Xiaochen Wang

Kimi Team, Tongtong Bai, Yifan Bai et al.

We introduce Kimi K2.5, an open-source multimodal agentic model designed to advance general agentic intelligence. K2.5 emphasizes the joint optimization of text and vision so that two modalities enhance each other. This includes a series of techniques such as joint text-vision pre-training, zero-vision SFT, and joint text-vision reinforcement learning. Building on this multimodal foundation, K2.5 introduces Agent Swarm, a self-directed parallel agent orchestration framework that dynamically decomposes complex tasks into heterogeneous sub-problems and executes them concurrently. Extensive evaluations show that Kimi K2.5 achieves state-of-the-art results across various domains including coding, vision, reasoning, and agentic tasks. Agent Swarm also reduces latency by up to $4.5\times$ over single-agent baselines. We release the post-trained Kimi K2.5 model checkpoint to facilitate future research and real-world applications of agentic intelligence.

Xinpeng Qiu, Wang Yihu, Zhifeng Liu et al.

Scientific peer review generation has attracted increasing attention for reducing reviewing burdens and providing timely feedback. However, existing Large Language Model (LLM)-based methods often produce generic comments with insufficient evidence support and weak source traceability, while complex multi-agent systems incur high inference costs. To address these challenges, we propose EGTR-Review, an Evidence-Grounded and Traceable Review Generation framework via Multi-Agent Teacher Distillation. EGTR-Review first constructs a multi-agent teacher that performs structure-aware paper decomposition, key-element extraction, external scholarly evidence retrieval, evidence-state labeling, verification reasoning, and review synthesis. It then distills both intermediate reasoning trajectories and final review comments into a lightweight student model through task-prefix-driven multi-task learning. An evidence-weighted objective further reduces the influence of weak, missing, or non-verifiable supervision. Experiments on public peer-review datasets show that EGTR-Review (Student) outperforms strong prompt-based, fine-tuned, and structured/agentic baselines across automatic metrics, LLM-as-Judge evaluation, and human evaluation, while maintaining strong factual grounding and source traceability with substantially lower token consumption and inference time. Our code, prompts, configurations, and sample data are available on GitHub.

Chengwen Wang, Qingxiu Dong, Xiaochen Wang et al. · pku

Datasets serve as crucial training resources and model performance trackers. However, existing datasets have exposed a plethora of problems, inducing biased models and unreliable evaluation results. In this paper, we propose a model-agnostic dataset evaluation framework for automatic dataset quality evaluation. We seek the statistical properties of the datasets and address three fundamental dimensions: reliability, difficulty, and validity, following a classical testing theory. Taking the Named Entity Recognition (NER) datasets as a case study, we introduce $9$ statistical metrics for a statistical dataset evaluation framework. Experimental results and human evaluation validate that our evaluation framework effectively assesses various aspects of the dataset quality. Furthermore, we study how the dataset scores on our statistical metrics affect the model performance, and appeal for dataset quality evaluation or targeted dataset improvement before training or testing models.

Yifan Pu, Jixuan Ying, Qixiu Li et al.

Vision Transformers (ViTs) have become a universal backbone for both image recognition and image generation. Yet their Multi-Head Self-Attention (MHSA) layer still performs a quadratic query-key interaction for every token pair, spending the bulk of computation on visually weak or redundant correlations. We introduce Visual-Contrast Attention (VCA), a drop-in replacement for MHSA that injects an explicit notion of discrimination while reducing the theoretical complexity from O(N N C) to O(N n C) with n << N. VCA first distils each head's dense query field into a handful of spatially pooled visual-contrast tokens, then splits them into a learnable positive and negative stream whose differential interaction highlights what truly separates one region from another. The module adds fewer than 0.3M parameters to a DeiT-Tiny backbone, requires no extra FLOPs, and is wholly architecture-agnostic. Empirically, VCA lifts DeiT-Tiny top-1 accuracy on ImageNet-1K from 72.2% to 75.6% (+3.4) and improves three strong hierarchical ViTs by up to 3.1%, while in class-conditional ImageNet generation it lowers FID-50K by 2.1 to 5.2 points across both diffusion (DiT) and flow (SiT) models. Extensive ablations confirm that (i) spatial pooling supplies low-variance global cues, (ii) dual positional embeddings are indispensable for contrastive reasoning, and (iii) combining the two in both stages yields the strongest synergy. VCA therefore offers a simple path towards faster and sharper Vision Transformers. The source code is available at https://github.com/LeapLabTHU/LinearDiff.

Xiaochen Wang, Junqing He, Zhe yang et al. · pku

Large Language Models (LLMs) with chain-of-thought (COT) prompting have demonstrated impressive abilities on simple nature language inference tasks. However, they tend to perform poorly on Multi-hop Question Answering (MHQA) tasks due to several challenges, including hallucination, error propagation and limited context length. We propose a prompting method, Finite State Machine (FSM) to enhance the reasoning capabilities of LLM for complex tasks in addition to improved effectiveness and trustworthiness. Different from COT methods, FSM addresses MHQA by iteratively decomposing a question into multi-turn sub-questions, and self-correcting in time, improving the accuracy of answers in each step. Specifically, FSM addresses one sub-question at a time and decides on the next step based on its current result and state, in an automaton-like format. Experiments on benchmarks show the effectiveness of our method. Although our method performs on par with the baseline on relatively simpler datasets, it excels on challenging datasets like Musique. Moreover, this approach mitigates the hallucination phenomenon, wherein the correct final answer can be recovered despite errors in intermediate reasoning. Furthermore, our method improves LLMs' ability to follow specified output format requirements, significantly reducing the difficulty of answer interpretation and the need for reformatting.

Yuan Zhong, Suhan Cui, Jiaqi Wang et al.

Health risk prediction is one of the fundamental tasks under predictive modeling in the medical domain, which aims to forecast the potential health risks that patients may face in the future using their historical Electronic Health Records (EHR). Researchers have developed several risk prediction models to handle the unique challenges of EHR data, such as its sequential nature, high dimensionality, and inherent noise. These models have yielded impressive results. Nonetheless, a key issue undermining their effectiveness is data insufficiency. A variety of data generation and augmentation methods have been introduced to mitigate this issue by expanding the size of the training data set through the learning of underlying data distributions. However, the performance of these methods is often limited due to their task-unrelated design. To address these shortcomings, this paper introduces a novel, end-to-end diffusion-based risk prediction model, named MedDiffusion. It enhances risk prediction performance by creating synthetic patient data during training to enlarge sample space. Furthermore, MedDiffusion discerns hidden relationships between patient visits using a step-wise attention mechanism, enabling the model to automatically retain the most vital information for generating high-quality data. Experimental evaluation on four real-world medical datasets demonstrates that MedDiffusion outperforms 14 cutting-edge baselines in terms of PR-AUC, F1, and Cohen's Kappa. We also conduct ablation studies and benchmark our model against GAN-based alternatives to further validate the rationality and adaptability of our model design. Additionally, we analyze generated data to offer fresh insights into the model's interpretability.

Bo Yan, Cheng Yang, Chuan Shi et al.

Abnormal event detection, which refers to mining unusual interactions among involved entities, plays an important role in many real applications. Previous works mostly over-simplify this task as detecting abnormal pair-wise interactions. However, real-world events may contain multi-typed attributed entities and complex interactions among them, which forms an Attributed Heterogeneous Information Network (AHIN). With the boom of social networks, abnormal event detection in AHIN has become an important, but seldom explored task. In this paper, we firstly study the unsupervised abnormal event detection problem in AHIN. The events are considered as star-schema instances of AHIN and are further modeled by hypergraphs. A novel hypergraph contrastive learning method, named AEHCL, is proposed to fully capture abnormal event patterns. AEHCL designs the intra-event and inter-event contrastive modules to exploit self-supervised AHIN information. The intra-event contrastive module captures the pair-wise and multivariate interaction anomalies within an event, and the inter-event module captures the contextual anomalies among events. These two modules collaboratively boost the performance of each other and improve the detection results. During the testing phase, a contrastive learning-based abnormal event score function is further proposed to measure the abnormality degree of events. Extensive experiments on three datasets in different scenarios demonstrate the effectiveness of AEHCL, and the results improve state-of-the-art baselines up to 12.0% in Average Precision (AP) and 4.6% in Area Under Curve (AUC) respectively.

Xiaochen Wang, Junyu Luo, Jiaqi Wang et al.

Pretraining has proven to be a powerful technique in natural language processing (NLP), exhibiting remarkable success in various NLP downstream tasks. However, in the medical domain, existing pretrained models on electronic health records (EHR) fail to capture the hierarchical nature of EHR data, limiting their generalization capability across diverse downstream tasks using a single pretrained model. To tackle this challenge, this paper introduces a novel, general, and unified pretraining framework called MEDHMP, specifically designed for hierarchically multimodal EHR data. The effectiveness of the proposed MEDHMP is demonstrated through experimental results on eight downstream tasks spanning three levels. Comparisons against eighteen baselines further highlight the efficacy of our approach.

Jiaqi Wang, Xiaochen Wang, Lingjuan Lyu et al.

This study introduces the Federated Medical Knowledge Injection (FEDMEKI) platform, a new benchmark designed to address the unique challenges of integrating medical knowledge into foundation models under privacy constraints. By leveraging a cross-silo federated learning approach, FEDMEKI circumvents the issues associated with centralized data collection, which is often prohibited under health regulations like the Health Insurance Portability and Accountability Act (HIPAA) in the USA. The platform is meticulously designed to handle multi-site, multi-modal, and multi-task medical data, which includes 7 medical modalities, including images, signals, texts, laboratory test results, vital signs, input variables, and output variables. The curated dataset to validate FEDMEKI covers 8 medical tasks, including 6 classification tasks (lung opacity detection, COVID-19 detection, electrocardiogram (ECG) abnormal detection, mortality prediction, sepsis prediction, and enlarged cardiomediastinum detection) and 2 generation tasks (medical visual question answering (MedVQA) and ECG noise clarification). This comprehensive dataset is partitioned across several clients to facilitate the decentralized training process under 16 benchmark approaches. FEDMEKI not only preserves data privacy but also enhances the capability of medical foundation models by allowing them to learn from a broader spectrum of medical knowledge without direct data exposure, thereby setting a new benchmark in the application of foundation models within the healthcare sector.

Xiaochen Wang, Jiaqi Wang, Houping Xiao et al.

Foundation models have demonstrated remarkable capabilities in handling diverse modalities and tasks, outperforming conventional artificial intelligence (AI) approaches that are highly task-specific and modality-reliant. In the medical domain, however, the development of comprehensive foundation models is constrained by limited access to diverse modalities and stringent privacy regulations. To address these constraints, this study introduces a novel knowledge injection approach, FedKIM, designed to scale the medical foundation model within a federated learning framework. FedKIM leverages lightweight local models to extract healthcare knowledge from private data and integrates this knowledge into a centralized foundation model using a designed adaptive Multitask Multimodal Mixture Of Experts (M3OE) module. This method not only preserves privacy but also enhances the model's ability to handle complex medical tasks involving multiple modalities. Our extensive experiments across twelve tasks in seven modalities demonstrate the effectiveness of FedKIM in various settings, highlighting its potential to scale medical foundation models without direct access to sensitive data.

Yuxuan Zhang, Yubo Wang, Yipeng Zhu et al.

AI agents may be able to automate your inbox, but can they automate other routine aspects of your life? Everyday online tasks offer a realistic yet unsolved testbed for evaluating the next generation of AI agents. To this end, we introduce ClawBench, an evaluation framework of 153 simple tasks that people need to accomplish regularly in their lives and work, spanning 144 live platforms across 15 categories, from completing purchases and booking appointments to submitting job applications. These tasks require demanding capabilities beyond existing benchmarks, such as obtaining relevant information from user-provided documents, navigating multi-step workflows across diverse platforms, and write-heavy operations like filling in many detailed forms correctly. Unlike existing benchmarks that evaluate agents in offline sandboxes with static pages, ClawBench operates on production websites, preserving the full complexity, dynamic nature, and challenges of real-world web interaction. A lightweight interception layer captures and blocks only the final submission request, ensuring safe evaluation without real-world side effects. Our evaluations of 7 frontier models show that both proprietary and open-source models can complete only a small portion of these tasks. For example, Claude Sonnet 4.6 achieves only 33.3%. Progress on ClawBench brings us closer to AI agents that can function as reliable general-purpose assistants.

Xiaochen Wang, Xiao Xiao, Ruhan Zhang et al.

Efficient and accurate product relevance assessment is critical for user experiences and business success. Training a proficient relevance assessment model requires high-quality query-product pairs, often obtained through negative sampling strategies. Unfortunately, current methods introduce pooling bias by mistakenly sampling false negatives, diminishing performance and business impact. To address this, we present Bias-mitigating Hard Negative Sampling (BHNS), a novel negative sampling strategy tailored to identify and adjust for false negatives, building upon our original False Negative Estimation algorithm. Our experiments in the Instacart search setting confirm BHNS as effective for practical e-commerce use. Furthermore, comparative analyses on public dataset showcase its domain-agnostic potential for diverse applications.

Xiaokun Zhang, Bo Xu, Zhaochun Ren et al.

Session-based recommendation aims to predict intents of anonymous users based on their limited behaviors. Modeling user behaviors involves two distinct rationales: co-occurrence patterns reflected by item IDs, and fine-grained preferences represented by item modalities (e.g., text and images). However, existing methods typically entangle these causes, leading to their failure in achieving accurate and explainable recommendations. To this end, we propose a novel framework DIMO to disentangle the effects of ID and modality in the task. At the item level, we introduce a co-occurrence representation schema to explicitly incorporate cooccurrence patterns into ID representations. Simultaneously, DIMO aligns different modalities into a unified semantic space to represent them uniformly. At the session level, we present a multi-view self-supervised disentanglement, including proxy mechanism and counterfactual inference, to disentangle ID and modality effects without supervised signals. Leveraging these disentangled causes, DIMO provides recommendations via causal inference and further creates two templates for generating explanations. Extensive experiments on multiple real-world datasets demonstrate the consistent superiority of DIMO over existing methods. Further analysis also confirms DIMO's effectiveness in generating explanations.

Xiangdi Meng, Damai Dai, Weiyao Luo et al. · pku

Supervised fine-tuning is the most common method to adapt large language models (LLMs) to downstream tasks, but full fine-tuning LLMs requires massive computational resources. Recently, parameter-efficient fine-tuning (PEFT) methods have been widely studied due to its cost-effectiveness. LoRA is one of the most widely used methods, which assumes that the optimization process is essentially low-dimensional. Although LoRA fine-tuning is effective, there is still a performance gap compared to full fine-tuning, since its weight update is limited to low-rank matrices. In order to break the low-rank bottleneck in LoRA Optimization, we propose PeriodicLoRA (PLoRA), which accumulates low-rank update matrices multiple times to achieve a higher update rank. PLoRA has multiple training stages. During each stage, we still update only the LoRA weights. However, at the end of each stage, we unload the LoRA weights into the backbone parameters and then reinitialize the LoRA states. Experimental results show that PLoRA has stronger learning ability, approximately 1.8 times that of LoRA's learning ability at most, but it does not increase memory usage. Further, we introduce a momentum-based unloading strategy for PLoRA to mitigate the training instability.

Junyu Luo, Xiaochen Wang, Jiaqi Wang et al.

Automatic International Classification of Diseases (ICD) coding plays a crucial role in the extraction of relevant information from clinical notes for proper recording and billing. One of the most important directions for boosting the performance of automatic ICD coding is modeling ICD code relations. However, current methods insufficiently model the intricate relationships among ICD codes and often overlook the importance of context in clinical notes. In this paper, we propose a novel approach, a contextualized and flexible framework, to enhance the learning of ICD code representations. Our approach, unlike existing methods, employs a dependent learning paradigm that considers the context of clinical notes in modeling all possible code relations. We evaluate our approach on six public ICD coding datasets and the experimental results demonstrate the effectiveness of our approach compared to state-of-the-art baselines.

Yihua Shao, Haojin He, Sijie Li et al.

Video Anomaly Detection~(VAD) focuses on identifying anomalies within videos. Supervised methods require an amount of in-domain training data and often struggle to generalize to unseen anomalies. In contrast, training-free methods leverage the intrinsic world knowledge of large language models (LLMs) to detect anomalies but face challenges in localizing fine-grained visual transitions and diverse events. Therefore, we propose EventVAD, an event-aware video anomaly detection framework that combines tailored dynamic graph architectures and multimodal LLMs through temporal-event reasoning. Specifically, EventVAD first employs dynamic spatiotemporal graph modeling with time-decay constraints to capture event-aware video features. Then, it performs adaptive noise filtering and uses signal ratio thresholding to detect event boundaries via unsupervised statistical features. The statistical boundary detection module reduces the complexity of processing long videos for MLLMs and improves their temporal reasoning through event consistency. Finally, it utilizes a hierarchical prompting strategy to guide MLLMs in performing reasoning before determining final decisions. We conducted extensive experiments on the UCF-Crime and XD-Violence datasets. The results demonstrate that EventVAD with a 7B MLLM achieves state-of-the-art (SOTA) in training-free settings, outperforming strong baselines that use 7B or larger MLLMs.

Jiaqi Wang, Junyu Luo, Muchao Ye et al.

The development of electronic health records (EHR) systems has enabled the collection of a vast amount of digitized patient data. However, utilizing EHR data for predictive modeling presents several challenges due to its unique characteristics. With the advancements in machine learning techniques, deep learning has demonstrated its superiority in various applications, including healthcare. This survey systematically reviews recent advances in deep learning-based predictive models using EHR data. Specifically, we begin by introducing the background of EHR data and providing a mathematical definition of the predictive modeling task. We then categorize and summarize predictive deep models from multiple perspectives. Furthermore, we present benchmarks and toolkits relevant to predictive modeling in healthcare. Finally, we conclude this survey by discussing open challenges and suggesting promising directions for future research.

Yiyang Du, Xiaochen Wang, Chi Chen et al. · tsinghua

Recently, model merging methods have demonstrated powerful strengths in combining abilities on various tasks from multiple Large Language Models (LLMs). While previous model merging methods mainly focus on merging homogeneous models with identical architecture, they meet challenges when dealing with Multimodal Large Language Models (MLLMs) with inherent heterogeneous property, including differences in model architecture and the asymmetry in the parameter space. In this work, we propose AdaMMS, a novel model merging method tailored for heterogeneous MLLMs. Our method tackles the challenges in three steps: mapping, merging and searching. Specifically, we first design mapping function between models to apply model merging on MLLMs with different architecture. Then we apply linear interpolation on model weights to actively adapt the asymmetry in the heterogeneous MLLMs. Finally in the hyper-parameter searching step, we propose an unsupervised hyper-parameter selection method for model merging. As the first model merging method capable of merging heterogeneous MLLMs without labeled data, extensive experiments on various model combinations demonstrated that AdaMMS outperforms previous model merging methods on various vision-language benchmarks.

Xiaochen Wang, Heming Xia, Jialin Song et al. · pku

Large Multimodal Models (LMMs) have achieved remarkable success across various visual-language tasks. However, existing benchmarks predominantly focus on single-image understanding, leaving the analysis of image sequences largely unexplored. To address this limitation, we introduce StripCipher, a comprehensive benchmark designed to evaluate capabilities of LMMs to comprehend and reason over sequential images. StripCipher comprises a human-annotated dataset and three challenging subtasks: visual narrative comprehension, contextual frame prediction, and temporal narrative reordering. Our evaluation of 16 state-of-the-art LMMs, including GPT-4o and Qwen2.5VL, reveals a significant performance gap compared to human capabilities, particularly in tasks that require reordering shuffled sequential images. For instance, GPT-4o achieves only 23.93% accuracy in the reordering subtask, which is 56.07% lower than human performance. Further quantitative analysis discuss several factors, such as input format of images, affecting the performance of LLMs in sequential understanding, underscoring the fundamental challenges that remain in the development of LMMs.

Junlong Li, Wenshuo Zhao, Jian Zhao et al. · cmu

Real-world language agents must handle complex, multi-step workflows across diverse Apps. For instance, an agent may manage emails by coordinating with calendars and file systems, or monitor a production database to detect anomalies and generate reports following an operating manual. However, existing language agent benchmarks often focus on narrow domains or simplified tasks that lack the diversity, realism, and long-horizon complexity required to evaluate agents' real-world performance. To address this gap, we introduce the Tool Decathlon (dubbed as Toolathlon), a benchmark for language agents offering diverse Apps and tools, realistic environment setup, and reliable execution-based evaluation. Toolathlon spans 32 software applications and 604 tools, ranging from everyday platforms such as Google Calendar and Notion to professional ones like WooCommerce, Kubernetes, and BigQuery. Most of the tools are based on a high-quality set of Model Context Protocol (MCP) servers that we may have revised or implemented ourselves. Unlike prior works, which primarily ensure functional realism but offer limited environment state diversity, we provide realistic initial environment states from real software, such as Canvas courses with dozens of students or real financial spreadsheets. This benchmark includes 108 manually sourced or crafted tasks in total, requiring interacting with multiple Apps over around 20 turns on average to complete. Each task is strictly verifiable through dedicated evaluation scripts. Comprehensive evaluation of SOTA models highlights their significant shortcomings: the best-performing model, Claude-4.5-Sonnet, achieves only a 38.6% success rate with 20.2 tool calling turns on average, while the top open-weights model DeepSeek-V3.2-Exp reaches 20.1%. We expect Toolathlon to drive the development of more capable language agents for real-world, long-horizon task execution.

Xiaochen Wang, Junqing He, Liang Chen et al. · pku

Large Language Models with chain-of-thought prompting, such as OpenAI-o1, have shown impressive capabilities in natural language inference tasks. However, Multi-hop Question Answering (MHQA) remains challenging for many existing models due to issues like hallucination, error propagation, and limited context length. To address these challenges and enhance LLMs' performance on MHQA, we propose the Self-Guiding prompting Finite State Machine (SG-FSM), designed to strengthen multi-hop reasoning abilities. Unlike traditional chain-of-thought methods, SG-FSM tackles MHQA by iteratively breaking down complex questions into sub-questions, correcting itself to improve accuracy. It processes one sub-question at a time, dynamically deciding the next step based on the current context and results, functioning much like an automaton. Experiments across various benchmarks demonstrate the effectiveness of our approach, outperforming strong baselines on challenging datasets such as Musique. SG-FSM reduces hallucination, enabling recovery of the correct final answer despite intermediate errors. It also improves adherence to specified output formats, simplifying evaluation significantly.

Yaqi Zhao, Xiaochen Wang, Li Dong et al.

Numerosity remains a challenge for state-of-the-art text-to-image generation models like FLUX and GPT-4o, which often fail to accurately follow counting instructions in text prompts. In this paper, we aim to study a fundamental yet often overlooked question: Can diffusion models inherently generate the correct number of objects specified by a textual prompt simply by scaling up the dataset and model size? To enable rigorous and reproducible evaluation, we construct a clean synthetic numerosity benchmark comprising two complementary datasets: GrayCount250 for controlled scaling studies, and NaturalCount6 featuring complex naturalistic scenes. Second, we empirically show that the scaling hypothesis does not hold: larger models and datasets alone fail to improve counting accuracy on our benchmark. Our analysis identifies a key reason: diffusion models tend to rely heavily on the noise initialization rather than the explicit numerosity specified in the prompt. We observe that noise priors exhibit biases toward specific object counts. In addition, we propose an effective strategy for controlling numerosity by injecting count-aware layout information into the noise prior. Our method achieves significant gains, improving accuracy on GrayCount250 from 20.0\% to 85.3\% and on NaturalCount6 from 74.8\% to 86.3\%, demonstrating effective generalization across settings.

Sai Wang, Senthilnathan Subramanian, Mudit Sahni et al.

Large-language-model (LLM) agents exhibit complex, context-sensitive behaviour that quickly renders static benchmarks and ad-hoc manual testing obsolete. We present Neo, a configurable, multi-agent framework that automates realistic, multi-turn evaluation of LLM-based systems. Neo couples a Question Generation Agent and an Evaluation Agent through a shared context-hub, allowing domain prompts, scenario controls and dynamic feedback to be composed modularly. Test inputs are sampled from a probabilistic state model spanning dialogue flow, user intent and emotional tone, enabling diverse, human-like conversations that adapt after every turn. Applied to a production-grade Seller Financial Assistant chatbot, Neo (i) uncovered edge-case failures across five attack categories with a 3.3% break rate close to the 5.8% achieved by expert human red-teamers, and (ii) delivered 10-12X higher throughput, generating 180 coherent test questions in around 45 mins versus 16h of human effort. Beyond security probing, Neo's stochastic policies balanced topic coverage and conversational depth, yielding broader behavioural exploration than manually crafted scripts. Neo therefore lays a foundation for scalable, self-evolving LLM QA: its agent interfaces, state controller and feedback loops are model-agnostic and extensible to richer factual-grounding and policy-compliance checks. We release the framework to facilitate reproducible, high-fidelity testing of emerging agentic systems.

Xiaochen Wang, Zongyu Wu, Yuan Zhong et al.

Graph retrieval-augmented generation (GRAG) places high demands on graph-specific retrievers. However, existing retrievers often rely on language models pretrained on plain text, limiting their effectiveness due to domain misalignment and structure ignorance. To address these challenges, we propose GPR, a graph-based retriever pretrained directly on knowledge graphs. GPR aligns natural language questions with relevant subgraphs through LLM-guided graph augmentation and employs a structure-aware objective to learn fine-grained retrieval strategies. Experiments on two datasets, three LLM backbones, and five baselines show that GPR consistently improves both retrieval quality and downstream generation, demonstrating its effectiveness as a robust retrieval solution for GRAG.

Xiaochen Wang, Yuan Zhong, Lingwei Zhang et al.

Medical deep learning models depend heavily on domain-specific knowledge to perform well on knowledge-intensive clinical tasks. Prior work has primarily leveraged unimodal knowledge graphs, such as the Unified Medical Language System (UMLS), to enhance model performance. However, integrating multimodal medical knowledge graphs remains largely underexplored, mainly due to the lack of resources linking imaging data with clinical concepts. To address this gap, we propose MEDMKG, a Medical Multimodal Knowledge Graph that unifies visual and textual medical information through a multi-stage construction pipeline. MEDMKG fuses the rich multimodal data from MIMIC-CXR with the structured clinical knowledge from UMLS, utilizing both rule-based tools and large language models for accurate concept extraction and relationship modeling. To ensure graph quality and compactness, we introduce Neighbor-aware Filtering (NaF), a novel filtering algorithm tailored for multimodal knowledge graphs. We evaluate MEDMKG across three tasks under two experimental settings, benchmarking twenty-four baseline methods and four state-of-the-art vision-language backbones on six datasets. Results show that MEDMKG not only improves performance in downstream medical tasks but also offers a strong foundation for developing adaptive and robust strategies for multimodal knowledge integration in medical artificial intelligence.

Yuan Zhong, Xiaochen Wang, Jiaqi Wang et al.

Synthesizing electronic health records (EHR) data has become a preferred strategy to address data scarcity, improve data quality, and model fairness in healthcare. However, existing approaches for EHR data generation predominantly rely on state-of-the-art generative techniques like generative adversarial networks, variational autoencoders, and language models. These methods typically replicate input visits, resulting in inadequate modeling of temporal dependencies between visits and overlooking the generation of time information, a crucial element in EHR data. Moreover, their ability to learn visit representations is limited due to simple linear mapping functions, thus compromising generation quality. To address these limitations, we propose a novel EHR data generation model called EHRPD. It is a diffusion-based model designed to predict the next visit based on the current one while also incorporating time interval estimation. To enhance generation quality and diversity, we introduce a novel time-aware visit embedding module and a pioneering predictive denoising diffusion probabilistic model (PDDPM). Additionally, we devise a predictive U-Net (PU-Net) to optimize P-DDPM.We conduct experiments on two public datasets and evaluate EHRPD from fidelity, privacy, and utility perspectives. The experimental results demonstrate the efficacy and utility of the proposed EHRPD in addressing the aforementioned limitations and advancing EHR data generation.

Arash Pakbin, Xiaochen Wang, Bobak J. Mortazavi et al.

Modern applications of survival analysis increasingly involve time-dependent covariates. The Python package BoXHED2.0 is a tree-boosted hazard estimator that is fully nonparametric, and is applicable to survival settings far more general than right-censoring, including recurring events and competing risks. BoXHED2.0 is also scalable to the point of being on the same order of speed as parametric boosted survival models, in part because its core is written in C++ and it also supports the use of GPUs and multicore CPUs. BoXHED2.0 is available from PyPI and also from www.github.com/BoXHED.

Xiaochen Wang, Arash Pakbin, Bobak J. Mortazavi et al.

The proliferation of medical monitoring devices makes it possible to track health vitals at high frequency, enabling the development of dynamic health risk scores that change with the underlying readings. Survival analysis, in particular hazard estimation, is well-suited to analyzing this stream of data to predict disease onset as a function of the time-varying vitals. This paper introduces the software package BoXHED (pronounced 'box-head') for nonparametrically estimating hazard functions via gradient boosting. BoXHED 1.0 is a novel tree-based implementation of the generic estimator proposed in Lee, Chen, Ishwaran (2017), which was designed for handling time-dependent covariates in a fully nonparametric manner. BoXHED is also the first publicly available software implementation for Lee, Chen, Ishwaran (2017). Applying BoXHED to cardiovascular disease onset data from the Framingham Heart Study reveals novel interaction effects among known risk factors, potentially resolving an open question in clinical literature.

Fang Zhang, Xiaochen Wang, Jingfei Han et al.

What are the most popular research topics in Artificial Intelligence (AI)? We formulate the problem as extracting top-$k$ topics that can best represent a given area with the help of knowledge base. We theoretically prove that the problem is NP-hard and propose an optimization model, FastKATE, to address this problem by combining both explicit and latent representations for each topic. We leverage a large-scale knowledge base (Wikipedia) to generate topic embeddings using neural networks and use this kind of representations to help capture the representativeness of topics for given areas. We develop a fast heuristic algorithm to efficiently solve the problem with a provable error bound. We evaluate the proposed model on three real-world datasets. Experimental results demonstrate our model's effectiveness, robustness, real-timeness (return results in $<1$s), and its superiority over several alternative methods.