Author name cluster

Haiyan Zhao

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.

11 papers

2 author rows

FM Conference 2026 Conference Paper

Formal Verification of Functional Correctness for the OpenHarmony LiteOS-M Kernel

Tianqi Zhao
Qinxiang Cao
Shenghua Feng
Minghui Zhou
Naijun Zhan
Yongzhi Cao
Junfeng Zhao
Haiyan Zhao

Abstract OpenHarmony LiteOS-M, a preemptive operating system (OS) kernel for the Internet of Things (IoT), is widely deployed in safety-critical domains, such as aerospace and transportation. As a rigorous method to assure software safety, formal verification has been applied to OS kernels in industry. However, entirely verified kernels with large codebases are rare, since such verification is typically performed within interactive theorem provers, requiring substantial human effort. In this paper, we present the functional correctness verification of LiteOS-M. First, to improve verification efficiency, we design a formal verification platform, Smart Verifier. The platform employs an annotation-based verifier as the front end, while the back end integrates Z3 and Rocq, combining automatic and interactive theorem proving techniques. Second, we tailor two verification methods, expressing program refinement as standard Hoare logic triples and modeling concurrency through state transition systems, to utilize the platform for verifying LiteOS-M. Our verified LiteOS-M kernel consists of 17, 000 lines of C. During the code review and verification, we find a total of 17 bugs, all confirmed and fixed by developers.

DOI

AAAI Conference 2025 Conference Paper

A Comprehensive Evaluation on Event Reasoning of Large Language Models

Zhengwei Tao
Zhi Jin
Yifan Zhang
Xiancai Chen
Haiyan Zhao
Jia Li
Bin Liang
Chongyang Tao

Event reasoning is a fundamental ability that underlies many applications. It requires event schema knowledge to perform global reasoning and needs to deal with the diversity of the inter-event relations and the reasoning paradigms. The extent to which LLMs excel in event reasoning across various relations and reasoning paradigms has not been thoroughly investigated. Additionally, it is still unclear whether LLMs utilize event knowledge in the same way humans do. To mitigate this disparity, we comprehensively evaluate the abilities of event reasoning of LLMs on different relations, paradigms, and levels of abstraction. We introduce a novel benchmark EV2 for EValuation of EVent reasoning. EV2 consists of two levels of evaluation on schema and instance and is comprehensive in relations and reasoning paradigms. We conduct extensive experiments on EV2. We find that 1) LLMs have abilities to accomplish event reasoning but their performances are far from satisfactory. 2) There are imbalances of event reasoning abilities on different relations and paradigms. 3) LLMs have event schema knowledge, however, they're not aligned with humans on how to utilize the knowledge. Based on these findings, we guide the LLMs in utilizing the event schema knowledge as memory leading to improvements in event reasoning.

PDF Details DOI

AAAI Conference 2025 Conference Paper

DABL: Detecting Semantic Anomalies in Business Processes Using Large Language Models

Wei Guan
Jian Cao
Jianqi Gao
Haiyan Zhao
Shiyou Qian

Detecting anomalies in business processes is crucial for ensuring operational success. While many existing methods rely on statistical frequency to detect anomalies, it's important to note that infrequent behavior doesn't necessarily imply undesirability. To address this challenge, detecting anomalies from a semantic viewpoint proves to be a more effective approach. However, current semantic anomaly detection methods treat a trace (i.e., process instance) as multiple event pairs, disrupting long-distance dependencies. In this paper, we introduce DABL, a novel approach for detecting semantic anomalies in business processes using large language models (LLMs). We collect 143,137 real-world process models from various domains. By generating normal traces through the playout of these process models and simulating both ordering and exclusion anomalies, we fine-tune Llama 2 using the resulting log. Through extensive experiments, we demonstrate that DABL surpasses existing state-of-the-art semantic anomaly detection methods in terms of both generalization ability and learning of given processes. Users can directly apply DABL to detect semantic anomalies in their own datasets without the need for additional training. Furthermore, DABL offers the ability to interpret anomalies' causes in natural language, providing valuable insights into the detected anomalies.

PDF Details DOI

AIIM Journal 2025 Journal Article

HMEA: A hierarchical medical knowledge graph entity alignment model fusing multi-aspect information

Weiguang Wang
Lijuan Ma
Wei Cai
Haiyan Zhao
Xia Zhang

Medical entity alignment is crucial for the integration and reasoning of medical knowledge, aiming to match semantically equivalent entities across different medical knowledge graphs. Unlike entities in general knowledge graphs, medical entities contain rich multi-aspect information, which not only includes structural and attribute information but also additional information such as ontology and descriptions. However, existing entity alignment methods overlook these additional pieces of information and lack exploration into the fusion of multi-aspect information. This leads to less-than-ideal performance in medical entity alignment. To address the aforementioned issues, in this paper, we propose a hierarchical medical knowledge graph entity alignment method, termed HMEA, which integrates multi-aspect information. Firstly, we represent the medical knowledge graph as a hierarchical heterogeneous graph to model the multi-aspect information of medical entities. Secondly, we design different representation learning methods according to the characteristics of multi-aspect information to obtain vector representations of entities in different dimensions. Subsequently, we devise a two-stage multi-aspect knowledge fusion mechanism to dynamically integrate multi-aspect information, enabling mutual complementarity. Finally, we utilize the fused entity vector representations to guide entity alignment. We compare our approach with state-of-the-art baseline models on ten different types of publicly available datasets and further conduct ablation and parameter analyses. Experimental results validate the effectiveness and robustness of the proposed model. In benchmark tests across all datasets, HMEA outperforms the current state-of-the-art methods significantly.

Details DOI

NeurIPS Conference 2025 Conference Paper

ParamMute: Suppressing Knowledge-Critical FFNs for Faithful Retrieval-Augmented Generation

Pengcheng Huang
Zhenghao Liu
Yukun Yan
Haiyan Zhao
Xiaoyuan Yi
Hao Chen
Zhiyuan Liu
Maosong Sun

Large language models (LLMs) integrated with retrieval-augmented generation (RAG) have improved factuality by grounding outputs in external evidence. However, they remain susceptible to unfaithful generation, where outputs contradict retrieved context despite its relevance and accuracy. Existing approaches aiming to improve faithfulness primarily focus on enhancing the utilization of external context, but often overlook the persistent influence of internal parametric knowledge during generation. In this work, we investigate the internal mechanisms behind unfaithful generation and identify a subset of mid-to-deep feed-forward networks (FFNs) that are disproportionately activated in such cases. Building on this insight, we propose Parametric Knowledge Muting through FFN Suppression (ParamMute), a framework that improves contextual faithfulness by suppressing the activation of unfaithfulness-associated FFNs and calibrating the model toward retrieved knowledge. To evaluate our approach, we introduce CoFaithfulQA, a benchmark specifically designed to evaluate faithfulness in scenarios where internal knowledge conflicts with accurate external evidence. Experimental results show that ParamMute significantly enhances faithfulness across both CoFaithfulQA and the established ConFiQA benchmark, achieving substantial reductions in reliance on parametric memory. These findings underscore the importance of mitigating internal knowledge dominance and provide a new direction for improving LLM trustworthiness in RAG. All codes are available at https: //github. com/OpenBMB/ParamMute.

PDF Details

NeurIPS Conference 2025 Conference Paper

RoMA: Scaling up Mamba-based Foundation Models for Remote Sensing

Fengxiang Wang
Yulin Wang
Mingshuo Chen
Haotian Wang
Hongzhen Wang
Haiyan Zhao
Yangang Sun
Shuo Wang

Recent advances in self-supervised learning for Vision Transformers (ViTs) have fueled breakthroughs in remote sensing (RS) foundation models. However, the quadratic complexity of self-attention poses a significant barrier to scalability, particularly for large models and high-resolution images. While the linear-complexity Mamba architecture offers a promising alternative, existing RS applications of Mamba remain limited to supervised tasks on small, domain-specific datasets. To address these challenges, we propose RoMA, a framework that enables scalable self-supervised pretraining of Mamba-based RS foundation models using large-scale, diverse, unlabeled data. RoMA enhances scalability for high-resolution images through a tailored auto-regressive learning strategy, incorporating two key innovations: 1) a rotation-aware pretraining mechanism combining adaptive cropping with angular embeddings to handle sparsely distributed objects with arbitrary orientations, and 2) multi-scale token prediction objectives that address the extreme variations in object scales inherent to RS imagery. Systematic empirical studies validate that Mamba adheres to RS data and parameter scaling laws, with performance scaling reliably as model and data size increase. Furthermore, experiments across scene classification, object detection, and semantic segmentation tasks demonstrate that RoMA-pretrained Mamba models consistently outperform ViT-based counterparts in both accuracy and computational efficiency. The source code and pretrained models have be released at https: //github. com/MiliLab/RoMA.

PDF Details

EAAI Journal 2024 Journal Article

A survey on deep reinforcement learning approaches for traffic signal control

Haiyan Zhao
Chengcheng Dong
Jian Cao
Qingkui Chen

In the domain of complex urban traffic networks, real-time Traffic Signal Control (TSC) serves as a pivotal strategy for mitigating congestion. Coordinating signal control across multiple intersections involves considerable complexity. Deep Reinforcement Learning (DRL) has emerged as a robust solution. In recent years, there has been rapid advancement in TSC methods, with numerous researchers employing various novel DRL methodologies. Yet, existing surveys lack timeliness and universality in capturing the latest research. There is a notable gap in current research surveys with respect to the latest developments in TSC. Therefore, the focus of this paper lies in analyzing the most recent papers from the past five years, with the aim to provide a comprehensive and multi-dimensional review of the evolution of DRL in TSC. The survey categorizes current research based on model setups, utilized algorithms, and application scenarios. Finally, this paper highlights potential directions for future TSC research.

Details DOI

TIST Journal 2024 Journal Article

Explainability for Large Language Models: A Survey

Haiyan Zhao
Hanjie Chen
Fan Yang
Ninghao Liu
Huiqi Deng
Hengyi Cai
Shuaiqiang Wang
Dawei Yin

Large language models (LLMs) have demonstrated impressive capabilities in natural language processing. However, their internal mechanisms are still unclear and this lack of transparency poses unwanted risks for downstream applications. Therefore, understanding and explaining these models is crucial for elucidating their behaviors, limitations, and social impacts. In this article, we introduce a taxonomy of explainability techniques and provide a structured overview of methods for explaining Transformer-based language models. We categorize techniques based on the training paradigms of LLMs: traditional fine-tuning-based paradigm and prompting-based paradigm. For each paradigm, we summarize the goals and dominant approaches for generating local explanations of individual predictions and global explanations of overall model knowledge. We also discuss metrics for evaluating generated explanations and discuss how explanations can be leveraged to debug models and improve performance. Lastly, we examine key challenges and emerging opportunities for explanation techniques in the era of LLMs in comparison to conventional deep learning models.

Details DOI

ICML Conference 2023 Conference Paper

Does Continual Learning Equally Forget All Parameters?

Haiyan Zhao
Tianyi Zhou 0001
Guodong Long
Jing Jiang 0002
Chengqi Zhang

Distribution shift (e. g. , task or domain shift) in continual learning (CL) usually results in catastrophic forgetting of previously learned knowledge. Although it can be alleviated by repeatedly replaying buffered data, the every-step replay is time-consuming. In this paper, we study which modules in neural networks are more prone to forgetting by investigating their training dynamics during CL. Our proposed metrics show that only a few modules are more task-specific and sensitive to task change, while others can be shared across tasks as common knowledge. Hence, we attribute forgetting mainly to the former and find that finetuning them only on a small buffer at the end of any CL method can bring non-trivial improvement. Due to the small number of finetuned parameters, such ”Forgetting Prioritized Finetuning (FPF)” is efficient in computation. We further propose a more efficient and simpler method that entirely removes the every-step replay and replaces them by only $k$-times of FPF periodically triggered during CL. Surprisingly, this ”$k$-FPF” performs comparably to FPF and outperforms the SOTA CL methods but significantly reduces their computational overhead and cost. In experiments on several benchmarks of class- and domain-incremental CL, FPF consistently improves existing CL methods by a large margin, and $k$-FPF further excels in efficiency without degrading the accuracy. We also empirically studied the impact of buffer size, epochs per task, and finetuning modules on the cost and accuracy of our methods.

Details

TMLR Journal 2022 Journal Article

Extracting Local Reasoning Chains of Deep Neural Networks

Haiyan Zhao
Tianyi Zhou
Guodong Long
Jing Jiang
Chengqi Zhang

We study how to explain the main steps of inference that a pre-trained deep neural net (DNN) relies on to produce predictions for a (sub)task and its data. This problem is related to network pruning and interpretable machine learning with the following highlighted differences: (1) fine-tuning of any neurons/filters is forbidden; (2) we target a very high pruning rate, e.g., ≥ 95%, for better interpretability; (3) the interpretation is for the whole inference process on a few data of a task rather than for individual neurons/filters or a single sample. In this paper, we introduce NeuroChains to extract the local inference chains by optimizing differentiable sparse scores for the filters and layers, which reflects their importance in preserving the outputs on a few data drawn from a given (sub)task. Thereby, NeuroChains can extract an extremely small sub-network composed of critical filters exactly copied from the original pre-trained DNN by removing the filters/layers with small scores. For samples from the same class, we can then visualize the inference pathway in the pre-trained DNN by applying existing interpretation techniques to the retained filters and layers. It reveals how the inference process stitches and integrates the information layer by layer and filter by filter. We provide detailed and insightful case studies together with several quantitative analyses over thousands of trials to demonstrate the quality, sparsity, fidelity and accuracy of the interpretation. In extensive empirical studies on VGG, ResNet, and ViT, NeuroChains significantly enriches the interpretation and makes the inner mechanism of DNNs more transparent.

PDF Details

IS Journal 2018 Journal Article

Neural Cross-Session Filtering: Next-Item Prediction Under Intra- and Inter-Session Context

Liang Hu
Qingkui Chen
Haiyan Zhao
Songlei Jian
Longbing Cao
Jian Cao

Classic recommender systems (RSs) often repeatedly recommend similar items to user historical profiles or recent purchases. For this, session-based RSs (SBRSs) are extensively studied in recent years. Current SBRSs often assume a rigid-order sequence, which does not fit in many real-world cases. In fact, the next-item recommendation depends on not only current session context but also historical sessions which are often neglected by current SBRSs. Accordingly, an SBRS over relaxed-order sequences with both intra- and inter-context is more pragmatic. Inspired by the successful experience in modern language modeling, we design an efficient neural architecture to model both intra- and inter-context for next item prediction.

Details DOI