Author name cluster

Sheng Zhong

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.

14 papers

2 author rows

AAAI Conference 2026 Conference Paper

Diverse Human Driving Vehicle Simulation in Background Traffic for Autonomous Driving Tests

Wendi Li
Hao Wu
Han Gao
Bing Mao
Fengyuan Xu
Sheng Zhong

Realistic background traffic is critical to the simulation platforms for autonomous driving (AD) testing. Given that most vehicles in reality are driven by human beings, introducing human driving (HD) vehicles to the background traffic is necessary to be able to discover more problems of the tested AD vehicle in the simulation stage. However, existing methods rely on ad-hoc rules or data-driven training to mimic partial human driver behaviors, which are not comprehensive and lack transparency. In this work, we design a smart human driving vehicle simulator HDSim which is empowered by cognitively inspired modeling and AI models. HDSim enables diverse, realistic, and scalable HD traffic simulation on AD testing platforms like CARLA in a non-intrusive manner. There are two novel components in HDSim. First, we introduce a driver model to guide the generation of diverse human driving styles by using different combinations of latent cognitive factors in a hierarchy. Second, we design a Perception-Mediated Behavior Influence (PMBI) mechanism to use LLM-assisted perceptual transformations to indirectly fuse driving actions with driving styles. Experiments show that HDSim traffic can help simulation platforms like CARLA to reveal 68% more failures of tested AD vehicles, and the explainability of reported accidents is also improved.

PDF Details DOI

AAAI Conference 2026 Conference Paper

TriFusion-IDS: A Multimodal Graph-Tabular-Text Contrastive Framework for Cross-Dataset Intrusion Detection

Qinxin Zhao
Sheng Zhong

Traditional Intrusion Detection Systems (IDS) are typically trained in specific network environments, and their performance often degrades significantly when deployed in new environments with different attack categories. To address this challenge, we propose and define the task of cross-dataset intrusion detection and design a novel multimodal contrastive learning framework named TriFusion-IDS. This framework represents network traffic from three complementary dimensions: a graph view to capture structural communication patterns, a tabular view to model statistical features, and a textual view to define the semantics of attacks. TriFusion-IDS fuses the graph and tabular representations and aligns them with textual descriptions in a shared embedding space using a CLIP-style contrastive loss function. This semantics-based alignment mechanism enables the model to overcome the effects of zero-shot categories and thus generalize to new network environments. Our extensive experiments on several mainstream datasets demonstrate that this method significantly outperforms existing baselines in cross-dataset intrusion detection scenarios.

PDF Details DOI

AAAI Conference 2025 Conference Paper

DriveEditor: A Unified 3D Information-Guided Framework for Controllable Object Editing in Driving Scenes

Yiyuan Liang
Zhiying Yan
Liqun Chen
Jiahuan Zhou
Luxin Yan
Sheng Zhong
Xu Zou

Vision-centric autonomous driving systems require diverse data for robust training and evaluation, which can be augmented by manipulating object positions and appearances within existing scene captures. While recent advancements in diffusion models have shown promise in video editing, their application to object manipulation in driving scenarios remains challenging due to imprecise positional control and difficulties in preserving high-fidelity object appearances. To address these challenges in position and appearance control, we introduce DriveEditor, the first diffusion-based framework for object editing in driving videos. DriveEditor offers a unified framework for comprehensive object editing operations, including repositioning, replacement, deletion, and insertion. These diverse manipulations are all achieved through a shared set of varying inputs, processed by identical position control and appearance maintenance modules. The position control module projects the given 3D bounding box while preserving depth information and hierarchically injects it into the diffusion process, enabling precise control over object position and orientation. The appearance maintenance module preserves consistent attributes with a single reference image by employing a three-tiered approach: low-level detail preservation, high-level semantic maintenance, and the integration of 3D priors from a novel view synthesis model. Extensive qualitative and quantitative evaluations on the nuScenes dataset demonstrate DriveEditor's exceptional fidelity and controllability in generating diverse driving scene edits, as well as its remarkable ability to facilitate downstream tasks.

PDF Details DOI

AAAI Conference 2025 Conference Paper

Iterative Self-Training with Class-Aware Text-to-Image Synthesis for Visual Task Learning

Xiang Zhang
Wanqing Zhao
Pengyang Li
Ying Liu
Hangzai Luo
Sheng Zhong
Jinye Peng
Jianping Fan

Generative models are widely used to produce synthetic images with annotations, alleviating the burden of image collection and annotation for training deep visual models. However, challenges such as limited image diversity, noisy pseudo labels, and domain gaps between synthetic and real images often undermine their effectiveness in downstream visual tasks. This paper introduces the Iterative Self-Training with Class-Aware Text-to-Image Synthesis (IST-CATS) framework, which addresses these challenges by integrating a class-aware text-to-image synthesis (CATS) component with an iterative self-training (IST) strategy. CATS innovatively introduces a class-aware chain approach to generate detailed descriptions. These descriptions act as prompts for a diffusion model, enabling the creation of a diverse of images accompanied by distinguishable objects against the background. The generated images can be easily pseudo-labeled by an unsupervised instance segmentation method, and then noisy pseudo labels can be effectively purified by a novel feature similarity-based filtering mechanism. The generated images underpin our IST, which progressively enhances vision models and refines pseudo labels through self-training and our proposed label filtering strategy (LabFilt). LabFilt meticulously improves the quality of pseudo labels by employing class-adaptive techniques at both the pixel and object levels, ensuring refined pseudo-label accuracy. IST-CATS demonstrates superior performance in object detection and semantic segmentation compared to traditional synthetic and semi/weakly-supervised methods, effectively addressing data collection and annotation challenges.

PDF Details DOI

IROS Conference 2025 Conference Paper

MemGS: Memory-Efficient Gaussian Splatting for Real-Time SLAM

Yinlong Bai
Hongxin Zhang
Sheng Zhong
Junkai Niu
Hai Li
Yijia He
Yi Zhou

Recent advancements in 3D Gaussian Splatting (3DGS) have made a significant impact on rendering and reconstruction techniques. Current research predominantly focuses on improving rendering performance and reconstruction quality using high-performance desktop GPUs, largely overlooking applications for embedded platforms like micro air vehicles (MAVs). These devices, with their limited computational resources and memory, often face a trade-off between system performance and reconstruction quality. In this paper, we improve existing methods in terms of GPU memory usage while enhancing rendering quality. Specifically, to address redundant 3D Gaussian primitives in SLAM, we propose merging them in voxel space based on geometric similarity. This reduces GPU memory usage without impacting system runtime performance. Furthermore, rendering quality is improved by initializing 3D Gaussian primitives via Patch-Grid (PG) point sampling, enabling more accurate modeling of the entire scene. Quantitative and qualitative evaluations on publicly available datasets demonstrate the effectiveness of our improvements.

Details

IJCAI Conference 2025 Conference Paper

SAP: Privacy-Preserving Fine-Tuning on Language Models with Split-and-Privatize Framework

Xicong Shen
Yang Liu
Yi Liu
Peiran Wang
Huiqi Liu
Jue Hong
Bing Duan
Zirui Huang

Pre-trained Language Models (PLM) have enabled a cost-effective approach to handling various downstream applications via Parameter-Efficient-Fine-Tuning (PEFT) techniques. In this context, service providers have introduced a popular fine-tuning-based product service known as Model-as-a-Service (MaaS). This service offers users access to extensive PLMs and training resources. With MaaS, users can fine-tune, deploy, and utilize their customized models seamlessly, leveraging a one-stop platform that allows them to work with their private datasets efficiently. However, this service paradigm has recently been exposed to the possibility of leaking user private data. To this end, we identify the data privacy leakage risks in MaaS-based PEFT and propose a Split-and-Privatize (SAP) framework, mitigating the privacy leakage by integrating split learning and differential privacy into MaaS PEFT. Furthermore, we propose Contributing-Token-Identification (CTI), a novel method to balance model utility degradation and privacy leakage. As a result, the proposed framework is comprehensively evaluated, demonstrating a 65% improvement in empirical privacy with only a 1% degradation in model performance on the Stanford Sentiment Treebank dataset, outperforming existing state-of-the-art baselines.

PDF Details DOI

ICRA Conference 2024 Conference Paper

IMU-Aided Event-based Stereo Visual Odometry

Junkai Niu
Sheng Zhong
Yi Zhou 0010

Direct methods for event-based visual odometry solve the mapping and camera pose tracking sub-problems by establishing implicit data association in a way that the generative model of events is exploited. The main bottlenecks faced by state-of-the-art work in this field include the high computational complexity of mapping and the limited accuracy of tracking. In this paper, we improve our previous direct pipeline Event-based Stereo Visual Odometry in terms of accuracy and efficiency. To speed up the mapping operation, we propose an efficient strategy of edge-pixel sampling according to the local dynamics of events. The mapping performance in terms of completeness and local smoothness is also improved by combining the temporal stereo results and the static stereo results. To circumvent the degeneracy issue of camera pose tracking in recovering the yaw component of general 6-DoF motion, we introduce as a prior the gyroscope measurements via pre-integration. Experiments on publicly available datasets justify our improvement. We release our pipeline as an open-source software for future research in this field.

Details

AAAI Conference 2024 Conference Paper

Make Lossy Compression Meaningful for Low-Light Images

Shilv Cai
Liqun Chen
Sheng Zhong
Luxin Yan
Jiahuan Zhou
Xu Zou

Low-light images frequently occur due to unavoidable environmental influences or technical limitations, such as insufficient lighting or limited exposure time. To achieve better visibility for visual perception, low-light image enhancement is usually adopted. Besides, lossy image compression is vital for meeting the requirements of storage and transmission in computer vision applications. To touch the above two practical demands, current solutions can be categorized into two sequential manners: ``Compress before Enhance (CbE)'' or ``Enhance before Compress (EbC)''. However, both of them are not suitable since: (1) Error accumulation in the individual models plagues sequential solutions. Especially, once low-light images are compressed by existing general lossy image compression approaches, useful information (e.g., texture details) would be lost resulting in a dramatic performance decrease in low-light image enhancement. (2) Due to the intermediate process, the sequential solution introduces an additional burden resulting in low efficiency. We propose a novel joint solution to simultaneously achieve a high compression rate and good enhancement performance for low-light images with much lower computational cost and fewer model parameters. We design an end-to-end trainable architecture, which includes the main enhancement branch and the signal-to-noise ratio (SNR) aware branch. Experimental results show that our proposed joint solution achieves a significant improvement over different combinations of existing state-of-the-art sequential ``Compress before Enhance'' or ``Enhance before Compress'' solutions for low-light images, which would make lossy low-light image compression more meaningful. The project is publicly available at: https://github.com/CaiShilv/Joint-IC-LL.

PDF Details DOI

NeurIPS Conference 2024 Conference Paper

Molecule Design by Latent Prompt Transformer

Deqian Kong
Yuhao Huang
Jianwen Xie
Edouardo Honig
Ming Xu
Shuanghong Xue
Pei Lin
Sanping Zhou

This work explores the challenging problem of molecule design by framing it as a conditional generative modeling task, where target biological properties or desired chemical constraints serve as conditioning variables. We propose the Latent Prompt Transformer (LPT), a novel generative model comprising three components: (1) a latent vector with a learnable prior distribution modeled by a neural transformation of Gaussian white noise; (2) a molecule generation model based on a causal Transformer, which uses the latent vector as a prompt; and (3) a property prediction model that predicts a molecule's target properties and/or constraint values using the latent prompt. LPT can be learned by maximum likelihood estimation on molecule-property pairs. During property optimization, the latent prompt is inferred from target properties and constraints through posterior sampling and then used to guide the autoregressive molecule generation. After initial training on existing molecules and their properties, we adopt an online learning algorithm to progressively shift the model distribution towards regions that support desired target properties. Experiments demonstrate that LPT not only effectively discovers useful molecules across single-objective, multi-objective, and structure-constrained optimization tasks, but also exhibits strong sample efficiency.

PDF Details DOI

AAAI Conference 2022 Conference Paper

Improving Human-Object Interaction Detection via Phrase Learning and Label Composition

Zhimin Li
Cheng Zou
Yu Zhao
Boxun Li
Sheng Zhong

Human-Object Interaction (HOI) detection is a fundamental task in high-level human-centric scene understanding. We propose PhraseHOI, containing a HOI branch and a novel phrase branch, to leverage language prior and improve relation expression. Specifically, the phrase branch is supervised by semantic embeddings, whose ground truths are automatically converted from the original HOI annotations without extra human efforts. Meanwhile, a novel label composition method is proposed to deal with the long-tailed problem in HOI, which composites novel phrase labels by semantic neighbors. Further, to optimize the phrase branch, a loss composed of a distilling loss and a balanced triplet loss is proposed. Extensive experiments are conducted to prove the effectiveness of the proposed PhraseHOI, which achieves significant improvement over the baseline and surpasses previous state-of-the-art methods on Full and NonRare on the challenging HICO-DET benchmark.

PDF Details

AAMAS Conference 2017 Conference Paper

On Repeated Stackelberg Security Game with the Cooperative Human Behavior Modelfor Wildlife Protection

Binru Wang
Yuan Zhang
Sheng Zhong

Inspired by successful deployments of Stackelberg Security Game in real life, researchers are working hard to optimize the game models to make them more practical. Recent security game work on wildlife protection makes a step forward by taking the possible cooperation among attackers into consideration. However, it models attackers to have complete rationality, which is not always possible in practice given they are human beings. We aim to tackle attackers’ bounded rationality in the complicated, cooperationenabled and multi-round security game for wildlife protection. Specifically, we construct a repeated Stackelberg game, and propose a novel adaptive human behavior model for attackers based on it. Despite generating defender’s optimal strategy requires to solve a non-linear and non-convex optimization problem, we are able to propose an efficient algorithm that approximately solve this problem. We perform extensive real-life experiments, and results show our solution effectively helps the defender to deal with attackers who might cooperate.

PDF

IJCAI Conference 2015 Conference Paper

Differentially Private Matrix Factorization

Jingyu Hua
Chang Xia
Sheng Zhong

Matrix factorization (MF) is a prevailing collaborative filtering method for building recommender systems. It requires users to upload their personal preferences to the recommender for performing MF, which raises serious privacy concerns. This paper proposes a differentially private MF mechanism that can prevent an untrusted recommender from learning any users’ ratings or profiles. Our design decouples computations upon users’ private data from the recommender to users, and makes the recommender aggregate local results in a privacypreserving way. It uses the objective perturbation to make sure that the final item profiles satisfy differential privacy and solves the challenge to decompose the noise component for objective perturbation into small pieces that can be determined locally and independently by users. We also propose a third-party based mechanism to reduce noises added in each iteration and adapt our online algorithm to the dynamic setting that allows users to leave and join. The experiments show that our proposal is efficient and introduces acceptable side effects on the precision of results.

PDF Details

AIIM Journal 2007 Journal Article

Gene Ontology analysis in multiple gene clusters under multiple hypothesis testing framework

Sheng Zhong
Dan Xie

Objective Gene Ontology (GO) has become a routine resource for functional analysis of gene lists. Although a number of tools have been provided to identify enriched GO terms in one or two gene lists, two technical challenges remain. First, how to handle multiple hypothesis testing in the analysis given that the tests are heavily correlated; second, how to identify GO terms that are enriched in a gene cluster, as compared to multiple other gene clusters. We provide a statistical procedure to rigorously treat these problems and offer a software tool for applying GO to the analysis of gene clusters. Methods We previously introduced a statistical procedure that handles hypothesis testing in a two-group comparison scenario. In this paper we extend the two-group comparison procedure into a general procedure that enables the analysis of any number of gene lists/clusters. This new procedure enables identification of GO terms enriched in any gene cluster, while it controls for multiple hypothesis testing. This procedure is implemented into a user-friendly analysis tool: GoSurfer. The current version of GoSurfer takes one or several gene lists as input, and it identifies the GO terms that are enriched in any of the input gene lists. GoSurfer estimates a conservative false discovery rate (FDR) for every GO term. The FDR estimation procedure in GoSurfer has two advantages: it does not rely on independence assumption, and it does not assume all the hypotheses are null hypothesis (complete null). Thus GoSurfer's FDR estimates are mildly conservative rather than overly conservative. Results We implemented the new procedure for GO analysis in multiple gene clusters into the GoSurfer software. We provide three examples on using GoSurfer to analyze time course gene expression data sets on the differentiation of embryonic stem cells. In the example of analysis of multiple gene clusters, we first used a typical clustering algorithm and identified five gene clusters, representing up-regulation, down-regulation and other patterns in the differentiation time course. Taking all the five gene clusters as input data, GoSurfer reports “cell adhesion” and “muscle contraction” as significant GO terms for the up-regulated cluster, “amino acids metabolism” as a significant GO term for the down-regulated gene cluster, and GoSurfer reports a number of GO terms related to RNA processing and RNA transport as significant terms to a cluster that is up-regulated in both early and late time points. This may suggest that genes for RNA processing and genes for RNA transport are coregulated in the differentiation process of embryonic stem cells. Conclusion The GoSurfer software is provided to analyze multiple gene clusters and identify GO terms that are enriched in any gene cluster. Gosurfer is available at: www. gosurfer. org.

Details DOI

TCS Journal 2007 Journal Article

Towards a theory of data entanglement

James Aspnes
Joan Feigenbaum
Aleksandr Yampolskiy
Sheng Zhong

We give a formal model for systems that store data in entangled form. We propose a new notion of entanglement, called all-or-nothing integrity (AONI) that binds the users’ data in a way that makes it hard to corrupt the data of any one user without corrupting the data of all users. AONI can be a useful defense against negligent or dishonest storage providers who might otherwise be tempted to discard documents belonging to users without much clout. We show that, if all users use a fixed standard recovery algorithm, we can implement AONI using a MAC, but, if some of the users adopt instead a non-standard recovery algorithm provided by the dishonest storage provider, AONI can no longer be achieved. However, even for the latter scenario, we describe a simple entangling mechanism that provides AONI for a restricted class of destructive adversaries.

Details DOI