Selected work

Representative Papers

CognitionNet: A Collaborative Neural Network for Play Style Discovery in Online Skill Gaming Platform

Aug 14, 2022Knowledge Discovery and Data Mining

This work addresses the challenge of modeling player behavior in online skill-gaming platforms (e.g., Rummy) by proposing a two-stage collaborative neural network framework that automatically discovers fine-grained behavioral patterns and stable, long-term playing styles from large-scale telemetry sequences. Methodologically, it introduces a novel “bridging loss” training mechanism that jointly optimizes latent-space clustering, sequential modeling, and supervised style classification—enabling end-to-end learning of psychological traits, tactical preferences, and engagement attribution. Key contributions include: (1) the first generalizable and interpretable taxonomy of playing styles grounded in behavioral psychology; (2) significant performance gains over state-of-the-art baselines in engagement prediction; and (3) psychologically informed behavioral diagnostics that support player experience understanding, progression analysis, and risk intervention.

5 citationsRead paper

ScarceGAN: Discriminative Classification Framework for Rare Class Identification for Longitudinal Data with Weak Prior

Oct 26, 2021International Conference on Information and Knowledge Management

This paper addresses the challenge of identifying extremely rare positive-class instances (e.g., novel attacks, high-risk users) in longitudinal telemetry data, where positive samples are severely scarce, negative samples are heterogeneous and weakly labeled across multiple sources, and abundant unlabeled data leads to insufficient prior knowledge. We propose the first weakly supervised, semi-supervised GAN framework jointly modeling sparse positives and multi-source heterogeneous negatives. Our method introduces a tolerance term to relax noisy negative-label constraints, designs a dual-path discriminator and generator loss to unify the modeling of diverse negative distributions and sparse positives, and integrates weakly supervised discriminative learning with generative data augmentation. Evaluated on a skill-game risk-control task, our approach achieves 85% recall for rare classes—60% higher than state-of-the-art baselines. On KDDCUP99, it successfully detects an attack class constituting only 0.09% of the data, establishing a new benchmark for ultra-rare-class detection.

3 citationsRead paper

Explainable and Interpretable Forecasts on Non-Smooth Multivariate Time Series for Responsible Gameplay

Aug 24, 2024Knowledge Discovery and Data Mining

This paper addresses the problem of predicting excessive engagement risk from non-smooth multivariate time series (MTS) in online gaming scenarios. We propose the Actionable Forecasting Network (AFN), the first model designed for actionable forecasting in this setting. Unlike existing approaches that assume temporal smoothness, AFN jointly optimizes high-accuracy forecasting, generation of smooth and interpretable behavioral trajectories, and multi-dimensional feature attribution via an integrated architecture comprising attention mechanisms, a differentiable attribution module, and trajectory regularization. Evaluated on real-world gaming data, AFN reduces mean squared error (MSE) by 25%, doubles the proportion of high-risk players identified four weeks in advance (achieving 23%), and precisely pinpoints critical risk-inducing timesteps to inform timely interventions—resulting in an 18% reduction in subsequent excessive engagement behaviors.

1 citationsRead paper

FAST-Q: Fast-track Exploration with Adversarially Balanced State Representations for Counterfactual Action Estimation in Offline Reinforcement Learning

Apr 30, 2025

In high-risk offline reinforcement learning for online gaming recommendation, counterfactual action estimation fails due to sparse state-space overlap between policies and experimental path bias. To address this, we propose a novel offline RL framework tailored to online game recommendation. Our contributions are threefold: (1) gradient reversal learning to construct adversarially balanced state representations, mitigating distributional shift; (2) a Q-value decomposition-based multi-objective optimization mechanism to jointly enhance interpretability and conservatism in recommendations; and (3) integration of adversarial regularization with improved conservative Q-learning, enabling parallel offline counterfactual exploration and exploitation. Deployed on a real-world volatile gaming platform, our method achieves a 0.15% increase in player return, a 2% uplift in user lifetime value (LTV), 0.4% and 2% gains in recommendation-driven engagement and platform session duration, respectively, and a 10% reduction in recommendation cost.

0 citationsRead paper

Recent publications

Latest Papers

FAST-Q: Fast-track Exploration with Adversarially Balanced State Representations for Counterfactual Action Estimation in Offline Reinforcement Learning

Apr 30, 2025

In high-risk offline reinforcement learning for online gaming recommendation, counterfactual action estimation fails due to sparse state-space overlap between policies and experimental path bias. To address this, we propose a novel offline RL framework tailored to online game recommendation. Our contributions are threefold: (1) gradient reversal learning to construct adversarially balanced state representations, mitigating distributional shift; (2) a Q-value decomposition-based multi-objective optimization mechanism to jointly enhance interpretability and conservatism in recommendations; and (3) integration of adversarial regularization with improved conservative Q-learning, enabling parallel offline counterfactual exploration and exploitation. Deployed on a real-world volatile gaming platform, our method achieves a 0.15% increase in player return, a 2% uplift in user lifetime value (LTV), 0.4% and 2% gains in recommendation-driven engagement and platform session duration, respectively, and a 10% reduction in recommendation cost.

0 citationsRead paper

Explainable and Interpretable Forecasts on Non-Smooth Multivariate Time Series for Responsible Gameplay

Aug 24, 2024Knowledge Discovery and Data Mining

This paper addresses the problem of predicting excessive engagement risk from non-smooth multivariate time series (MTS) in online gaming scenarios. We propose the Actionable Forecasting Network (AFN), the first model designed for actionable forecasting in this setting. Unlike existing approaches that assume temporal smoothness, AFN jointly optimizes high-accuracy forecasting, generation of smooth and interpretable behavioral trajectories, and multi-dimensional feature attribution via an integrated architecture comprising attention mechanisms, a differentiable attribution module, and trajectory regularization. Evaluated on real-world gaming data, AFN reduces mean squared error (MSE) by 25%, doubles the proportion of high-risk players identified four weeks in advance (achieving 23%), and precisely pinpoints critical risk-inducing timesteps to inform timely interventions—resulting in an 18% reduction in subsequent excessive engagement behaviors.

1 citationsRead paper

CognitionNet: A Collaborative Neural Network for Play Style Discovery in Online Skill Gaming Platform

Aug 14, 2022Knowledge Discovery and Data Mining

This work addresses the challenge of modeling player behavior in online skill-gaming platforms (e.g., Rummy) by proposing a two-stage collaborative neural network framework that automatically discovers fine-grained behavioral patterns and stable, long-term playing styles from large-scale telemetry sequences. Methodologically, it introduces a novel “bridging loss” training mechanism that jointly optimizes latent-space clustering, sequential modeling, and supervised style classification—enabling end-to-end learning of psychological traits, tactical preferences, and engagement attribution. Key contributions include: (1) the first generalizable and interpretable taxonomy of playing styles grounded in behavioral psychology; (2) significant performance gains over state-of-the-art baselines in engagement prediction; and (3) psychologically informed behavioral diagnostics that support player experience understanding, progression analysis, and risk intervention.

5 citationsRead paper

ScarceGAN: Discriminative Classification Framework for Rare Class Identification for Longitudinal Data with Weak Prior

Oct 26, 2021International Conference on Information and Knowledge Management

This paper addresses the challenge of identifying extremely rare positive-class instances (e.g., novel attacks, high-risk users) in longitudinal telemetry data, where positive samples are severely scarce, negative samples are heterogeneous and weakly labeled across multiple sources, and abundant unlabeled data leads to insufficient prior knowledge. We propose the first weakly supervised, semi-supervised GAN framework jointly modeling sparse positives and multi-source heterogeneous negatives. Our method introduces a tolerance term to relax noisy negative-label constraints, designs a dual-path discriminator and generator loss to unify the modeling of diverse negative distributions and sparse positives, and integrates weakly supervised discriminative learning with generative data augmentation. Evaluated on a skill-game risk-control task, our approach achieves 85% recall for rare classes—60% higher than state-of-the-art baselines. On KDDCUP99, it successfully detects an attack class constituting only 0.09% of the data, establishing a new benchmark for ultra-rare-class detection.

3 citationsRead paper