Source author record

Hua Yang

Hua Yang appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

hep-ph physics.soc-ph Artificial Intelligence Computation and Language Computer Vision Machine Learning physics.data-an physics.plasm-ph quant-ph Robotics

Catalog footprint

What is connected

9works

10topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

MAD-OPD: Breaking the Ceiling in On-Policy Distillation via Multi-Agent Debate

On-policy distillation (OPD) trains a student on its own trajectories under token-level teacher supervision, but existing methods are capped by a single-teacher capability ceiling: when the teacher errs, the student inherits the error. OPD also remains largely unexplored in agentic tasks, where per-step errors compound across long trajectories and destabilize training. We propose MAD-OPD (Multi-Agent Debate-driven On-Policy Distillation), which breaks this ceiling by recasting the distillation teacher as a deliberative collective of teachers that debate over the student's on-policy state; the debate produces an emergent collective intelligence that supplies token-level supervision, with each teacher's contribution weighted by its post-debate confidence. To extend OPD to agentic tasks, we also introduce On-Policy Agentic Distillation (OPAD), which adds step-level sampling to stabilize training under multi-step error compounding. We additionally derive a task-adaptive divergence principle, selecting JSD (Jensen-Shannon divergence) for agentic stability and reverse KL (Kullback-Leibler) divergence for code generation, and verify it both theoretically and empirically. Across six teacher-student configurations (Qwen3 and Qwen3.5; 1.7B-14B students, 8B-32B teachers) and five agentic and code benchmarks, MAD-OPD ranks first across all six configurations; on the 14B+8B$\to$4B setting it lifts the agentic average by $+2.4\%$ and the code average by $+3.7\%$ over the stronger single-teacher OPD.

preprint2022arXiv

Autonomous Electric Vehicle Battery Disassembly Based on NeuroSymbolic Computing

The booming of electric vehicles demands efficient battery disassembly for recycling to be environment-friendly. Due to the unstructured environment and high uncertainties, battery disassembly is still primarily done by humans, probably assisted by robots. It is highly desirable to design autonomous solutions to improve work efficiency and lower human risks in high voltage and toxic environments. This paper proposes a novel framework of the NeuroSymbolic task and motion planning method to disassemble batteries in an unstructured environment using robots automatically. It enables robots to independently locate and disassemble battery bolts, with or without obstacles. This study not only provides a solution for intelligently disassembling electric vehicle batteries but also verifies its feasibility through a set of test results with the robot accomplishing the disassembly tasks in a complex and dynamic environment.

preprint2020arXiv

Weighted Bilinear Coding over Salient Body Parts for Person Re-identification

Deep convolutional neural networks (CNNs) have demonstrated dominant performance in person re-identification (Re-ID). Existing CNN based methods utilize global average pooling (GAP) to aggregate intermediate convolutional features for Re-ID. However, this strategy only considers the first-order statistics of local features and treats local features at different locations equally important, leading to sub-optimal feature representation. To deal with these issues, we propose a novel weighted bilinear coding (WBC) framework for local feature aggregation in CNN networks to pursue more representative and discriminative feature representations, which can adapt to other state-of-the-art methods and improve their performance. In specific, bilinear coding is used to encode the channel-wise feature correlations to capture richer feature interactions. Meanwhile, a weighting scheme is applied on the bilinear coding to adaptively adjust the weights of local features at different locations based on their importance in recognition, further improving the discriminability of feature aggregation. To handle the spatial misalignment issue, we use a salient part net (spatial attention module) to derive salient body parts, and apply the WBC model on each part. The final representation, formed by concatenating the WBC encoded features of each part, is both discriminative and resistant to spatial misalignment. Experiments on three benchmarks including Market-1501, DukeMTMC-reID and CUHK03 evidence the favorable performance of our method against other outstanding methods.

preprint2015arXiv

A new electromagnetic code for ICRF antenna in EAST

The demand for an effective tool to help in the design of ion cyclotron radio frequency (ICRF) antenna system for fusion experiment has driven the development of predictive codes. A new electromagnetic code based on the method of moments (MOM) is described in the paper. The code computes the electromagnetic field by the solution of the electric field integral equation. The structure of ICRF antennas are discretized with triangular mesh. By using the new code, the scattering parameter and the surface current are given and compared with the result by commercial code CST. Moreover, the power spectra are studied with different toroidal phases for heating and current drive. Good agreement of simulation results between the new code and CST are obtained. The code has been validated against CST for EAST ICRF antenna.

preprint2012arXiv

Experimental realization of large-alphabet quantum key distribution protocol using orbital angular momentum entanglement

We experimentally demonstrate a quantum key distribution (QKD) protocol using photon pairs entangled in orbit angular momentum (OAM). In our protocol, Alice and Bob modulate their OAM states on each entangled pair with spatial light modulators (SLMs), respectively. Alice uses a fixed phase hologram in her SLM, while Bob designs $N$ different suitable phase holograms and uses them to represent his $N$-based information in his SLM. With coincidences, Alice can fully retrieve the key stream sent by Bob without information reconciliation or privacy amplification. We report the experiment results with N=3 and the sector states with OAM eigenmodes l=1 and l=-1. Our experiment shows that the coincidence rates are in relatively distinct value regions for the three different key elements. Alice could recover fully Bob's keys by the protocol. Finally, we discuss the security of the protocol both form the light way and against the general attacks.

preprint2009arXiv

Measuring Significance of Community Structure in Complex Networks

Many complex systems can be represented as networks and separating a network into communities could simplify the functional analysis considerably. Recently, many approaches have been proposed for finding communities, but none of them can evaluate the communities found are significant or trivial definitely. In this paper, we propose an index to evaluate the significance of communities in networks. The index is based on comparing the similarity between the original community structure in network and the community structure of the network after perturbed, and is defined by integrating all the similarities. Many artificial networks and real-world networks are tested. The results show that the index is independent from the size of network and the number of communities. Moreover, we find the clear communities always exist in social networks, but don't find significative communities in proteins interaction networks and metabolic networks.

preprint2009arXiv

Scaling properties in spatial networks and its effects on topology and traffic dynamics

Empirical studies on the spatial structures in several real transport networks reveal that the distance distribution in these networks obeys power law. To discuss the influence of the power-law exponent on the network's structure and function, a spatial network model is proposed. Based on a regular network and subject to a limited cost $C$, long range connections are added with power law distance distribution $P(r)=ar^{-δ}$. Some basic topological properties of the network with different $δ$ are studied. It is found that the network has the smallest average shortest path when $δ=2$. Then a traffic model on this network is investigated. It is found that the network with $δ=1.5$ is best for the traffic process. All of these results give us some deep understandings about the relationship between spatial structure and network function.

preprint2006arXiv

The correction of the littlest Higgs model to the Higgs production process $e^+e^-\to e^+e^-H$ at the ILC

The littlest Higgs model is the most economical one among various little Higgs models. In the context of the littlest Higgs model, we study the process $e^{+}e^{-}\to e^{+}e^{-}H$ at the ILC and calculate the correction of the littlest Higgs model to the cross section of this process. The results show that, in the favorable parameter spaces preferred by the electroweak precision data, the value of the relative correction is in the range from a few percent to tens percent. In most case, the correction is large enough to reach the measurement precision of the ILC. Therefore, the correction of the littlest Higgs model to the process $e^{+}e^{-}\to e^{+}e^{-}H$ might be detected at the ILC which will give an ideal way to test the model.

preprint2002arXiv

On QCD Predictions for the Chiral Lagrangian Coefficients

Based on a previous study of deriving the chiral Lagrangian (CL) from QCD, we illustrate the main feature of QCD predictions for the CL coefficients (CLC) in certain approximations. We first show that, in the large-N(c) limit, the anomaly part contributions to the CLC are exactly cancelled by certain terms in the normal part contributions (NPC), so that the final results only concern the remaining NPC depending on QCD interactions. We then do the calculation in a simple approach with further approximations. The obtained CLC and quark condensate are consistent with the experiments.