Rong Pan

AAAI Conference 2026 Conference Paper

StoryBox: Collaborative Multi-Agent Simulation for Hybrid Bottom-Up Long-Form Story Generation Using Large Language Models

• Zehao Chen
• Rong Pan
• Haoran Li

Human writers often begin their stories with an overarching mental scene, where they envision the interactions between characters and their environment. Inspired by this creative process, we propose a novel approach to long-form story generation, termed hybrid bottom-up long-form story generation, using multi-agent simulations. In our method, agents interact within a dynamic sandbox environment, where their behaviors and interactions with one another and the environment generate emergent events. These events form the foundation for the story, enabling organic character development and plot progression. Unlike traditional top-down approaches that impose rigid structures, our hybrid bottom-up approach allows for the natural unfolding of events, fostering more spontaneous and engaging storytelling. The system is capable of generating stories exceeding 10,000 words while maintaining coherence and consistency, addressing some of the key challenges faced by current story generation models. We achieve state-of-the-art performance across several metrics. This approach offers a scalable and innovative solution for creating dynamic, immersive long-form stories that evolve organically from agent-driven interactions.

TMLR Journal 2026 Journal Article

Topology-Guided Graph Pre-training and Prompt Learning on Directed Graphs

• Peiyu Liang
• Chenguang Yang
• Yixuan He
• Rong Pan
• Yuzhou Chen

In recent years, graph neural networks (GNNs) have been the dominant approach for graph representation learning, leading to new state-of-the-art results on many classification and prediction tasks. However, they are limited by the fact that they cannot effectively learn expressive node representations without the guide of labels, thus suffering from the labeled data scarcity problem. To address the challenges of labeling costs and improve robustness in few-shot scenarios, pre-training on self-supervised tasks has garnered significant attention. Additionally, numerous prompting methods have been proposed as effective ways to bridge the gap between pretext tasks and downstream applications. Although graph pre-training and prompt tuning methods have explored various downstream tasks on undirected graphs, directed graphs have been largely under-explored, and these models suffer limitations in capturing directional and topological information in directed graphs. In this paper, we propose a novel topology-guided directed graph pre-training and prompt tuning model, named TopoDIG, that can effectively capture intrinsic directional structural and local topological features in directed graphs. These features play essential roles in transferring knowledge from a pre-trained model to downstream tasks. TopoDIG consists of an encoder in the form of a magnetic Laplacian matrix, a topological encoder, and a graph prompt learning function. Experimental results on both real-world and synthetic directed graphs demonstrate the superior performance of TopoDIG compared to prominent baseline methods.

JBHI Journal 2025 Journal Article

A Trusted Medical Image Zero-Watermarking Scheme Based on DCNN and Hyperchaotic System

• Ruotong Xiang
• Gang Liu
• Min Dang
• Quan Wang
• Rong Pan

The zero-watermarking methods provide a means of lossless, which was adopted to protect medical image copyright requiring high integrity. However, most existing studies have only focused on robustness and there has been little discussion about the analysis and experiment on discriminability. Therefore, this paper proposes a trusted robust zero-watermarking scheme for medical images based on Deep convolution neural network (DCNN) and the hyperchaotic encryption system. Firstly, the medical image is converted into several feature map matrices by the specific convolution layer of DCNN. Then, a stable Gram matrix is obtained by calculating the colinear correlation between different channels in feature map matrices. Finally, the Gram matrixes of the medical image and the feature map matrixes of the watermark image are fused by the trained DCNN to generate the zero-watermark. Meanwhile, we propose two feature evaluation criteria for finding differentiated eigenvalues. The eigenvalue is used as the explicit key to encrypt the generated zero-watermark by Lorenz hyperchaotic encryption, which enhances security and discriminability. The experimental results show that the proposed scheme can resist common image attacks and geometric attacks, and is distinguishable in experiments, being applicable for the copyright protection of medical images.

AAAI Conference 2025 Conference Paper

Causal Discovery by Interventions via Integer Programming

• Abdelmonem Elrefaey
• Rong Pan

Causal discovery is essential across various scientific fields to uncover causal structures within data. Traditional methods relying on observational data have limitations due to confounding variables. This paper presents an optimization-based approach using integer programming (IP) to design minimal intervention sets that ensure causal structure identifiability. Our method provides exact and modular solutions, adaptable to different experimental settings and constraints. We demonstrate its effectiveness through comparative analysis across different settings demonstrating its applicability and robustness.

AAAI Conference 2025 Conference Paper

SVGBuilder: Component-Based Colored SVG Generation with Text-Guided Autoregressive Transformers

• Zehao Chen
• Rong Pan

Scalable Vector Graphics (SVG) are essential XML-based formats for versatile graphics, offering resolution independence and scalability. Unlike raster images, SVGs use geometric shapes and support interactivity, animation, and manipulation via CSS and JavaScript. Current SVG generation methods face challenges related to high computational costs and complexity. In contrast, human designers use component-based tools for efficient SVG creation. Inspired by this, SVGBuilder introduces a component-based, autoregressive model for generating high-quality colored SVGs from textual input. It significantly reduces computational overhead and improves efficiency compared to traditional methods. Our model generates SVGs up to 604 times faster than optimization-based approaches. To address the limitations of existing SVG datasets and support our research, we introduce ColorSVG-100K, the first large-scale dataset of colored SVGs, comprising 100,000 graphics. This dataset fills the gap in color information for SVG generation models and enhances diversity in model training. Evaluation against state-of-the-art models demonstrates SVGBuilder's superior performance in practical applications, highlighting its efficiency and quality in generating complex SVG graphics.

IJCAI Conference 2022 Conference Paper

FedCG: Leverage Conditional GAN for Protecting Privacy and Maintaining Competitive Performance in Federated Learning

• Yuezhou Wu
• Yan Kang
• Jiahuan Luo
• Yuanqin He
• Lixin Fan
• Rong Pan
• Qiang Yang

Federated learning (FL) aims to protect data privacy by enabling clients to build machine learning models collaboratively without sharing their private data. Recent works demonstrate that information exchanged during FL is subject to gradient-based privacy attacks and, consequently, a variety of privacy-preserving methods have been adopted to thwart such attacks. However, these defensive methods either introduce orders of magnitudes more computational and communication overheads (e. g. , with homomorphic encryption) or incur substantial model performance losses in terms of prediction accuracy (e. g. , with differential privacy). In this work, we propose FEDCG, a novel federated learning method that leverages conditional generative adversarial networks to achieve high-level privacy protection while still maintaining competitive model performance. FEDCG decomposes each client's local network into a private extractor and a public classifier and keeps the extractor local to protect privacy. Instead of exposing extractors, FEDCG shares clients' generators with the server for aggregating clients' shared knowledge aiming to enhance the performance of each client's local networks. Extensive experiments demonstrate that FEDCG can achieve competitive model performance compared with FL baselines, and privacy analysis shows that FEDCG has a high-level privacy-preserving capability.

AAAI Conference 2019 Conference Paper

Logic Attention Based Neighborhood Aggregation for Inductive Knowledge Graph Embedding

• Peifeng Wang
• Jialong Han
• Chenliang Li
• Rong Pan

Knowledge graph embedding aims at modeling entities and relations with low-dimensional vectors. Most previous methods require that all entities should be seen during training, which is unpractical for real-world knowledge graphs with new entities emerging on a daily basis. Recent efforts on this issue suggest training a neighborhood aggregator in conjunction with the conventional entity and relation embeddings, which may help embed new entities inductively via their existing neighbors. However, their neighborhood aggregators neglect the unordered and unequal natures of an entity’s neighbors. To this end, we summarize the desired properties that may lead to effective neighborhood aggregators. We also introduce a novel aggregator, namely, Logic Attention Network (LAN), which addresses the properties by aggregating neighbors with both rules- and network-based attention weights. By comparing with conventional aggregators on two knowledge graph completion tasks, we experimentally validate LAN’s superiority in terms of the desired properties.

AAAI Conference 2018 Conference Paper

Incorporating GAN for Negative Sampling in Knowledge Representation Learning

• Peifeng Wang
• Shuangyin Li
• Rong Pan

Knowledge representation learning aims at modeling knowledge graph by encoding entities and relations into a low dimensional space. Most of the traditional works for knowledge embedding need negative sampling to minimize a marginbased ranking loss. However, those works construct negative samples through a random mode, by which the samples are often too trivial to fit the model efficiently. In this paper, we propose a novel knowledge representation learning framework based on Generative Adversarial Networks (GAN). In this GAN-based framework, we take advantage of a generator to obtain high-quality negative samples. Meanwhile, the discriminator in GAN learns the embeddings of the entities and relations in knowledge graph. Thus, we can incorporate the proposed GAN-based framework into various traditional models to improve the ability of knowledge representation learning. Experimental results show that our proposed GANbased framework outperforms baselines on triplets classification and link prediction tasks.

AAAI Conference 2018 Conference Paper

Mention and Entity Description Co-Attention for Entity Disambiguation

• Feng Nie
• Yunbo Cao
• Jinpeng Wang
• Chin-Yew Lin
• Rong Pan

For the task of entity disambiguation, mention contexts and entity descriptions both contain various kinds of information content while only a subset of them are helpful for disambiguation. In this paper, we propose a type-aware co-attention model for entity disambiguation, which tries to identify the most discriminative words from mention contexts and most relevant sentences from corresponding entity descriptions simultaneously. To bridge the semantic gap between mention contexts and entity descriptions, we further incorporate entity type information to enhance the co-attention mechanism. Our evaluation shows that the proposed model outperforms the state-of-the-arts on three public datasets. Further analysis also confirms that both the co-attention mechanism and the type-aware mechanism are effective.

AAAI Conference 2017 Conference Paper

Automatic Emphatic Information Extraction from Aligned Acoustic Data and Its Application on Sentence Compression

• Yanju Chen
• Rong Pan

We introduce a novel method to extract and utilize the semantic information from acoustic data. By automatic Speech-To- Text alignment techniques, we are able to detect word-based acoustic durations that can prosodically emphasize specific words in an utterance. We model and analyze the sentencebased emphatic patterns by predicting the emphatic levels using only the lexical features, and demonstrate the potential ability of emphatic information produced by such an unsupervised method to improve the performance of NLP tasks, such as sentence compression, by providing weak supervision on multi-task learning based on LSTMs.

AAAI Conference 2017 Conference Paper

Recurrent Attentional Topic Model

• Shuangyin Li
• Yu Zhang
• Rong Pan
• Mingzhi Mao
• Yang Yang

In a document, the topic distribution of a sentence depends on both the topics of preceding sentences and its own content, and it is usually affected by the topics of the preceding sentences with different weights. It is natural that a document can be treated as a sequence of sentences. Most existing works for Bayesian document modeling do not take these points into consideration. To fill this gap, we propose a Recurrent Attentional Topic Model (RATM) for document embedding. The RATM not only takes advantage of the sequential orders among sentence but also use the attention mechanism to model the relations among successive sentences. In RATM, we propose a Recurrent Attentional Bayesian Process (RABP) to handle the sequences. Based on the RABP, RATM fully utilizes the sequential information of the sentences in a document. Experiments on two copora show that our model outperforms state-of-the-art methods on document modeling and classification.

IJCAI Conference 2017 Conference Paper

Self-paced Compensatory Deep Boltzmann Machine for Semi-Structured Document Embedding

• Shuangyin Li
• Rong Pan
• Jun Yan

In the last decade, there has been a huge amount of documents with different types of rich metadata information, which belongs to the Semi-Structured Documents (SSDs), appearing in many real applications. It is an interesting research work to model this type of text data following the way how humans understand text with informative metadata. In the paper, we introduce a Self-paced Compensatory Deep Boltzmann Machine (SCDBM) architecture that learns a deep neural network by using metadata information to learn deep structure layer-wisely for Semi-Structured Documents (SSDs) embedding in a self-paced way. Inspired by the way how humans understand text, the model defines a deep process of document vector extraction beyond the space of words by jointing the metadata where each layer selects different types of metadata. We present efficient learning and inference algorithms for the SCDBM model and empirically demonstrate that using the representation discovered by this model has better performance on semi-structured document classification and retrieval, and tag prediction comparing with state-of-the-art baselines.

UAI Conference 2016 Conference Paper

Correlated Tag Learning in Topic Model

• Shuangyin Li
• Rong Pan
• Yu Zhang 0006
• Qiang Yang 0001

It is natural to expect that the documents in a corpus will be correlated, and these correlations are reflected by not only the words but also the observed tags in each document. Most previous works model this type of corpus, which are called the semi-structured corpus, without considering the correlations among the tags. In this work, we develop a Correlated Tag Learning (CTL) model for semi-structured corpora based on the topic model to enable the construction of the correlation graph among tags via a logistic normal participation process. For the inference of the CTL model, we devise a variational inference algorithm to approximate the posterior. In experiments, we visualize the tag correlation graph generated by the CTL model on the DBLP corpus and for the tasks of document retrieval and classification, the correlation graph among tags is helpful to improve the generalization performance compared with the state-of-the-art baselines.

AAAI Conference 2015 Conference Paper

Personalized Tag Recommendation through Nonlinear Tensor Factorization Using Gaussian Kernel

• Xiaomin Fang
• Rong Pan
• Guoxiang Cao
• Xiuqiang He
• Wenyuan Dai

Personalized tag recommendation systems recommend a list of tags to a user when he is about to annotate an item. It exploits the individual preference and the characteristic of the items. Tensor factorization techniques have been applied to many applications, such as tag recommendation. Models based on Tucker Decomposition can achieve good performance but require a lot of computation power. On the other hand, models based on Canonical Decomposition can run in linear time and are more feasible for online recommendation. In this paper, we propose a novel method for personalized tag recommendation, which can be considered as a nonlinear extension of Canonical Decomposition. Different from linear tensor factorization, we exploit Gaussian radial basis function to increase the model’s capacity. The experimental results show that our proposed method outperforms the state-of-the-art methods for tag recommendation on real datasets and perform well even with a small number of features, which verifies that our models can make better use of features.

IJCAI Conference 2013 Conference Paper

Tag-Weighted Topic Model for Mining Semi-Structured Documents

• Shuangyin Li
• Jiefei Li
• Rong Pan

In the last decade, latent Dirichlet allocation (LDA) successfully discovers the statistical distribution of the topics over a unstructured text corpus. Meanwhile, more and more document data come up with rich human-provided tag information during the evolution of the Internet, which called semistructured data. The semi-structured data contain both unstructured data (e. g. , plain text) and metadata, such as papers with authors and web pages with tags. In general, different tags in a document play different roles with their own weights. To model such semi-structured documents is nontrivial. In this paper, we propose a novel method to model tagged documents by a topic model, called Tag-Weighted Topic Model (TWTM). TWTM is a framework that leverages the tags in each document to infer the topic components for the documents. This allows not only to learn document-topic distributions, but also to infer the tag-topic distributions for text mining (e. g. , classification, clustering, and recommendations). Moreover, TWTM automatically infers the probabilistic weights of tags for each document. We present an efficient variational inference method with an EM algorithm for estimating the model parameters. The experimental results show that our TWTM approach outperforms the baseline algorithms over three corpora in document modeling and text classification.

YNIMG Journal 2012 Journal Article

A prior feature SVM-MRF based method for mouse brain segmentation

• Teresa Wu
• Min Hyeok Bae
• Min Zhang
• Rong Pan
• Alexandra Badea

We introduce an automated method, called prior feature Support Vector Machine-Markov Random Field (pSVMRF), to segment three-dimensional mouse brain Magnetic Resonance Microscopy (MRM) images. Our earlier work, extended MRF (eMRF) integrated Support Vector Machine (SVM) and Markov Random Field (MRF) approaches, leading to improved segmentation accuracy; however, the computation of eMRF is very expensive, which may limit its performance on segmentation and robustness. In this study pSVMRF reduces training and testing time for SVM, while boosting segmentation performance. Unlike the eMRF approach, where MR intensity information and location priors are linearly combined, pSVMRF combines this information in a nonlinear fashion, and enhances the discriminative ability of the algorithm. We validate the proposed method using MR imaging of unstained and actively stained mouse brain specimens, and compare segmentation accuracy with two existing methods: eMRF and MRF. C57BL/6 mice are used for training and testing, using cross validation. For formalin fixed C57BL/6 specimens, pSVMRF outperforms both eMRF and MRF. The segmentation accuracy for C57BL/6 brains, stained or not, was similar for larger structures like hippocampus and caudate putamen, (~87%), but increased substantially for smaller regions like susbtantia nigra (from 78. 36% to 91. 55%), and anterior commissure (from ~50% to ~80%). To test segmentation robustness against increased anatomical variability we add two strains, BXD29 and a transgenic mouse model of Alzheimer's disease. Segmentation accuracy for new strains is 80% for hippocampus, and caudate putamen, indicating that pSVMRF is a promising approach for phenotyping mouse models of human brain disorders.

IJCAI Conference 2011 Conference Paper

Bi-Weighting Domain Adaptation for Cross-Language Text Classification

• Chang Wan
• Rong Pan
• Jiefei Li

Text classification is widely used in many real-world applications. To obtain satisfied classification performance, most traditional data mining methods require lots of labeled data, which can be costly in terms of both time and human efforts. In reality, there are plenty of such resources in English since it has the largest population in the Internet world, which is not true in many other languages. In this paper, we present a novel transfer learning approach to tackle the cross-language text classification problems. We first align the feature spaces in both domains utilizing some on-line translation service, which makes the two feature spaces under the same coordinate. Although the feature sets in both domains are the same, the distributions of the instances in both domains are different, which violates the i. i. d. assumption in most traditional machine learning methods. For this issue, we propose an iterative feature and instance weighting (Bi-Weighting) method for domain adaptation. We empirically evaluate the effectiveness and efficiency of our approach. The experimental results show that our approach outperforms some baselines including four transfer learning algorithms.

ICAPS Conference 2011 Conference Paper

Cross-Domain Action-Model Acquisition for Planning via Web Search

• Hankz Hankui Zhuo
• Qiang Yang 0001
• Rong Pan
• Lei Li 0022

Applying learning techniques to acquire action models is an area of intense research interest. Most previous works in this area have assumed that there is a significant amount of training data available in a planning domain of interest, which we call target domain, where action models are to be learned. However, it is often difficult to acquire sufficient training data to ensure that the learned action models are of high quality. In this paper, we develop a novel approach to learning action models with limited training data in the target domain by transferring knowledge from related auxiliary or source domains. We assume that the action models in the source domains have already been created before, and seek to transfer as much of the the available information from the source domains as possible to help our learning task. We first exploit a Web searching method to bridge the target and source domains, such that transferrable knowledge from source domains is identified. We then encode the transferred knowledge together with the available data from the target domain as constraints in a maximum satisfiability problem, and solve these constraints using a weighted MAX-SAT solver. We finally transform the solutions thus obtained into high-quality target-domain action models. We empirically show that our transfer-learning based framework is effective in several domains, including the International Planning Competition (IPC) domains and some synthetic domains.

YNIMG Journal 2009 Journal Article

Automated segmentation of mouse brain images using extended MRF

• Min Hyeok Bae
• Rong Pan
• Teresa Wu
• Alexandra Badea

We introduce an automated segmentation method, extended Markov random field (eMRF), to classify 21 neuroanatomical structures of mouse brain based on three dimensional (3D) magnetic resonance images (MRI). The image data are multispectral: T2-weighted, proton density-weighted, diffusion x, y and z weighted. Earlier research (Ali, A. A. , Dale, A. M. , Badea, A. , Johnson, G. A. , 2005. Automated segmentation of neuroanatomical structures in multispectral MR microscopy of the mouse brain. NeuroImage 27 (2), 425–435) successfully explored the use of MRF for mouse brain segmentation. In this research, we study the use of information generated from support vector machine (SVM) to represent the probabilistic information. Since SVM in general has a stronger discriminative power than the Gaussian likelihood method and is able to handle nonlinear classification problems, integrating SVM into MRF improved the classification accuracy. The eMRF employs the posterior probability distribution obtained from SVM to generate a classification based on the MR intensity. Secondly, the eMRF introduces a new potential function based on location information. Third, to maximize the classification performance, the eMRF uses the contribution weights optimally determined for each of the three potential functions: observation, location and contextual functions, which are traditionally equally weighted. We use the voxel overlap percentage and volume difference percentage to evaluate the accuracy of eMRF segmentation and compare the algorithm with three other segmentation methods — mixed ratio sampling SVM (MRS-SVM), atlas-based segmentation and MRF. Validation using classification accuracy indices between automatically segmented and manually traced data shows that eMRF outperforms other methods.

AIJ Journal 2007 Journal Article

Mining competent case bases for case-based reasoning

• Rong Pan
• Qiang Yang
• Sinno Jialin Pan

Case-based reasoning relies heavily on the availability of a highly competent case base to make high-quality decisions. However, good case bases are difficult to come by. In this paper, we present a novel algorithm for automatically mining a high-quality case base from a raw case set that can preserve and sometimes even improve the competence of case-based reasoning. In this paper, we analyze two major problems in previous case-mining algorithms. The first problem is caused by noisy cases such that the nearest neighbor cases of a problem may not provide correct solutions. The second problem is caused by uneven case distribution, such that similar problems may have dissimilar solutions. To solve these problems, we develop a theoretical framework for the error bound in case-based reasoning, and propose a novel case-base mining algorithm guided by the theoretical results that returns a high-quality case base from raw data efficiently. We support our theory and algorithm with extensive empirical evaluation using different benchmark data sets.

AAAI Conference 2005 Short Paper

A Framework for Bayesian Network Mapping

• Rong Pan

This research is motivated by the need to support inference across multiple intelligence systems involving uncertainty. Our objective is to develop a theoretical framework and related inference methods to map semantically similar variables between separate Bayesian networks in a principled way. The work is to be conducted in two steps. In the first step, we investigate the problem of formalizing the mapping between variables in two separate BNs with different semantics and distributions as pairwise linkages. In the second step, we aim to justify the mapping between networks as a set of selected variable linkages, and then conduct in-ference along it.

AAAI Conference 2005 Conference Paper

Competence Driven Case-Base Mining

• Rong Pan
• Junfeng Pan

We present a novel algorithm for extracting a high-quality case base from raw data while preserving and sometimes improving the competence of case-based reasoning. We extend the framework of Smyth and Keane’s case-deletion policy with two additional features. First, we build a case base using a statistical distribution that is mined from the input data so that the case-base competence can be preserved or even increased for future problems. Second, we introduce a nonlinear transformation of the data set so that the case-base sizes can be further reduced while ensuring that the competence be preserved and even increased. We show that Smyth and Keane’s deletion-based algorithm is sensitive to noisy cases, and that our solution solves this problem more satisfactorily. We show the theoretical foundation and empirical evaluation on several data sets.

AAAI Conference 2005 System Paper

Swoogle: Searching for Knowledge on the Semantic Web

• Tim Finin
• Rong Pan
• Pranam Kolari

The Semantic Web’s distributed nature raises significant data access problems — how can an agent discover, index, search and navigate knowledge on the Semantic Web? Swoogle was developed to facilitate webscale semantic web data access by providing these services to both human and software agents. It focuses on two levels of knowledge granularity: URI based semantic web vocabulary and semantic web documents (SWDs), i.e., RDF and OWL documents encoded in XML, NTriples or N3.

IJCAI Conference 2003 Conference Paper

TAGA: Travel Market Framework in Agentcities

• Youyong Zou
• Tim Finin
• Li Ding
• HARRY CHEN
• Rong Pan