Jie An

FM Conference 2026 Conference Paper

Exact Moment Estimation of Stochastic Differential Dynamics

• Shenghua Feng
• Jie An
• Naijun Zhan
• Fanjiang Xu

Abstract Moment estimation for stochastic differential equations (SDEs) is fundamental to the formal reasoning and verification of stochastic dynamical systems, yet remains challenging and is rarely available in closed form. In this paper, we study time-homogeneous SDEs with polynomial drift and diffusion, and investigate when their moments can be computed exactly. We formalize the notion of moment-solvable SDEs and propose a generic symbolic procedure that, for a given monomial, attempts to construct a finite-dimensional linear ordinary differential equation (ODE) system governing its moment, thereby enabling exact computation. We introduce a syntactic class of pro-solvable SDEs, characterized by a block-triangular structure, and prove that all polynomial moments of any pro-solvable SDE admit such finite ODE representations. This class strictly generalizes linear SDEs and includes many nonlinear models. Experimental results demonstrate the effectiveness of our approach.

TACAS Conference 2026 Conference Paper

Quantifier Elimination Meets Treewidth

• Hao Wu
• Jiyu Zhu
• Amir Kafshdar Goharshady
• Jie An
• Bican Xia
• Naijun Zhan

AAAI Conference 2026 Conference Paper

RESTL: Reinforcement Learning Guided by Multi-Aspect Rewards for Signal Temporal Logic Transformation

• Yue Fang
• Zhi Jin
• Jie An
• Hongshen Chen
• Xiaohong Chen
• Naijun Zhan

Signal Temporal Logic (STL) is a powerful formal language for specifying real-time specifications of Cyber-Physical Systems (CPS). Transforming specifications written in natural language into STL formulas automatically has attracted increasing attention. Existing rule-based methods depend heavily on rigid pattern matching and domain-specific knowledge, limiting their generalizability and scalability. Recently, Supervised Fine-Tuning (SFT) of large language models (LLMs) has been successfully applied to transform natural language into STL. However, the lack of fine-grained supervision on atomic proposition correctness, semantic fidelity, and formula readability often leads SFT-based methods to produce formulas misaligned with the intended meaning. To address these issues, we propose RESTL, a reinforcement learning (RL)-based framework for the transformation from natural language to STL. RESTL introduces multiple independently trained reward models that provide fine-grained, multi-faceted feedback from four perspectives, i.e., atomic proposition consistency, semantic alignment, formula succinctness, and symbol matching. These reward models are trained with a curriculum learning strategy to improve their feedback accuracy, and their outputs are aggregated into a unified signal that guides the optimization of the STL generator via Proximal Policy Optimization (PPO). Experimental results demonstrate that RESTL significantly outperforms state-of-the-art methods in both automatic metrics and human evaluations.

AAAI Conference 2026 Conference Paper

Runtime Safety and Reach-avoid Prediction of Stochastic Systems via Observation-aware Barrier Functions

• Shenghua Feng
• Jie An
• Fanjiang Xu

Stochastic dynamical systems have emerged as fundamental models across numerous application domains, providing powerful mathematical representations for capturing uncertain system behavior. In this paper, we address the problem of runtime safety and reach-avoid probability prediction for discrete-time stochastic systems with online observations, i.e., estimating the probability that the system satisfies a given safety or reach-avoid specification. Unlike traditional approaches that rely solely on offline models, we propose a framework that incorporates real-time observations to dynamically refine probability estimates for safety and reach-avoid events. By introducing observation-aware barrier functions, our method adaptively updates probability bounds as new observations are collected, combining efficient offline computation with online backward iteration. This approach enables rigorous and responsive prediction of safety and reach-avoid probabilities under uncertainty. In addition to the theoretical guarantees, experimental results on benchmark systems demonstrate the practical effectiveness of the proposed method.

FM Conference 2026 Conference Paper

STLts-Div: Diversified Trace Synthesis from STL Specifications Using MILP

• Martin Jouve-Genty
• Han Su
• Sota Sato
• Jie An
• Zhenya Zhang
• Ichiro Hasuo

Abstract Modern cyber-physical systems are complex, and requirements are often written in Signal Temporal Logic (STL). Writing the right STL is difficult in practice; engineers benefit from concrete executions that illustrate what a specification actually admits. Trace synthesis addresses this need, but a single witness rarely suffices to understand intent or explore edge cases—diverse satisfying behaviors are far more informative. We introduce diversified trace synthesis: the automatic generation of sets of behaviorally diverse traces that satisfy a given STL formula. Building on a MILP encoding of STL and system model, we formalize three complementary diversification objectives—Boolean distance, random Boolean distance, and value distance—all captured by an objective function and solved iteratively. We implement these ideas in STLts-Div, a lightweight Python tool that integrates with Gurobi.

NeurIPS Conference 2025 Conference Paper

On Inductive Biases That Enable Generalization in Diffusion Transformers

• Jie An
• De Wang
• Pengsheng Guo
• Jiebo Luo
• Alex Schwing

Recent work studying the generalization of diffusion models with locally linear UNet-based denoisers reveals inductive biases that can be expressed via geometry-adaptive harmonic bases. For such locally linear UNets, these geometry-adaptive harmonic bases can be conveniently visualized through the eigen-decomposition of a UNet’s Jacobian matrix. In practice, however, more recent denoising networks are often transformer-based, e. g. , the diffusion transformer (DiT). Due to the presence of nonlinear operations, similar eigen-decomposition analyses cannot be used to reveal the inductive biases of transformer-based denoisers. This motivates our search for alternative ways to explain the strong generalization ability observed in DiT models. Investigating a DiT’s pivotal attention modules, we find that locality of attention maps in a DiT’s early layers are closely associated with generalization. To verify this finding, we modify the generalization of a DiT by restricting its attention windows and observe an improvement in generalization. Furthermore, we empirically find that both the placement and the effective attention size of these local attention windows are crucial factors. Experimental results on the CelebA, ImageNet, MSCOCO, and LSUN data show that strengthening the inductive bias of a DiT can improve both generalization and generation quality when less training data is available. Source code is available at https: //github. com/DiT-Generalization/DiT-Generalization.

AAAI Conference 2025 Conference Paper

Ouroboros-Diffusion: Exploring Consistent Content Generation in Tuning-free Long Video Diffusion

• Jingyuan Chen
• Fuchen Long
• Jie An
• Zhaofan Qiu
• Ting Yao
• Jiebo Luo
• Tao Mei

The first-in-first-out (FIFO) video diffusion, built on a pre-trained text-to-video model, has recently emerged as an effective approach for tuning-free long video generation. This technique maintains a queue of video frames with progressively increasing noise, continuously producing clean frames at the queue's head while Gaussian noise is enqueued at the tail. However, FIFO-Diffusion often struggles to keep long-range temporal consistency in the generated videos due to the lack of correspondence modeling across frames. In this paper, we propose Ouroboros-Diffusion, a novel video denoising framework designed to enhance structural and content (subject) consistency, enabling the generation of consistent videos of arbitrary length. Specifically, we introduce a new latent sampling technique at the queue tail to improve structural consistency, ensuring perceptually smooth transitions among frames. To enhance subject consistency, we devise a Subject-Aware Cross-Frame Attention (SACFA) mechanism, which aligns subjects across frames within short segments to achieve better visual coherence. Furthermore, we introduce self-recurrent guidance. This technique leverages information from all previous cleaner frames at the front of the queue to guide the denoising of noisier frames at the end, fostering rich and contextual global information interaction. Extensive experiments of long video generation on the VBench benchmark demonstrate the superiority of our Ouroboros-Diffusion, particularly in terms of subject consistency, motion smoothness, and temporal consistency.

IJCAI Conference 2024 Conference Paper

Bring Metric Functions into Diffusion Models

• Jie An
• Zhengyuan Yang
• Jianfeng Wang
• Linjie Li
• Zicheng Liu
• Lijuan Wang
• Jiebo Luo

We introduce a Cascaded Diffusion Model (Cas-DM) that improves a Denoising Diffusion Probabilistic Model (DDPM) by effectively incorporating additional metric functions in training. Metric functions such as the LPIPS loss have been proven highly effective in consistency models derived from the score matching. However, for the diffusion counterparts, the methodology and efficacy of adding extra metric functions remain unclear. One major challenge is the mismatch between the noise predicted by a DDPM at each step and the desired clean image that the metric function works well on. To address this problem, we propose Cas-DM, a network architecture that cascades two network modules to effectively apply metric functions to the diffusion model training. The first module, similar to a standard DDPM, learns to predict the added noise and is unaffected by the metric function. The second cascaded module learns to predict the clean image, thereby facilitating the metric function computation. Experiment results show that the proposed diffusion model backbone enables the effective use of the LPIPS loss, improving the image quality (FID, sFID) of diffusion models on various established benchmarks.

AAAI Conference 2020 Conference Paper

Ultrafast Photorealistic Style Transfer via Neural Architecture Search

• Jie An
• Haoyi Xiong
• Jun Huan
• Jiebo Luo

The key challenge in photorealistic style transfer is that an algorithm should faithfully transfer the style of a reference photo to a content photo while the generated image should look like one captured by a camera. Although several photorealistic style transfer algorithms have been proposed, they need to rely on post- and/or pre-processing to make the generated images look photorealistic. If we disable the additional processing, these algorithms would fail to produce plausible photorealistic stylization in terms of detail preservation and photorealism. In this work, we propose an effective solution to these issues. Our method consists of a construction step (C-step) to build a photorealistic stylization network and a pruning step (P-step) for acceleration. In the C-step, we propose a dense auto-encoder named PhotoNet based on a carefully designed pre-analysis. PhotoNet integrates a feature aggregation module (BFA) and instance normalized skip links (INSL). To generate faithful stylization, we introduce multiple style transfer modules in the decoder and INSLs. PhotoNet significantly outperforms existing algorithms in terms of both efficiency and effectiveness. In the P-step, we adopt a neural architecture search method to accelerate PhotoNet. We propose an automatic network pruning framework in the manner of teacher-student learning for photorealistic stylization. The network architecture named PhotoNAS resulted from the search achieves significant acceleration over PhotoNet while keeping the stylization effects almost intact. We conduct extensive experiments on both image and video transfer. The results show that our method can produce favorable results while achieving 20-30 times acceleration in comparison with the existing state-of-the-art approaches. It is worth noting that the proposed algorithm accomplishes better performance without any pre- or post-processing.

YNICL Journal 2015 Journal Article

Multivariate pattern analysis reveals anatomical connectivity differences between the left and right mesial temporal lobe epilepsy

• Peng Fang
• Jie An
• Ling-Li Zeng
• Hui Shen
• Fanglin Chen
• Wensheng Wang
• Shijun Qiu
• Dewen Hu

Previous studies have demonstrated differences of clinical signs and functional brain network organizations between the left and right mesial temporal lobe epilepsy (mTLE), but the anatomical connectivity differences underlying functional variance between the left and right mTLE remain uncharacterized. We examined 43 (22 left, 21 right) mTLE patients with hippocampal sclerosis and 39 healthy controls using diffusion tensor imaging. After the whole-brain anatomical networks were constructed for each subject, multivariate pattern analysis was applied to classify the left mTLE from the right mTLE and extract the anatomical connectivity differences between the left and right mTLE patients. The classification results reveal 93.0% accuracy for the left mTLE versus the right mTLE, 93.4% accuracy for the left mTLE versus controls and 90.0% accuracy for the right mTLE versus controls. Compared with the right mTLE, the left mTLE exhibited a different connectivity pattern in the cortical-limbic network and cerebellum. The majority of the most discriminating anatomical connections were located within or across the cortical-limbic network and cerebellum, thereby indicating that these disease-related anatomical network alterations may give rise to a portion of the complex of emotional and memory deficit between the left and right mTLE. Moreover, the orbitofrontal gyrus, cingulate cortex, hippocampus and parahippocampal gyrus, which exhibit high discriminative power in classification, may play critical roles in the pathophysiology of mTLE. The current study demonstrated that anatomical connectivity differences between the left mTLE and the right mTLE may have the potential to serve as a neuroimaging biomarker to guide personalized diagnosis of the left and right mTLE.