recommender systems.
- 1 March 2024
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in Information Systems Research
- Vol. 35 (1) , 339-362
Consumer Acquisition for Recommender Systems: A Theoretical Framework and Empirical Evaluations
How to acquire the most valuable consumers to grow your recommender system? We propose a dynamic consumer acquisition model to enable value-driven acquisition decisions. We build a model of consumer acquisition that takes into account the value that a consumer contributes to the recommender system, the cost of their participation (e.g., privacy loss), and the value of their participation to other consumers (via network externality). We also propose data-driven procedures to estimate this model to enable informed, value-driven acquisition decisions. On three different data sets, we perform comprehensive simulation-based evaluations to demonstrate the performance of this dynamic consumer acquisition model. We find nuanced relationships between the firm’s choice of incentive strategies and acquisition outcomes. Neither a constant pricing strategy nor a greedy pricing strategy may be optimal. Instead, under a moderately greedy strategy, where the firm only partially extracts the network externality from consumers, the dynamic acquisition sequence can outperform random acquisition sequences on firm utility, recommender system performance, and consumer surplus simultaneously. Our work contributes a novel theoretical framework, practical insights, and design artifacts to facilitate effective consumer acquisition in recommender systems.
- Open Access
RAW: A Robust and Agile Plug-and-Play Watermark Framework for AI-Generated Images with Provable Guarantees
Safeguarding intellectual property and preventing potential misuse of AI-generated images are of paramount importance. This paper introduces a robust and agile plug-and-play watermark detection framework, dubbed as RAW. As a departure from traditional encoder-decoder methods, which incorporate fixed binary codes as watermarks within latent representations, our approach introduces learnable watermarks directly into the original image data. Subsequently, we employ a classifier that is jointly trained with the watermark to detect the presence of the watermark. The proposed framework is compatible with various generative architectures and supports on-the-fly watermark injection after training. By incorporating state-of-the-art smoothing techniques, we show that the framework provides provable guarantees regarding the false positive rate for misclassifying a watermarked image, even in the presence of certain adversarial attacks targeting watermark removal. Experiments on a diverse range of images generated by state-of-the-art diffusion models reveal substantial performance enhancements compared to existing approaches. For instance, our method demonstrates a notable increase in AUROC, from 0.48 to 0.82, when compared to state-of-the-art approaches in detecting watermarked images under adversarial attacks, while maintaining image quality, as indicated by closely aligned FID and CLIP scores.
Open AccessRAW: A Robust and Agile Plug-and-Play Watermark Framework for AI-Generated Images with Provable Guarantees
Safeguarding intellectual property and preventing potential misuse of AI-generated images are of paramount importance. This paper introduces a robust and agile plug-and-play watermark detection framework, dubbed as RAW. As a departure from traditional encoder-decoder methods, which incorporate fixed binary codes as watermarks within latent representations, our approach introduces learnable watermarks directly into the original image data. Subsequently, we employ a classifier that is jointly trained with the watermark to detect the presence of the watermark. The proposed framework is compatible with various generative architectures and supports on-the-fly watermark injection after training. By incorporating state-of-the-art smoothing techniques, we show that the framework provides provable guarantees regarding the false positive rate for misclassifying a watermarked image, even in the presence of certain adversarial attacks targeting watermark removal. Experiments on a diverse range of images generated by state-of-the-art diffusion models reveal substantial performance enhancements compared to existing approaches. For instance, our method demonstrates a notable increase in AUROC, from 0.48 to 0.82, when compared to state-of-the-art approaches in detecting watermarked images under adversarial attacks, while maintaining image quality, as indicated by closely aligned FID and CLIP scores.
- Open Access
Welfare and Fairness Dynamics in Federated Learning: A Client Selection Perspective
Federated learning (FL) is a privacy-preserving learning technique that enables distributed computing devices to train shared learning models across data silos collaboratively. Existing FL works mostly focus on designing advanced FL algorithms to improve the model performance. However, the economic considerations of the clients, such as fairness and incentive, are yet to be fully explored. Without such considerations, self-motivated clients may lose interest and leave the federation. To address this problem, we designed a novel incentive mechanism that involves a client selection process to remove low-quality clients and a money transfer process to ensure a fair reward distribution. Our experimental results strongly demonstrate that the proposed incentive mechanism can effectively improve the duration and fairness of the federation.
Open AccessWelfare and Fairness Dynamics in Federated Learning: A Client Selection Perspective
Federated learning (FL) is a privacy-preserving learning technique that enables distributed computing devices to train shared learning models across data silos collaboratively. Existing FL works mostly focus on designing advanced FL algorithms to improve the model performance. However, the economic considerations of the clients, such as fairness and incentive, are yet to be fully explored. Without such considerations, self-motivated clients may lose interest and leave the federation. To address this problem, we designed a novel incentive mechanism that involves a client selection process to remove low-quality clients and a money transfer process to ensure a fair reward distribution. Our experimental results strongly demonstrate that the proposed incentive mechanism can effectively improve the duration and fairness of the federation.
- 17 June 2022
- Published by Elsevier in Journal of Econometrics
- Vol. 235 (2) , 444-453
Distribution-invariant differential privacy
- 17 June 2022
- Published by Elsevier in Journal of Econometrics
- Vol. 235 (2) , 444-453
Distribution-invariant differential privacy
- Open Access
Distribution-Invariant Differential Privacy
Differential privacy is becoming one gold standard for protecting the privacy of publicly shared data. It has been widely used in social science, data science, public health, information technology, and the U.S. decennial census. Nevertheless, to guarantee differential privacy, existing methods may unavoidably alter the conclusion of the original data analysis, as privatization often changes the sample distribution. This phenomenon is known as the trade-off between privacy protection and statistical accuracy. In this work, we mitigate this trade-off by developing a distribution-invariant privatization (DIP) method to reconcile both high statistical accuracy and strict differential privacy. As a result, any downstream statistical or machine learning task yields essentially the same conclusion as if one used the original data. Numerically, under the same strictness of privacy protection, DIP achieves superior statistical accuracy in a wide range of simulation studies and real-world benchmarks.
Open AccessDistribution-Invariant Differential Privacy
Differential privacy is becoming one gold standard for protecting the privacy of publicly shared data. It has been widely used in social science, data science, public health, information technology, and the U.S. decennial census. Nevertheless, to guarantee differential privacy, existing methods may unavoidably alter the conclusion of the original data analysis, as privatization often changes the sample distribution. This phenomenon is known as the trade-off between privacy protection and statistical accuracy. In this work, we mitigate this trade-off by developing a distribution-invariant privatization (DIP) method to reconcile both high statistical accuracy and strict differential privacy. As a result, any downstream statistical or machine learning task yields essentially the same conclusion as if one used the original data. Numerically, under the same strictness of privacy protection, DIP achieves superior statistical accuracy in a wide range of simulation studies and real-world benchmarks.