arXiv subjects

Shuo Shen

Publications and source records attributed to Shuo Shen.

5 recordsLinked to original sources

arXiv2025

Complex Instruction Following with Diverse Style Policies in Football Games

Despite advancements in language-controlled reinforcement learning (LC-RL) for basic domains and straightforward commands (e.g., object manipulation and navigation), effectively extending LC-RL to comprehend and execute high-level or abstract instructions in complex, multi-agent environments, such as football games, remains a significant challenge. To address this gap, we introduce Language-Controlled Diverse Style Policies (LCDSP), a novel LC-RL paradigm specifically designed for complex scenarios. LCDSP comprises two key components: a Diverse Style Training (DST) method and a Style Interpreter (SI). The DST method efficiently trains a single policy capable of exhibiting a wide range of diverse behaviors by modulating agent actions through style parameters (SP). The SI is designed to accurately and rapidly translate high-level language instructions into these corresponding SP. Through extensive experiments in a complex 5v5 football environment, we demonstrate that LCDSP effectively comprehends abstract tactical instructions and accurately executes the desired diverse behavioral styles, showcasing its potential for complex, real-world applications.

cs.MA↗

arXiv2025

Noise-Resilient Symbolic Regression with Dynamic Gating Reinforcement Learning

Symbolic regression (SR) has emerged as a pivotal technique for uncovering the intrinsic information within data and enhancing the interpretability of AI models. However, current state-of-the-art (sota) SR methods struggle to perform correct recovery of symbolic expressions from high-noise data. To address this issue, we introduce a novel noise-resilient SR (NRSR) method capable of recovering expressions from high-noise data. Our method leverages a novel reinforcement learning (RL) approach in conjunction with a designed noise-resilient gating module (NGM) to learn symbolic selection policies. The gating module can dynamically filter the meaningless information from high-noise data, thereby demonstrating a high noise-resilient capability for the SR process. And we also design a mixed path entropy (MPE) bonus term in the RL process to increase the exploration capabilities of the policy. Experimental results demonstrate that our method significantly outperforms several popular baselines on benchmarks with high-noise data. Furthermore, our method also can achieve sota performance on benchmarks with clean data, showcasing its robustness and efficacy in SR tasks.

cs.LG↗

arXiv2023

Diversity is Strength: Mastering Football Full Game with Interactive Reinforcement Learning of Multiple AIs

Training AI with strong and rich strategies in multi-agent environments remains an important research topic in Deep Reinforcement Learning (DRL). The AI's strength is closely related to its diversity of strategies, and this relationship can guide us to train AI with both strong and rich strategies. To prove this point, we propose Diversity is Strength (DIS), a novel DRL training framework that can simultaneously train multiple kinds of AIs. These AIs are linked through an interconnected history model pool structure, which enhances their capabilities and strategy diversities. We also design a model evaluation and screening scheme to select the best models to enrich the model pool and obtain the final AI. The proposed training method provides diverse, generalizable, and strong AI strategies without using human data. We tested our method in an AI competition based on Google Research Football (GRF) and won the 5v5 and 11v11 tracks. The method enables a GRF AI to have a high level on both 5v5 and 11v11 tracks for the first time, which are under complex multi-agent environments. The behavior analysis shows that the trained AI has rich strategies, and the ablation experiments proved that the designed modules benefit the training process.

cs.AI↗

arXiv2022

Spatial data sharing with secure multi-party computation for exploratory spatial data analysis

Spatial data sharing plays a significant role in opening data research and promoting government agency transparency. However, valuable spatial data, like high-precision geographic information and personal traffic records, cannot be made public because they may incur leakage risks such as intrusion, theft, and the unauthorised sale of proprietary information. When participants with confidential data distrust each other but want to use the other datasets for calculations, the most common solution is to provide their original data to a trusted third party. However, the trusted third party frequently risks being attacked and having the data leaked. To maintain data controllability, most companies and organisations refuse to share their data. In this study, we introduce secure multi-party computation to spatial data sharing to address the sharing problem. Additionally, we describe the design and implementation of the protocols of two exploratory spatial data analyses: global Moran's I and local Moran's I. Furthermore, we build a system to demonstrate process realisation and results visualisation. Comparing our system with existing data-sharing schemes, our system Identifies the correct result without incurring leaking risks during spatial data sharing.

cs.CR↗

arXiv2013

Tuning the phase sensitivity of a double-lambda system with a static magnetic field

We study the effect of a DC magnetic field on the phase sensitivity of a double-lambda system coupled by two laser fields, a probe and a pump. It is demonstrated that the gain and the refractive index of the probe can be controlled by either the magnetic field or the relative phase between the two laser fields. More interestingly, when the system reduces to a single-lambda system, turning on the magnetic field transforms the system from a phase-insensitive process to a phase-sensitive one. In the pulsed-probe regime, we observed switching between slow and fast light when the magnetic field or the relative phase was adjusted. Experiments using a coated 87Rb vapor cell produced results in good agreement with our numerical simulation. This work provides a novel and simple means to manipulate phase sensitive electromagnetically-induced-transparency or four-wave mixing, and could be useful for applications in quantum optics, nonlinear optics and magnetometery based on such systems.

quant-ph↗