BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Yicheng Pan

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

ResearcherUnclaimed source record
physics.soc-phSocial and Information NetworksArtificial IntelligenceComputer VisionMachine Learning

Catalog footprint

What is connected

9works
5topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

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.

Published work

9 published item(s)

preprint2026arXiv

Unlocking Complex Visual Generation via Closed-Loop Verified Reasoning

Despite rapid advancements, current text-to-image (T2I) models predominantly rely on a single-step generation paradigm, which struggles with complex semantics and faces diminishing returns from parameter scaling. While recent multi-step reasoning approaches show promise, they are hindered by ungrounded planning hallucinations lacking verification, monolithic post-hoc reflection, long-context optimization instabilities, and prohibitive inference latency. To overcome these bottlenecks, we propose the Closed-Loop Visual Reasoning (CLVR) framework, a comprehensive system that deeply couples visual-language logical planning with pixel-level diffusion generation. CLVR introduces an automated data engine with step-level visual verification to synthesize reliable reasoning trajectories, and proposes Proxy Prompt Reinforcement Learning (PPRL) to resolve long-context optimization instabilities by distilling interleaved multimodal histories into explicit reward signals for accurate causal attribution. Furthermore, to mitigate the severe latency bottleneck caused by iterative denoising, we propose $Δ$-Space Weight Merge (DSWM), a theoretically grounded method that fuses alignment weights with off-the-shelf distillation priors, reducing the per-step inference cost to just 4 NFEs without requiring expensive re-distillation. Extensive experiments demonstrate that CLVR outperforms existing open-source baselines across multiple benchmarks and approaches the performance of proprietary commercial models, unlocking general test-time scaling capabilities for complex visual generation.

0 citations0 reviewsNo code linkedOpen access
preprint2025arXiv

Structure-Guided Allocation of 2D Gaussians for Image Representation and Compression

Recent advances in 2D Gaussian Splatting (2DGS) have demonstrated its potential as a compact image representation with millisecond-level decoding. However, existing 2DGS-based pipelines allocate representation capacity and parameter precision largely oblivious to image structure, limiting their rate-distortion (RD) efficiency at low bitrates. To address this, we propose a structure-guided allocation principle for 2DGS, which explicitly couples image structure with both representation capacity and quantization precision, while preserving native decoding speed. First, we introduce a structure-guided initialization that assigns 2D Gaussians according to spatial structural priors inherent in natural images, yielding a localized and semantically meaningful distribution. Second, during quantization-aware fine-tuning, we propose adaptive bitwidth quantization of covariance parameters, which grants higher precision to small-scale Gaussians in complex regions and lower precision elsewhere, enabling RD-aware optimization, thereby reducing redundancy without degrading edge quality. Third, we impose a geometry-consistent regularization that aligns Gaussian orientations with local gradient directions to better preserve structural details. Extensive experiments demonstrate that our approach substantially improves both the representational power and the RD performance of 2DGS while maintaining over 1000 FPS decoding. Compared with the baseline GSImage, we reduce BD-rate by 43.44% on Kodak and 29.91% on DIV2K.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

A Simple yet Effective Method for Graph Classification

In deep neural networks, better results can often be obtained by increasing the complexity of previously developed basic models. However, it is unclear whether there is a way to boost performance by decreasing the complexity of such models. Intuitively, given a problem, a simpler data structure comes with a simpler algorithm. Here, we investigate the feasibility of improving graph classification performance while simplifying the learning process. Inspired by structural entropy on graphs, we transform the data sample from graphs to coding trees, which is a simpler but essential structure for graph data. Furthermore, we propose a novel message passing scheme, termed hierarchical reporting, in which features are transferred from leaf nodes to root nodes by following the hierarchical structure of coding trees. We then present a tree kernel and a convolutional network to implement our scheme for graph classification. With the designed message passing scheme, the tree kernel and convolutional network have a lower runtime complexity of $O(n)$ than Weisfeiler-Lehman subtree kernel and other graph neural networks of at least $O(hm)$. We empirically validate our methods with several graph classification benchmarks and demonstrate that they achieve better performance and lower computational consumption than competing approaches.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Community Structures Are Definable in Networks, and Universal in Real World

Community detecting is one of the main approaches to understanding networks \cite{For2010}. However it has been a longstanding challenge to give a definition for community structures of networks. Here we found that community structures are definable in networks, and are universal in real world. We proposed the notions of entropy- and conductance-community structure ratios. It was shown that the definitions of the modularity proposed in \cite{NG2004}, and our entropy- and conductance-community structures are equivalent in defining community structures of networks, that randomness in the ER model \cite{ER1960} and preferential attachment in the PA \cite{Bar1999} model are not mechanisms of community structures of networks, and that the existence of community structures is a universal phenomenon in real networks. Our results demonstrate that community structure is a universal phenomenon in the real world that is definable, solving the challenge of definition of community structures in networks. This progress provides a foundation for a structural theory of networks.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Community Structures Are Definable in Networks: A Structural Theory of Networks

We found that neither randomness in the ER model nor the preferential attachment in the PA model is the mechanism of community structures of networks, that community structures are universal in real networks, that community structures are definable in networks, that communities are interpretable in networks, and that homophyly is the mechanism of community structures and a structural theory of networks. We proposed the notions of entropy- and conductance-community structures. It was shown that the two definitions of the entropy- and conductance-community structures and the notion of modularity proposed by physicists are all equivalent in defining community structures of networks, that neither randomness in the ER model nor preferential attachment in the PA model is the mechanism of community structures of networks, and that the existence of community structures is a universal phenomenon in real networks. This poses a fundamental question: What are the mechanisms of community structures of real networks? To answer this question, we proposed a homophyly model of networks. It was shown that networks of our model satisfy a series of new topological, probabilistic and combinatorial principles, including a fundamental principle, a community structure principle, a degree priority principle, a widths principle, an inclusion and infection principle, a king node principle and a predicting principle etc. The new principles provide a firm foundation for a structural theory of networks. Our homophyly model demonstrates that homophyly is the underlying mechanism of community structures of networks, that nodes of the same community share common features, that power law and small world property are never obstacles of the existence of community structures in networks, that community structures are {\it definable} in networks, and that (natural) communities are {\it interpretable}.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Dimensions, Structures and Security of Networks

One of the main issues in modern network science is the phenomenon of cascading failures of a small number of attacks. Here we define the dimension of a network to be the maximal number of functions or features of nodes of the network. It was shown that there exist linear networks which are provably secure, where a network is linear, if it has dimension one, that the high dimensions of networks are the mechanisms of overlapping communities, that overlapping communities are obstacles for network security, and that there exists an algorithm to reduce high dimensional networks to low dimensional ones which simultaneously preserves all the network properties and significantly amplifies security of networks. Our results explore that dimension is a fundamental measure of networks, that there exist linear networks which are provably secure, that high dimensional networks are insecure, and that security of networks can be amplified by reducing dimensions.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Homophyly and Randomness Resist Cascading Failure in Networks

The universal properties of power law and small world phenomenon of networks seem unavoidably obstacles for security of networking systems. Existing models never give secure networks. We found that the essence of security is the security against cascading failures of attacks and that nature solves the security by mechanisms. We proposed a model of networks by the natural mechanisms of homophyly, randomness and preferential attachment. It was shown that homophyly creates a community structure, that homophyly and randomness introduce ordering in the networks, and that homophyly creates inclusiveness and introduces rules of infections. These principles allow us to provably guarantee the security of the networks against any attacks. Our results show that security can be achieved provably by structures, that there is a tradeoff between the roles of structures and of thresholds in security engineering, and that power law and small world property are never obstacles for security of networks.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Homophyly Networks -- A Structural Theory of Networks

A grand challenge in network science is apparently the missing of a structural theory of networks. The authors have showed that the existence of community structures is a universal phenomenon in real networks, and that neither randomness nor preferential attachment is a mechanism of community structures of network \footnote{A. Li, J. Li, and Y. Pan, Community structures are definable in networks, and universal in the real world, To appear.}. This poses a fundamental question: What are the mechanisms of community structures of real networks? Here we found that homophyly is the mechanism of community structures and a structural theory of networks. We proposed a homophyly model. It was shown that networks of our model satisfy a series of new topological, probabilistic and combinatorial principles, including a fundamental principle, a community structure principle, a degree priority principle, a widths principle, an inclusion and infection principle, a king node principle, and a predicting principle etc, leading to a structural theory of networks. Our model demonstrates that homophyly is the underlying mechanism of community structures of networks, that nodes of the same community share common features, that power law and small world property are never obstacles of the existence of community structures in networks, and that community structures are definable in networks.

0 citations0 reviewsNo code linkedOpen access
preprint2013arXiv

Provable Security of Networks

We propose a definition of {\it security} and a definition of {\it robustness} of networks against the cascading failure models of deliberate attacks and random errors respectively, and investigate the principles of the security and robustness of networks. We propose a {\it security model} such that networks constructed by the model are provably secure against any attacks of small sizes under the cascading failure models, and simultaneously follow a power law, and have the small world property with a navigating algorithm of time complex $O(\log n)$. It is shown that for any network $G$ constructed from the security model, $G$ satisfies some remarkable topological properties, including: (i) the {\it small community phenomenon}, that is, $G$ is rich in communities of the form $X$ of size poly logarithmic in $\log n$ with conductance bounded by $O(\frac{1}{|X|^β})$ for some constant $β$, (ii) small diameter property, with diameter $O(\log n)$ allowing a navigation by a $O(\log n)$ time algorithm to find a path for arbitrarily given two nodes, and (iii) power law distribution, and satisfies some probabilistic and combinatorial principles, including the {\it degree priority theorem}, and {\it infection-inclusion theorem}. By using these principles, we show that a network $G$ constructed from the security model is secure for any attacks of small scales under both the uniform threshold and random threshold cascading failure models. Our security theorems show that networks constructed from the security model are provably secure against any attacks of small sizes, for which natural selections of {\it homophyly, randomness} and {\it preferential attachment} are the underlying mechanisms.

0 citations0 reviewsNo code linkedOpen access