Author name cluster

Yu Shang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

6 papers

2 author rows

EAAI Journal 2026 Journal Article

Exploring the sustainable development path of global digital service trade stability: A hybrid approach perspective

Shikang Kang
Yu Shang

As the digitization process accelerates, unbalanced sustainable development conditions exacerbate the instability and risks of global digital service trade. Based on the theoretical framework of sustainable development, this study takes panel data of 161 economies from 2014 to 2023 as a sample and employs a hybrid approach of empirical analysis-dynamic fuzzy set qualitative comparisons (dynamic QCA)-artificial neural network (ANN) to identify the sustainability capabilities that drive the digital service trade stability (Dts), and to explore the sustainability portfolio paths that generate high Dts and the cases. The results show that 11 drivers are significantly and positively correlated with Dts and that a single sustainability capability does not constitute high Dts. Three combination paths exist to achieve high Dts, with industrialized innovation capabilities distributed across each path. Economic coherence, ecological sustainability, social peace and inclusion, and sustainable health and well-being as alternatives to the combination paths. The most influential antecedent condition is industrialization innovation capacity, followed by ecological sustainability. The findings demonstrate that tailored combinations of sustainable capabilities, rather than any single factor, underpin trade resilience. This study proposes and validates a hybrid research framework for artificial intelligence (AI) empowerment. This framework not only reveals the multiple driving paths of the stability of digital service trade, enriches the research of sustainable development, but also provides a new AI methodology paradigm for the interpretable causal discovery of complex socio-economic systems.

Details DOI

NeurIPS Conference 2025 Conference Paper

AgentRecBench: Benchmarking LLM Agent-based Personalized Recommender Systems

Yu Shang
Peijie Liu
Yuwei Yan
Zijing Wu
Leheng Sheng
Yuanqing Yu
Chumeng Jiang
An Zhang

The emergence of agentic recommender systems powered by Large Language Models (LLMs) represents a paradigm shift in personalized recommendations, leveraging LLMs’ advanced reasoning and role-playing capabilities to enable autonomous, adaptive decision-making. Unlike traditional recommendation approaches, agentic recommender systems can dynamically gather and interpret user-item interactions from complex environments, generating robust recommendation strategies that generalize across diverse scenarios. However, the field currently lacks standardized evaluation protocols to systematically assess these methods. To address this critical gap, we propose: (1) an interactive textual recommendation simulator incorporating rich user and item metadata and three typical evaluation scenarios (classic, evolving-interest, and cold-start recommendation tasks); (2) a unified modular framework for developing agentic recommender systems; and (3) the first comprehensive benchmark comparing over 10 classical and agentic recommendation methods. Our findings demonstrate the superiority of agentic systems and establish actionable design guidelines for their core components. The benchmark environment has been rigorously validated through an open challenge and remains publicly available with a maintained leaderboard at https: //tsinghua-fib-lab. github. io/AgentSocietyChallenge/pages/overview. html. The benchmark is available at: https: //huggingface. co/datasets/SGJQovo/AgentRecBench.

PDF Details

ICLR Conference 2025 Conference Paper

AgentSquare: Automatic LLM Agent Search in Modular Design Space

Yu Shang
Yu Li 0022
Keyu Zhao
Likai Ma
Jiahe Liu
Fengli Xu
Yong Li 0008

Recent advancements in Large Language Models (LLMs) have led to a rapid growth of agentic systems capable of handling a wide range of complex tasks. However, current research largely relies on manual, task-specific design, limiting their adaptability to novel tasks. In this paper, we introduce a new research problem: Modularized LLM Agent Search (MoLAS). We propose a modular design space that abstracts existing LLM agent designs into four fundamental modules with uniform IO interface: Planning, Reasoning, Tool Use, and Memory. Building on this design space, we present a novel LLM agent search framework called AgentSquare, which introduces two core mechanisms, i.e., module evolution and recombination, to efficiently search for optimized LLM agents. To further accelerate the process, we design a performance predictor that uses in-context surrogate models to skip unpromising agent designs. Extensive experiments across six benchmarks, covering the diverse scenarios of web, embodied, tool use and game applications, show that AgentSquare substantially outperforms hand-crafted agents, achieving an average performance gain of 17.2% against best-known human designs. Moreover, AgentSquare can generate interpretable design insights, enabling a deeper understanding of agentic architecture and its impact on task performance. We believe that the modular design space and AgentSquare search framework offer a platform for fully exploiting the potential of prior successful designs and consolidate the collective efforts of research community. Code repo is available at https://github.com/tsinghua-fib-lab/AgentSquare.

Details

NeurIPS Conference 2025 Conference Paper

RoboScape: Physics-informed Embodied World Model

Yu Shang
Xin Zhang
Yinzhou Tang
Lei Jin
Chen Gao
Wei Wu
Yong Li

World models have become indispensable tools for embodied intelligence, serving as powerful simulators capable of generating realistic robotic videos while addressing critical data scarcity challenges. However, current embodied world models exhibit limited physical awareness, particularly in modeling 3D geometry and motion dynamics, resulting in unrealistic video generation for contact-rich robotic scenarios. In this paper, we present RoboScape, a unified physics-informed world model that jointly learns RGB video generation and physics knowledge within an integrated framework. We introduce two key physics-informed joint training tasks: temporal depth prediction that enhances 3D geometric consistency in video rendering, and keypoint dynamics learning that implicitly encodes physical properties (e. g. , object shape and material characteristics) while improving complex motion modeling. Extensive experiments demonstrate that RoboScape generates videos with superior visual fidelity and physical plausibility across diverse robotic scenarios. We further validate its practical utility through downstream applications including robotic policy training with generated data and policy evaluation. Our work provides new insights for building efficient physics-informed world models to advance embodied intelligence research. Our code and demos are available at: https: //github. com/tsinghua-fib-lab/RoboScape.

PDF Details

AAAI Conference 2024 Conference Paper

Transferable Adversarial Attacks for Object Detection Using Object-Aware Significant Feature Distortion

Xinlong Ding
Jiansheng Chen
Hongwei Yu
Yu Shang
Yining Qin
Huimin Ma

Transferable black-box adversarial attacks against classifiers by disturbing the intermediate-layer features have been extensively studied in recent years. However, these methods have not yet achieved satisfactory performances when directly applied to object detectors. This is largely because the features of detectors are fundamentally different from that of the classifiers. In this study, we propose a simple but effective method to improve the transferability of adversarial examples for object detectors by leveraging the properties of spatial consistency and limited equivariance of object detectors’ features. Specifically, we combine a novel loss function and deliberately designed data augmentation to distort the backbone features of object detectors by suppressing significant features corresponding to objects and amplifying the surrounding vicinal features corresponding to object boundaries. As such the target object and background area on the generated adversarial samples are more likely to be confused by other detectors. Extensive experimental results show that our proposed method achieves state-of-the-art black-box transferability for untargeted attacks on various models, including one/two-stage, CNN/Transformer-based, and anchor-free/anchor-based detectors.

PDF Details DOI

YNIMG Journal 2012 Journal Article

Noninvasive optical evaluation of spontaneous low frequency oscillations in cerebral hemodynamics

Ran Cheng
Yu Shang
Don Hayes
Sibu P. Saha
Guoqiang Yu

Spontaneous low frequency oscillations (LFOs) around 0. 1Hz have been observed in mean arterial pressure (MAP) and cerebral blood flow velocity (CBFV). Previous studies have shown that cerebral autoregulation in major arteries can be assessed by quantification of the phase shift between LFOs of MAP and CBFV. However, many cerebral diseases are associated with abnormal microvasculature and tissue dysfunction in brain, and quantification of these abnormalities requires direct measurement of cerebral tissue hemodynamics. This pilot study used a novel hybrid near-infrared diffuse optical instrument to noninvasively and simultaneously detect LFOs of cerebral blood flow (CBF) and cerebral oxygenation (i. e. , oxygenated/deoxygenated/total hemoglobin concentration: [HbO2]/[Hb]/THC) in human prefrontal cortex. Using the hybrid instrument and a finger plethysmograph, the dynamic changes of CBF, [HbO2], [Hb], THC and MAP were concurrently measured in 15 healthy subjects at rest, during 70° head-up-tilting (HUT) and during enforced breathing at 0. 1Hz. The LFOs were extracted from the measured variables using power spectral analysis, and the phase shifts and coherences of LFOs between MAP and each of the measured hemodynamic variables were calculated from the corresponding transfer functions. Levels of coherence (>0. 4) were used to judge the success of LFO measurements. We found that CBF, [HbO2] and THC were reliable hemodynamic parameters in detecting LFOs and HUT was the most robust and stable protocol for quantifying phase shifts of hemodynamic LFOs. Comparing with other relevant studies, similar success rates for detecting cerebral LFOs have been achieved in our study. The phase shifts of LFOs in CBF were also close to those in CBFV reported by other groups, although the results in cerebral oxygenation measurements during enforced breathing varied across studies. Future study will investigate cerebral LFOs in patients with cerebral impairment and evaluate their cerebral autoregulation capabilities and neurocognitive functions via the quantification of LFO phase shifts.

Details DOI