arXiv subjects

Shukuan Wang

Publications and source records attributed to Shukuan Wang.

5 recordsLinked to original sources

arXiv2026

AcceRL: A Distributed Asynchronous Reinforcement Learning and World Model Framework for Vision-Language-Action Models

Reinforcement learning (RL) for large-scale Vision-Language-Action (VLA) models is severely bottlenecked by synchronization barriers and the high cost of environment data acquisition. To overcome these challenges, we propose AcceRL, a distributed asynchronous RL framework that physically isolates environment rollouts, model inference, and gradient updates. By eliminating the cascading long-tail idle bubbles inherent in synchronous systems, AcceRL maximizes hardware utilization and ensures scalable throughput. Furthermore, AcceRL features a modular design that supports the integration of diverse, plug-and-play world models into its distributed pipeline. Extensive experiments demonstrate that the base framework achieves highly competitive performance across all four LIBERO~\cite{liu2023libero} task suites. Systematically, the asynchronous architecture delivers a $2.4\times$ throughput speedup over leading synchronous baselines. Algorithmically, by leveraging a world model pre-trained on 1,000 offline trajectories, AcceRL achieves up to a $200\times$ improvement in online sample efficiency on LIBERO-Spatial, establishing a robust framework that is both sample-efficient and time-efficient for embodied AI. Code is included in the supplementary material. Code is available at https://github.com/distanceLu/AcceRL.

cs.LG↗

arXiv2026

GIPO: Gaussian Importance Sampling Policy Optimization

Post-training with reinforcement learning (RL) has recently shown strong promise for advancing multimodal agents beyond supervised imitation. However, RL remains limited by poor data efficiency, particularly in settings where interaction data are scarce and quickly become outdated. To address this challenge, GIPO (Gaussian Importance sampling Policy Optimization) is proposed as a policy optimization objective based on truncated importance sampling, replacing hard clipping with a log-ratio-based Gaussian trust weight to softly damp extreme importance ratios while maintaining non-zero gradients. Theoretical analysis shows that GIPO introduces an implicit, tunable constraint on the update magnitude, while concentration bounds guarantee robustness and stability under finite-sample estimation. Experimental results show that GIPO achieves state-of-the-art performance among clipping-based baselines across a wide range of replay buffer sizes, from near on-policy to highly stale data, while exhibiting superior bias--variance trade-off, high training stability and improved sample efficiency. Code is available at https://github.com/distanceLu/GIPO.

cs.LG↗

arXiv2025

DUET: Dual Model Co-Training for Entire Space CTR Prediction

The pre-ranking stage plays a pivotal role in large-scale recommender systems but faces an intrinsic trade-off between model expressiveness and computational efficiency. Owing to the massive candidate pool and strict latency constraints, industry systems often rely on lightweight two-tower architectures, which are computationally efficient yet limited in estimation capability. As a result, they struggle to capture the complex synergistic and suppressive relationships among candidate items, which are essential for producing contextually coherent and diverse recommendation lists. Moreover, this simplicity further amplifies the Sample Selection Bias (SSB) problem, as coarse-grained models trained on biased exposure data must generalize to a much larger candidate space with distinct distributions. To address these issues, we propose \textbf{DUET} (\textbf{DU}al Model Co-Training for \textbf{E}ntire Space C\textbf{T}R Prediction), a set-wise pre-ranking framework that achieves expressive modeling under tight computational budgets. Instead of scoring items independently, DUET performs set-level prediction over the entire candidate subset in a single forward pass, enabling information-aware interactions among candidates while amortizing the computational cost across the set. Moreover, a dual model co-training mechanism extends supervision to unexposed items via mutual pseudo-label refinement, effectively mitigating SSB. Validated through extensive offline experiments and online A/B testing, DUET consistently outperforms state-of-the-art baselines and achieves improvements across multiple core business metrics. At present, DUET has been fully deployed in Kuaishou and Kuaishou Lite Apps, serving the main traffic for hundreds of millions of users.

cs.IR↗

arXiv2025

MARS: Modality-Aligned Retrieval for Sequence Augmented CTR Prediction

Click-through rate (CTR) prediction serves as a cornerstone of recommender systems. Despite the strong performance of current CTR models based on user behavior modeling, they are still severely limited by interaction sparsity, especially in low-active user scenarios. To address this issue, data augmentation of user behavior is a promising research direction. However, existing data augmentation methods heavily rely on collaborative signals while overlooking the rich multimodal features of items, leading to insufficient modeling of low-active users. To alleviate this problem, we propose a novel framework \textbf{MARS} (\textbf{M}odality-\textbf{A}ligned \textbf{R}etrieval for \textbf{S}equence Augmented CTR Prediction). MARS utilizes a Stein kernel-based approach to align text and image features into a unified and unbiased semantic space to construct multimodal user embeddings. Subsequently, each low-active user's behavior sequence is augmented by retrieving, filtering, and concentrating the most similar behavior sequence of high-active users via multimodal user embeddings. Validated by extensive offline experiments and online A/B tests, our framework MARS consistently outperforms state-of-the-art baselines and achieves substantial growth on core business metrics within Kuaishou~\footnote{https://www.kuaishou.com/}. Consequently, MARS has been successfully deployed, serving the main traffic for hundreds of millions of users. To ensure reproducibility, we provide anonymous access to the implementation code~\footnote{https://github.com/wangshukuan/MARS}.

cs.IR↗

arXiv2024

Monte Carlo Tree Search based Space Transfer for Black-box Optimization

Bayesian optimization (BO) is a popular method for computationally expensive black-box optimization. However, traditional BO methods need to solve new problems from scratch, leading to slow convergence. Recent studies try to extend BO to a transfer learning setup to speed up the optimization, where search space transfer is one of the most promising approaches and has shown impressive performance on many tasks. However, existing search space transfer methods either lack an adaptive mechanism or are not flexible enough, making it difficult to efficiently identify promising search space during the optimization process. In this paper, we propose a search space transfer learning method based on Monte Carlo tree search (MCTS), called MCTS-transfer, to iteratively divide, select, and optimize in a learned subspace. MCTS-transfer can not only provide a well-performing search space for warm-start but also adaptively identify and leverage the information of similar source tasks to reconstruct the search space during the optimization process. Experiments on synthetic functions, real-world problems, Design-Bench and hyper-parameter optimization show that MCTS-transfer can demonstrate superior performance compared to other search space transfer methods under different settings. Our code is available at \url{https://github.com/lamda-bbo/mcts-transfer}.

cs.LG↗