Yun Peng

Jun Gao, Yun Peng, Qian Qiao et al.

Despite strong performance on code generation tasks, it remains unclear whether large language models (LLMs) genuinely reason about code execution. Existing code reasoning benchmarks primarily evaluate final output correctness under a single canonical implementation, leaving two critical aspects underexplored: (1) whether LLMs can maintain consistency to functionally equivalent implementations, and (2) whether LLMs can accurately reason about intermediate execution states. We introduce \textbf{CoRE}, a \textbf{Co}de \textbf{Re}asoning benchmark that evaluates code reasoning through \textbf{implementation invariance} and \textbf{process transparency}. Extensive evaluations on eight frontier LLMs reveal two fundamental limitations. First, models exhibit a substantial \textbf{robustness gap}, with performance varying significantly across equivalent implementations. Second, we observe \textbf{superficial execution}, where models arrive at correct final outputs without correctly reasoning about intermediate execution states. Together, these findings demonstrate that output-only evaluations are insufficient for assessing code reasoning and position CoRE as a necessary benchmark for evaluating robust and faithful code reasoning.\footnote{Data and code are available at https://github.com/ZJUSig/CoRE.}

Chaozheng Wang, Yuanhang Yang, Cuiyun Gao et al.

Pre-trained models have been shown effective in many code intelligence tasks. These models are pre-trained on large-scale unlabeled corpus and then fine-tuned in downstream tasks. However, as the inputs to pre-training and downstream tasks are in different forms, it is hard to fully explore the knowledge of pre-trained models. Besides, the performance of fine-tuning strongly relies on the amount of downstream data, while in practice, the scenarios with scarce data are common. Recent studies in the natural language processing (NLP) field show that prompt tuning, a new paradigm for tuning, alleviates the above issues and achieves promising results in various NLP tasks. In prompt tuning, the prompts inserted during tuning provide task-specific knowledge, which is especially beneficial for tasks with relatively scarce data. In this paper, we empirically evaluate the usage and effect of prompt tuning in code intelligence tasks. We conduct prompt tuning on popular pre-trained models CodeBERT and CodeT5 and experiment with three code intelligence tasks including defect prediction, code summarization, and code translation. Our experimental results show that prompt tuning consistently outperforms fine-tuning in all three tasks. In addition, prompt tuning shows great potential in low-resource scenarios, e.g., improving the BLEU scores of fine-tuning by more than 26\% on average for code summarization. Our results suggest that instead of fine-tuning, we could adapt prompt tuning for code intelligence tasks to achieve better performance, especially when lacking task-specific data.

Dawei Tian, Jiakun Liu, Yun Peng et al.

Modern software systems evolve rapidly under CI/CD practices, where tests are critical for quality. However, substantial code changes often render existing test cases obsolete, causing pipeline disruptions, reduced productivity, and compromised quality. Recent automatic test update approaches leverage LLMs to refine test cases via execution feedback and exact-matching context retrieval, prioritizing executability and line coverage but suffering three limitations: (1) neglecting test assertion adequacy, weakening fault detection; (2) relying on coarse line coverage instead of specific uncovered lines/branches; (3) using exact-matching retrieval, which fails for LLM hallucinated queries. To address these, we propose MuMuTestUp, a mutation-guided multi-agent framework with three specialized agents: Mutation Analysis (strengthens assertions via surviving mutants), Coverage Analysis (generates targeted repair instructions for uncovered lines/branches), and Semantic Retrieval (handles hallucinations via semantic-similarity search). We also construct PRBENCH, a 571-sample pull-request-level dataset from 10 open-source Java projects (validated for cross-commit update scenarios). Evaluations against state-of-the-art baselines use both open-source (Deepseek-V3.2) and closed-source (GPT-4.1) LLMs.

Shangyu Li, Juyong Jiang, Meibo Ren et al.

Transpilation, or code translation, aims to convert source code from one programming language (PL) to another. It is beneficial for many downstream applications, from modernizing large legacy codebases to augmenting data for low-resource PLs. Recent large language model (LLM)-based approaches have demonstrated immense potential for code translation. Among these approaches, training-based methods are particularly important because LLMs currently do not effectively adapt to domain-specific settings that suffer from a lack of knowledge without targeted training. This limitation is evident in transpilation tasks involving low-resource PLs. However, existing training-based approaches rely on a pairwise transpilation paradigm, making it impractical to support a diverse range of PLs. This limitation is particularly prominent for low-resource PLs due to a scarcity of training data. Furthermore, these methods suffer from suboptimal reinforcement learning (RL) reward formulations. To address these limitations, we propose CodePivot, a training framework that leverages Python as an intermediate representation (IR), augmented by a novel RL reward mechanism, Aggressive-Partial-Functional reward, to bootstrap the model's multilingual transpilation ability without requiring parallel corpora. Experiments involving 10 PLs show that the resulting 7B model, trained on Python-to-Others tasks, consistently improves performance across both general and low-resource PL-related transpilation tasks. It outperforms substantially larger mainstream models with hundreds of billions more parameters, such as Deepseek-R1 and Qwen3-235B-A22B-Instruct-2507, on Python-to-Others tasks and Others-to-All tasks, respectively. In addition, it outperforms its counterpart trained directly on Any-to-Any tasks on general transpilation tasks. The code and data are available at https://github.com/lishangyu-hkust/CodePivot.

Feng Zhou, Yanjie Zhou, Longjie Wang et al.

Traditional one-shot medical image segmentation (MIS) methods use registration networks to propagate labels from a reference atlas or rely on comprehensive sampling strategies to generate synthetic labeled data for training. However, these methods often struggle with registration errors and low-quality synthetic images, leading to poor performance and generalization. To overcome this, we introduce a novel one-shot MIS framework based on knowledge distillation, which allows the network to directly 'see' real images through a distillation process guided by image reconstruction. It focuses on anatomical structures in a single labeled image and a few unlabeled ones. A registration-based data augmentation network creates realistic, labeled samples, while a feature distillation module helps the student network learn segmentation from these samples, guided by the teacher network. During inference, the streamlined student network accurately segments new images. Evaluations on three public datasets (OASIS for T1 brain MRI, BCV for abdomen CT, and VerSe for vertebrae CT) show superior segmentation performance and generalization across different medical image datasets and modalities compared to leading methods. Our code is available at https://github.com/NoviceFodder/OS-MedSeg.

Wei Ai, Yun Peng, Yuntao Shou et al.

With the rapid growth of IoT devices and latency-sensitive applications, the demand for both real-time and energy-efficient computing has surged, placing significant pressure on traditional cloud computing architectures. Mobile edge computing (MEC), an emerging paradigm, effectively alleviates the load on cloud centers and improves service quality by offloading computing tasks to edge servers closer to end users. However, the limited computing resources, non-continuous power provisioning (e.g., battery-powered nodes), and highly dynamic systems of edge servers complicate efficient task scheduling and resource allocation. To address these challenges, this paper proposes a multi-agent deep reinforcement learning algorithm, TG-DCMADDPG, and constructs a collaborative computing framework for multiple edge servers, aiming to achieve joint optimization of fine-grained task partitioning and offloading. This approach incorporates a temporal graph neural network (TimeGNN) to model and predict time series of multi-dimensional server state information, thereby reducing the frequency of online interactions and improving policy predictability. Furthermore, a multi-agent deterministic policy gradient algorithm (DC-MADDPG) in a discrete-continuous hybrid action space is introduced to collaboratively optimize task partitioning ratios, transmission power, and priority scheduling strategies. Extensive simulation experiments confirm that TG-DCMADDPG achieves markedly faster policy convergence, superior energy-latency optimization, and higher task completion rates compared with existing state-of-the-art methods, underscoring its robust scalability and practical effectiveness in dynamic and constrained MEC scenarios.

Zhanyue Qin, Jia Feng, Yibo Lyu et al.

Code reasoning refers to the task of predicting the output of a program given its source code and specific inputs. It can measure the reasoning capability of large language models (LLMs) and also benefit downstream tasks such as code generation and mathematical reasoning. Existing work has verified the effectiveness of reinforcement learning on the task. However, these methods design rewards solely based on final outputs or coarse-grained signals, and neglect the inherent consistency of the stepwise reasoning process in the task. Therefore, these methods often result in sparse reward or reward hacking, which limits the full play of enhanced learning capabilities. To alleviate these issues, we propose CodeThinker, a consistency-driven reinforcement learning framework for code reasoning. Specifically, CodeThinker has three key components: (1) a stepwise reasoning-aware model training module, which utilizes a consistency tracing paradigm as a template to synthesize training data that captures the stepwise reasoning process; (2) a dynamic beam sampling strategy, which aims to improve the quality of sampled outputs under a fixed sampling budget; and (3) a consistency reward mechanism that can effectively alleviate reward hacking. Experiments on three popular benchmarks show that CodeThinker achieves state-of-the-art performance across multiple LLMs. For instance, it outperforms the strongest baseline by 4.3% in accuracy when deployed on Qwen2.5-Coder-7B-Instruct. We also validate the effectiveness of CodeThinker on downstream tasks. Results show that, without additional training, CodeThinker obtains average accuracy gains of 5.33 and 3.11 percentage points on mathematical reasoning and code reasoning tasks covering 17 programming languages, respectively.

Yonghe Zhao, Qiang Huang, Siwei Wu et al.

Causal inference plays a vital role in diverse domains like epidemiology, healthcare, and economics. De-confounding and counterfactual prediction in observational data has emerged as a prominent concern in causal inference research. While existing models tackle observed confounders, the presence of unobserved confounders remains a significant challenge, distorting causal inference and impacting counterfactual outcome accuracy. To address this, we propose a novel variational learning model of unobserved confounders for counterfactual inference (VLUCI), which generates the posterior distribution of unobserved confounders. VLUCI relaxes the unconfoundedness assumption often overlooked by most causal inference methods. By disentangling observed and unobserved confounders, VLUCI constructs a doubly variational inference model to approximate the distribution of unobserved confounders, which are used for inferring more accurate counterfactual outcomes. Extensive experiments on synthetic and semi-synthetic datasets demonstrate VLUCI's superior performance in inferring unobserved confounders. It is compatible with state-of-the-art counterfactual inference models, significantly improving inference accuracy at both group and individual levels. Additionally, VLUCI provides confidence intervals for counterfactual outcomes, aiding decision-making in risk-sensitive domains. We further clarify the considerations when applying VLUCI to cases where unobserved confounders don't strictly conform to our model assumptions using the public IHDP dataset as an example, highlighting the practical advantages of VLUCI.

Xiao Lin, Yun Peng, Liuyi Wang et al.

Category-level object pose estimation aims to recover the rotation, translation and size of unseen instances within predefined categories. In this task, deep neural network-based methods have demonstrated remarkable performance. However, previous studies show they suffer from spurious correlations raised by "unclean" confounders in models, hindering their performance on novel instances with significant variations. To address this issue, we propose CleanPose, a novel approach integrating causal learning and knowledge distillation to enhance category-level pose estimation. To mitigate the negative effect of unobserved confounders, we develop a causal inference module based on front-door adjustment, which promotes unbiased estimation by reducing potential spurious correlations. Additionally, to further improve generalization ability, we devise a residual-based knowledge distillation method that has proven effective in providing comprehensive category information guidance. Extensive experiments across multiple benchmarks (REAL275, CAMERA25 and HouseCat6D) hightlight the superiority of proposed CleanPose over state-of-the-art methods. Code will be available at https://github.com/chrislin0621/CleanPose.

Yun Peng, Akhilesh Deepak Gotmare, Michael Lyu et al. · salesforce

Large Language Models (LLMs) are widely adopted for assisting in software development tasks, yet their performance evaluations have narrowly focused on the functional correctness of generated code. Human programmers, however, require LLM-generated code to be not only correct but also optimally efficient. We propose PerfCodeGen, a training-free framework that enhances the performance of LLM-generated code by incorporating feedback based on runtime during test case execution into the self-refinement iterations. With PerfCodeGen, we achieve speedups for a significantly higher proportion of problems compared to using the base LLM with sophisticated prompting techniques. Applied to open language models like Phi-3-mini, PerfCodeGen achieves runtime efficiency comparable to prompting powerful closed models like GPT-4. We achieve state-of-the-art runtime efficiency on benchmarks such as HumanEval, MBPP, and APPS, frequently surpassing the ground truth reference solutions with PerfCodeGen using GPT-3.5 and GPT-4. Additionally, we demonstrate the effectiveness of our approach in enhancing code quality across a range of open LLMs of varying sizes including Phi-3-mini, Llama 3 8B, Mixtral 8x7B, Command R, and Llama 3 70B.

Yichen Li, Yun Peng, Yintong Huo et al.

Large Language Models (LLMs) have achieved remarkable success in code completion, as evidenced by their essential roles in developing code assistant services such as Copilot. Being trained on in-file contexts, current LLMs are quite effective in completing code for single source files. However, it is challenging for them to conduct repository-level code completion for large software projects that require cross-file information. Existing research on LLM-based repository-level code completion identifies and integrates cross-file contexts, but it suffers from low accuracy and limited context length of LLMs. In this paper, we argue that Integrated Development Environments (IDEs) can provide direct, accurate and real-time cross-file information for repository-level code completion. We propose IDECoder, a practical framework that leverages IDE native static contexts for cross-context construction and diagnosis results for self-refinement. IDECoder utilizes the rich cross-context information available in IDEs to enhance the capabilities of LLMs of repository-level code completion. We conducted preliminary experiments to validate the performance of IDECoder and observed that this synergy represents a promising trend for future exploration.

Jun Gao, Yun Peng, Xiaoxue Ren

Large Language Models (LLMs) have achieved remarkable progress in code-related tasks. Despite their advancement, empirical evidence reveals that they still struggle with \emph{deductive code reasoning}, the ability to reason about the program execution process. While prior studies have recognized this limitation, the underlying causes remain largely underexplored. In this paper, we begin by presenting a comprehensive empirical study that reveals three key challenges undermining deductive code reasoning: (1) an intrinsic gap between generation and reasoning abilities, (2) a consistent bias towards code sources, and (3) weak zero-shot generalization on complex benchmarks. In light of these challenges, we propose \texttt{ReMind}, a multi-agent framework composed of \texttt{Mutator}, \texttt{Executor}, and \texttt{Inspector}. The \texttt{Mutator} generates code variants to mitigate bias towards code sources, the \texttt{Executor} traces variable states step-by-step to expose inconsistency, and the \texttt{Inspector} identifies problematic reasoning steps and provides control-flow refinement to bridge the intrinsic reasoning gap. Through their coordinated collaboration, \texttt{ReMind} systematically identifies and refines reasoning flaws, achieving outstanding performance and enabling robust zero-shot generalization. Extensive experiments on two benchmarks with five LLMs demonstrate the superior advantages of \texttt{ReMind} compared to baseline approaches in deductive code reasoning.

Yun Peng, Sen Lin, Qian Chen et al.

Graph analysis is fundamental in real-world applications. Traditional approaches rely on SPARQL-like languages or clicking-and-dragging interfaces to interact with graph data. However, these methods either require users to possess high programming skills or support only a limited range of graph analysis functionalities. To address the limitations, we propose a large language model (LLM)-based framework called ChatGraph. With ChatGraph, users can interact with graphs through natural language, making it easier to use and more flexible than traditional approaches. The core of ChatGraph lies in generating chains of graph analysis APIs based on the understanding of the texts and graphs inputted in the user prompts. To achieve this, ChatGraph consists of three main modules: an API retrieval module that searches for relevant APIs, a graph-aware LLM module that enables the LLM to comprehend graphs, and an API chain-oriented finetuning module that guides the LLM in generating API chains.

Shuqing Li, Anson Y. Lam, Yun Peng et al.

Graphical user interface (UI) software has undergone a fundamental transformation from traditional two-dimensional (2D) desktop/web/mobile interfaces to spatial three-dimensional (3D) environments. While existing work has made remarkable success in automated 2D software generation, such as HTML/CSS and mobile app interface code synthesis, the generation of 3D software still remains under-explored. Current methods for 3D software generation usually generate the 3D environments as a whole and cannot modify or control specific elements in the software. Furthermore, these methods struggle to handle the complex spatial and semantic constraints inherent in the real world. To address the challenges, we present Scenethesis, a novel requirement-sensitive 3D software synthesis approach that maintains formal traceability between user specifications and generated 3D software. Scenethesis is built upon ScenethesisLang, a domain-specific language that serves as a granular constraint-aware intermediate representation (IR) to bridge natural language requirements and executable 3D software. It serves both as a comprehensive scene description language enabling fine-grained modification of 3D software elements and as a formal constraint-expressive specification language capable of expressing complex spatial constraints. By decomposing 3D software synthesis into stages operating on ScenethesisLang, Scenethesis enables independent verification, targeted modification, and systematic constraint satisfaction. Our evaluation demonstrates that Scenethesis accurately captures over 80% of user requirements and satisfies more than 90% of hard constraints while handling over 100 constraints simultaneously. Furthermore, Scenethesis achieves a 42.8% improvement in BLIP-2 visual evaluation scores compared to the state-of-the-art method.

Zihan Wu, Jie Xu, Yun Peng et al.

Large Language Models (LLMs) struggle to automate real-world vulnerability detection due to two key limitations: the heterogeneity of vulnerability patterns undermines the effectiveness of a single unified model, and manual prompt engineering for massive weakness categories is unscalable. To address these challenges, we propose \textbf{MulVul}, a retrieval-augmented multi-agent framework designed for precise and broad-coverage vulnerability detection. MulVul adopts a coarse-to-fine strategy: a \emph{Router} agent first predicts the top-$k$ coarse categories and then forwards the input to specialized \emph{Detector} agents, which identify the exact vulnerability types. Both agents are equipped with retrieval tools to actively source evidence from vulnerability knowledge bases to mitigate hallucinations. Crucially, to automate the generation of specialized prompts, we design \emph{Cross-Model Prompt Evolution}, a prompt optimization mechanism where a generator LLM iteratively refines candidate prompts while a distinct executor LLM validates their effectiveness. This decoupling mitigates the self-correction bias inherent in single-model optimization. Evaluated on 130 CWE types, MulVul achieves 34.79\% Macro-F1, outperforming the best baseline by 41.5\%. Ablation studies validate cross-model prompt evolution, which boosts performance by 51.6\% over manual prompts by effectively handling diverse vulnerability patterns.

Shuqing Li, Jiayi Yan, Chenyu Niu et al. · pku, tencent-ai

Virtual Reality (VR) games require players to translate high-level semantic actions into precise device manipulations using controllers and head-mounted displays (HMDs). While humans intuitively perform this translation based on common sense and embodied understanding, whether Large Language Models (LLMs) can effectively replicate this ability remains underexplored. This paper introduces a benchmark, ComboBench, evaluating LLMs' capability to translate semantic actions into VR device manipulation sequences across 262 scenarios from four popular VR games: Half-Life: Alyx, Into the Radius, Moss: Book II, and Vivecraft. We evaluate seven LLMs, including GPT-3.5, GPT-4, GPT-4o, Gemini-1.5-Pro, LLaMA-3-8B, Mixtral-8x7B, and GLM-4-Flash, compared against annotated ground truth and human performance. Our results reveal that while top-performing models like Gemini-1.5-Pro demonstrate strong task decomposition capabilities, they still struggle with procedural reasoning and spatial understanding compared to humans. Performance varies significantly across games, suggesting sensitivity to interaction complexity. Few-shot examples substantially improve performance, indicating potential for targeted enhancement of LLMs' VR manipulation capabilities. We release all materials at https://sites.google.com/view/combobench.

Yi Yin, Yuntao Shou, Zao Dai et al.

In recent years, multimodal medical data-based survival analysis has attracted much attention. However, real-world datasets often suffer from the problem of incomplete modality, where some patient modality information is missing due to acquisition limitations or system failures. Existing methods typically infer missing modalities directly from observed ones using deep neural networks, but they often ignore the distributional discrepancy across modalities, resulting in inconsistent and unreliable modality reconstruction. To address these challenges, we propose a novel framework that combines a low-rank Transformer with a flow-based generative model for robust and flexible multimodal survival prediction. Specifically, we first formulate the concerned problem as incomplete multimodal survival analysis using the multi-instance representation of whole slide images (WSIs) and genomic profiles. To realize incomplete multimodal survival analysis, we propose a class-specific flow for cross-modal distribution alignment. Under the condition of class labels, we model and transform the cross-modal distribution. By virtue of the reversible structure and accurate density modeling capabilities of the normalizing flow model, the model can effectively construct a distribution-consistent latent space of the missing modality, thereby improving the consistency between the reconstructed data and the true distribution. Finally, we design a lightweight Transformer architecture to model intra-modal dependencies while alleviating the overfitting problem in high-dimensional modality fusion by virtue of the low-rank Transformer. Extensive experiments have demonstrated that our method not only achieves state-of-the-art performance under complete modality settings, but also maintains robust and superior accuracy under the incomplete modalities scenario.

Wenxuan Wang, Yongjiang Wu, Junyuan Zhang et al.

The rapid growth of audio-centric platforms and applications such as WhatsApp and Twitter has transformed the way people communicate and share audio content in modern society. However, these platforms are increasingly misused to disseminate harmful audio content, such as hate speech, deceptive advertisements, and explicit material, which can have significant negative consequences (e.g., detrimental effects on mental health). In response, researchers and practitioners have been actively developing and deploying audio content moderation tools to tackle this issue. Despite these efforts, malicious actors can bypass moderation systems by making subtle alterations to audio content, such as modifying pitch or inserting noise. Moreover, the effectiveness of modern audio moderation tools against such adversarial inputs remains insufficiently studied. To address these challenges, we propose MTAM, a Metamorphic Testing framework for Audio content Moderation software. Specifically, we conduct a pilot study on 2000 audio clips and define 14 metamorphic relations across two perturbation categories: Audio Features-Based and Heuristic perturbations. MTAM applies these metamorphic relations to toxic audio content to generate test cases that remain harmful while being more likely to evade detection. In our evaluation, we employ MTAM to test five commercial textual content moderation software and an academic model against three kinds of toxic content. The results show that MTAM achieves up to 38.6%, 18.3%, 35.1%, 16.7%, and 51.1% error finding rates (EFR) when testing commercial moderation software provided by Gladia, Assembly AI, Baidu, Nextdata, and Tencent, respectively, and it obtains up to 45.7% EFR when testing the state-of-the-art algorithms from the academy.

Shuqing Li, Cuiyun Gao, Jianping Zhang et al.

The quality of Virtual Reality (VR) apps is vital, particularly the rendering quality of the VR Graphical User Interface (GUI). Different from traditional 2D apps, VR apps create a 3D digital scene for users, by rendering two distinct 2D images for the user's left and right eyes, respectively. Stereoscopic visual inconsistency (denoted as "SVI") issues, however, undermine the rendering process of the user's brain, leading to user discomfort and even adverse health effects. Such issues commonly exist but remain underexplored. We conduct an empirical analysis on 282 SVI bug reports from 15 VR platforms, summarizing 15 types of manifestations. The empirical analysis reveals that automatically detecting SVI issues is challenging, mainly because: (1) lack of training data; (2) the manifestations of SVI issues are diverse, complicated, and often application-specific; (3) most accessible VR apps are closed-source commercial software. Existing pattern-based supervised classification approaches may be inapplicable or ineffective in detecting the SVI issues. To counter these challenges, we propose an unsupervised black-box testing framework named StereoID to identify the stereoscopic visual inconsistencies, based only on the rendered GUI states. StereoID generates a synthetic right-eye image based on the actual left-eye image and computes distances between the synthetic right-eye image and the actual right-eye image to detect SVI issues. We propose a depth-aware conditional stereo image translator to power the image generation process, which captures the expected perspective shifts between left-eye and right-eye images. We build a large-scale unlabeled VR stereo screenshot dataset with larger than 171K images from 288 real-world VR apps for experiments. After substantial experiments, StereoID demonstrates superior performance for detecting SVI issues in both user reports and wild VR apps.

Yun Peng, Xiao Lin, Nachuan Ma et al.

Visual anomaly detection is vital in real-world applications, such as industrial defect detection and medical diagnosis. However, most existing methods focus on local structural anomalies and fail to detect higher-level functional anomalies under logical conditions. Although recent studies have explored logical anomaly detection, they can only address simple anomalies like missing or addition and show poor generalizability due to being heavily data-driven. To fill this gap, we propose SAM-LAD, a zero-shot, plug-and-play framework for logical anomaly detection in any scene. First, we obtain a query image's feature map using a pre-trained backbone. Simultaneously, we retrieve the reference images and their corresponding feature maps via the nearest neighbor search of the query image. Then, we introduce the Segment Anything Model (SAM) to obtain object masks of the query and reference images. Each object mask is multiplied with the entire image's feature map to obtain object feature maps. Next, an Object Matching Model (OMM) is proposed to match objects in the query and reference images. To facilitate object matching, we further propose a Dynamic Channel Graph Attention (DCGA) module, treating each object as a keypoint and converting its feature maps into feature vectors. Finally, based on the object matching relations, an Anomaly Measurement Model (AMM) is proposed to detect objects with logical anomalies. Structural anomalies in the objects can also be detected. We validate our proposed SAM-LAD using various benchmarks, including industrial datasets (MVTec Loco AD, MVTec AD), and the logical dataset (DigitAnatomy). Extensive experimental results demonstrate that SAM-LAD outperforms existing SoTA methods, particularly in detecting logical anomalies.

Yongzhi Huang, Fengjun Xi, Liyun Tu et al.

Accurate segmentation of multiple organs in Computed Tomography (CT) images plays a vital role in computer-aided diagnosis systems. While various supervised learning approaches have been proposed recently, these methods heavily depend on a large amount of high-quality labeled data, which are expensive to obtain in practice. To address this challenge, we propose a label-efficient framework using knowledge transfer from a pre-trained diffusion model for CT multi-organ segmentation. Specifically, we first pre-train a denoising diffusion model on 207,029 unlabeled 2D CT slices to capture anatomical patterns. Then, the model backbone is transferred to the downstream multi-organ segmentation task, followed by fine-tuning with few labeled data. In fine-tuning, two fine-tuning strategies, linear classification and fine-tuning decoder, are employed to enhance segmentation performance while preserving learned representations. Quantitative results show that the pre-trained diffusion model is capable of generating diverse and realistic 256x256 CT images (Fréchet inception distance (FID): 11.32, spatial Fréchet inception distance (sFID): 46.93, F1-score: 73.1%). Compared to state-of-the-art methods for multi-organ segmentation, our method achieves competitive performance on the FLARE 2022 dataset, particularly in limited labeled data scenarios. After fine-tuning with 1% and 10% labeled data, our method achieves dice similarity coefficients (DSCs) of 71.56% and 78.51%, respectively. Remarkably, the method achieves a DSC score of 51.81% using only four labeled CT slices. These results demonstrate the efficacy of our approach in overcoming the limitations of supervised learning approaches that is highly dependent on large-scale labeled data.

Zi Gong, Cuiyun Gao, Yasheng Wang et al.

Source code summarization aims at generating concise and clear natural language descriptions for programming languages. Well-written code summaries are beneficial for programmers to participate in the software development and maintenance process. To learn the semantic representations of source code, recent efforts focus on incorporating the syntax structure of code into neural networks such as Transformer. Such Transformer-based approaches can better capture the long-range dependencies than other neural networks including Recurrent Neural Networks (RNNs), however, most of them do not consider the structural relative correlations between tokens, e.g., relative positions in Abstract Syntax Trees (ASTs), which is beneficial for code semantics learning. To model the structural dependency, we propose a Structural Relative Position guided Transformer, named SCRIPT. SCRIPT first obtains the structural relative positions between tokens via parsing the ASTs of source code, and then passes them into two types of Transformer encoders. One Transformer directly adjusts the input according to the structural relative distance; and the other Transformer encodes the structural relative positions during computing the self-attention scores. Finally, we stack these two types of Transformer encoders to learn representations of source code. Experimental results show that the proposed SCRIPT outperforms the state-of-the-art methods by at least 1.6%, 1.4% and 2.8% with respect to BLEU, ROUGE-L and METEOR on benchmark datasets, respectively. We further show that how the proposed SCRIPT captures the structural relative dependencies.

Yun Peng, Shuqing Li, Wenwei Gu et al.

Application Programming Interfaces (APIs), which encapsulate the implementation of specific functions as interfaces, greatly improve the efficiency of modern software development. As numbers of APIs spring up nowadays, developers can hardly be familiar with all the APIs, and usually need to search for appropriate APIs for usage. So lots of efforts have been devoted to improving the API recommendation task. However, it has been increasingly difficult to gauge the performance of new models due to the lack of a uniform definition of the task and a standardized benchmark. For example, some studies regard the task as a code completion problem; while others recommend relative APIs given natural language queries. To reduce the challenges and better facilitate future research, in this paper, we revisit the API recommendation task and aim at benchmarking the approaches. Specifically, the paper groups the approaches into two categories according to the task definition, i.e., query-based API recommendation and code-based API recommendation. We study 11 recently-proposed approaches along with 4 widely-used IDEs. One benchmark named as APIBench is then built for the two respective categories of approaches. Based on APIBench, we distill some actionable insights and challenges for API recommendation. We also achieve some implications and directions for improving the performance of recommending APIs, including data source selection, appropriate query reformulation, low resource setting, and cross-domain adaptation.

Chaochao Zhou, Thomas Cha, Yun Peng et al.

Registration of 3D anatomic structures to their 2D dual fluoroscopic X-ray images is a widely used motion tracking technique. However, deep learning implementation is often impeded by a paucity of medical images and ground truths. In this study, we proposed a transfer learning strategy for 3D-to-2D registration using deep neural networks trained from an artificial dataset. Digitally reconstructed radiographs (DRRs) and radiographic skull landmarks were automatically created from craniocervical CT data of a female subject. They were used to train a residual network (ResNet) for landmark detection and a cycle generative adversarial network (GAN) to eliminate the style difference between DRRs and actual X-rays. Landmarks on the X-rays experiencing GAN style translation were detected by the ResNet, and were used in triangulation optimization for 3D-to-2D registration of the skull in actual dual-fluoroscope images (with a non-orthogonal setup, point X-ray sources, image distortions, and partially captured skull regions). The registration accuracy was evaluated in multiple scenarios of craniocervical motions. In walking, learning-based registration for the skull had angular/position errors of 3.9 +- 2.1 deg / 4.6 +- 2.2 mm. However, the accuracy was lower during functional neck activity, due to overly small skull regions imaged on the dual fluoroscopic images at end-range positions. The methodology to strategically augment artificial training data can tackle the complicated skull registration scenario, and has potentials to extend to widespread registration scenarios.

Yun Peng, Cuiyun Gao, Zongjie Li et al.

Type inference for dynamic programming languages such as Python is an important yet challenging task. Static type inference techniques can precisely infer variables with enough static constraints but are unable to handle variables with dynamic features. Deep learning (DL) based approaches are feature-agnostic, but they cannot guarantee the correctness of the predicted types. Their performance significantly depends on the quality of the training data (i.e., DL models perform poorly on some common types that rarely appear in the training dataset). It is interesting to note that the static and DL-based approaches offer complementary benefits. Unfortunately, to our knowledge, precise type inference based on both static inference and neural predictions has not been exploited and remains an open challenge. In particular, it is hard to integrate DL models into the framework of rule-based static approaches. This paper fills the gap and proposes a hybrid type inference approach named HiTyper based on both static inference and deep learning. Specifically, our key insight is to record type dependencies among variables in each function and encode the dependency information in type dependency graphs (TDGs). Based on TDGs, we can easily integrate type inference rules in the nodes to conduct static inference and type rejection rules to inspect the correctness of neural predictions. HiTyper iteratively conducts static inference and DL-based prediction until the TDG is fully inferred. Experiments on two benchmark datasets show that HiTyper outperforms state-of-the-art DL models by exactly matching 10% more human annotations. HiTyper also achieves an increase of more than 30% on inferring rare types. Considering only the static part of HiTyper, it infers 2x ~ 3x more types than existing static type inference tools.

Yun Peng, Byron Choi, Jianliang Xu

Graphs have been widely used to represent complex data in many applications. Efficient and effective analysis of graphs is important for graph-based applications. However, most graph analysis tasks are combinatorial optimization (CO) problems, which are NP-hard. Recent studies have focused a lot on the potential of using machine learning (ML) to solve graph-based CO problems. Most recent methods follow the two-stage framework. The first stage is graph representation learning, which embeds the graphs into low-dimension vectors. The second stage uses ML to solve the CO problems using the embeddings of the graphs learned in the first stage. The works for the first stage can be classified into two categories, graph embedding (GE) methods and end-to-end (E2E) learning methods. For GE methods, learning graph embedding has its own objective, which may not rely on the CO problems to be solved. The CO problems are solved by independent downstream tasks. For E2E learning methods, the learning of graph embeddings does not have its own objective and is an intermediate step of the learning procedure of solving the CO problems. The works for the second stage can also be classified into two categories, non-autoregressive methods and autoregressive methods. Non-autoregressive methods predict a solution for a CO problem in one shot. A non-autoregressive method predicts a matrix that denotes the probability of each node/edge being a part of a solution of the CO problem. The solution can be computed from the matrix. Autoregressive methods iteratively extend a partial solution step by step. At each step, an autoregressive method predicts a node/edge conditioned to current partial solution, which is used to its extension. In this survey, we provide a thorough overview of recent studies of the graph learning-based CO methods. The survey ends with several remarks on future research directions.

Yun Peng, Zhongli Ding

This paper deals with the following problem: modify a Bayesian network to satisfy a given set of probability constraints by only change its conditional probability tables, and the probability distribution of the resulting network should be as close as possible to that of the original network. We propose to solve this problem by extending IPFP (iterative proportional fitting procedure) to probability distributions represented by Bayesian networks. The resulting algorithm E-IPFP is further developed to D-IPFP, which reduces the computational cost by decomposing a global EIPFP into a set of smaller local E-IPFP problems. Limited analysis is provided, including the convergence proofs of the two algorithms. Computer experiments were conducted to validate the algorithms. The results are consistent with the theoretical analysis.