AI Open最新文献

{"title":"Human professional level driving agent for race car simulation environments","authors":"Gergely Bári ,&nbsp;László Palkovics","doi":"10.1016/j.aiopen.2026.02.007","DOIUrl":"10.1016/j.aiopen.2026.02.007","url":null,"abstract":"<div><div>Precise vehicle control at the limits of tire adhesion is paramount for both competitive motorsport performance and the safe execution of emergency maneuvers in road vehicles. Mastering this “grip-limit driving” presents significant challenges due to highly non-linear vehicle dynamics and sensitivity to changing conditions, often exceeding the capabilities of traditional controllers and driver models. This paper investigates the efficacy of Deep Reinforcement Learning (DRL), specifically the Proximal Policy Optimisation (PPO) algorithm, as a data-driven approach to learn expert-level driving skills within the TORCS high-fidelity race car simulation environment. An agent was trained end-to-end, utilizing “realworld-friendly” state signals (such as speeds, accelerations, and yaw rate, simple LiDaR, etc.) as input to determine continuous steering and pedal commands. Notably, the trained Agent achieved lap times comparable to a human e-sport world champion on the target track, demonstrating the potential of this methodology while also highlighting how agents can exploit idealized simulation to achieve superhuman control. Furthermore, this work presents the formulation of the time-optimal driving task as a DRL problem and offers a novel justification for the commonly used “progress reward” function, demonstrating its conceptual link to the time-difference feedback mechanisms human drivers use for performance optimization. These findings provide valuable insights into AI-driven vehicle control under extreme conditions and contribute to the development of more capable autonomous agents for simulation and potentially, real-world applications.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 123-132"},"PeriodicalIF":14.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147540173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DualFlexKAN: Dual-stage Kolmogorov-Arnold Networks with independent function control","authors":"Andrés Ortiz,&nbsp;Nicolás J. Gallego-Molina,&nbsp;Carmen Jiménez-Mesa,&nbsp;Juan M. Górriz,&nbsp;Javier Ramírez","doi":"10.1016/j.aiopen.2026.08.003","DOIUrl":"10.1016/j.aiopen.2026.08.003","url":null,"abstract":"<div><div>Multi-Layer Perceptrons (MLPs) rely on pre-defined, fixed activation functions, imposing a static inductive bias that forces the network to approximate complex topologies solely through increased depth and width. Kolmogorov–Arnold Networks (KANs) address this limitation through edge-centric learnable functions, yet their formulation suffers from quadratic parameter scaling and architectural rigidity that hinders the effective integration of standard regularisation techniques. This paper presents DualFlexKAN (DFKAN), a flexible architecture that employs a dual-stage mechanism to independently manage pre-linear input transformations and post-linear output activations. This decoupling allows hybrid networks to more effectively balance expressiveness and computational cost. Unlike conventional formulations, DFKAN accommodates various basis function families, including orthogonal polynomials, B-splines, and radial basis functions, while offering configurable regularisation options that stabilise training dynamics. Comprehensive evaluations across regression benchmarks, physics-informed tasks, function approximation, higher-dimensional tabular data, and classification demonstrate that DFKAN outperforms both MLPs and conventional KANs in accuracy, convergence speed, and gradient fidelity on the regression and physics-informed tasks, while remaining competitive with far larger baselines at a fraction of their parameter cost on the classification benchmarks. The proposed hybrid configurations achieve superior performance with one to two orders of magnitude fewer parameters than standard KANs, effectively mitigating the parameter explosion problem while preserving KAN-style expressiveness. DFKAN provides a principled, scalable framework for incorporating adaptive non-linearities. The proposed design space formally subsumes recent lightweight KAN variants (such as Wav-KAN, Chebyshev-based formulations, and RBF-based formulations such as FastKAN) and coarse hybrid KAN-MLP architectures as special instances, whilst providing strictly finer-grained, per-layer control over the expressiveness-efficiency trade-off. The resulting architecture proves particularly advantageous for data-efficient learning and interpretable function discovery in scientific applications.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 227-247"},"PeriodicalIF":8.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148776835","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"WorldArena: A unified benchmark for evaluating perception and functional utility of embodied world models","authors":"Yu Shang,&nbsp;Zhuohang Li,&nbsp;Yiding Ma,&nbsp;Weikang Su,&nbsp;Xin Jin,&nbsp;Ziyou Wang,&nbsp;Lei Jin,&nbsp;Xin Zhang,&nbsp;Yinzhou Tang,&nbsp;Haisheng Su,&nbsp;Chen Gao,&nbsp;Wei Wu,&nbsp;Xihui Liu,&nbsp;Dhruv Shah,&nbsp;Zhaoxiang Zhang,&nbsp;Zhibo Chen,&nbsp;Jun Zhu,&nbsp;Yonghong Tian,&nbsp;Tat-Seng Chua,&nbsp;Xin Wang,&nbsp;Yong Li","doi":"10.1016/j.aiopen.2026.07.002","DOIUrl":"10.1016/j.aiopen.2026.07.002","url":null,"abstract":"<div><div>While world models have emerged as a cornerstone of embodied intelligence by enabling agents to reason about environmental dynamics through action-conditioned prediction, their evaluation remains fragmented. Current evaluation of embodied world models has largely focused on perceptual fidelity (e.g., video generation quality), overlooking the functional utility of these models in downstream decision-making tasks. In this work, we introduce WorldArena, a unified benchmark designed to systematically evaluate embodied world models across both perceptual and functional dimensions. WorldArena assesses models through three dimensions: video perception quality, measured with 16 metrics across six sub-dimensions; embodied task functionality, which evaluates world models as data engines, policy evaluators, and action planners integrating with subjective human evaluation. Furthermore, we propose EWMScore, a holistic metric integrating multi-dimensional performance into a single interpretable index. Through extensive experiments on 14 representative models, we reveal a significant perception–functionality gap, showing that high visual quality does not necessarily translate into strong embodied task capability. WorldArena benchmark with the public leaderboard is released at <span><span>https://world-arena.ai/</span><svg><path></path></svg></span>, providing a framework for tracking progress toward truly functional world models in embodied AI.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 208-226"},"PeriodicalIF":8.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148776834","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DCF: Uniformity optimization for dual collaborative filtering","authors":"Dan Zhang,&nbsp;Xi Wu,&nbsp;Yifan Zhu,&nbsp;Wei Wang,&nbsp;Chao Zhou,&nbsp;Yangli-ao Geng,&nbsp;Jibing Gong","doi":"10.1016/j.aiopen.2026.08.002","DOIUrl":"10.1016/j.aiopen.2026.08.002","url":null,"abstract":"<div><div>Recently, Alignment and Uniformity (AU) has been acknowledged as an effective strategy for enhancing user and item representation in collaborative filtering (CF) for recommender systems. Unlike traditional CF losses such as Bayesian Personalized Ranking (BPR), which rely on negative sampling, AU focuses on enforcing uniform distributions for user and item representations, aligning positive user–item pairs directly. However, existing AU-based methods have analytically discovered that the widely used “uniformity” loss does not always lead to a truly uniform distribution. It can end up in a local minimum, where representations are clustered within a low-dimensional subspace. This issue stems from overlooking the distinct statistical characteristics of user and item uniformity. This oversight fails to account for the significant variances in the number of users and items, along with their interaction distributions, necessitating different treatments. To tackle these challenges, there is a need to distinguish between user and item uniformity, as they should be treated differently due to their distinct statistical characteristics. Motivated by this, biased uniformity and redundancy reduction approaches are proposed and incorporated into AU-based CF loss, leading to the development of new objectives, referred to as dual collaborative filtering (DCF). The proposed DCF approach has shown state-of-the-art performance on widely used benchmarks, outperforming AU-based baselines, thus confirming its effectiveness.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 276-288"},"PeriodicalIF":8.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149004458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TRiSM for Agentic AI: A review of Trust, Risk, and Security Management in LLM-based Agentic Multi-Agent Systems","authors":"Shaina Raza ,&nbsp;Ranjan Sapkota ,&nbsp;Manoj Karkee ,&nbsp;Christos Emmanouilidis","doi":"10.1016/j.aiopen.2026.02.006","DOIUrl":"10.1016/j.aiopen.2026.02.006","url":null,"abstract":"<div><div>Agentic AI systems, built upon large language models (LLMs) and deployed in multi-agent configurations, are redefining intelligence, autonomy, collaboration, and decision-making across enterprise and societal domains. This review presents a structured analysis of Trust, Risk, and Security Management (TRiSM) in the context of LLM-based Agentic Multi-Agent Systems (AMAS). We begin by examining the conceptual foundations of Agentic AI and highlight its architectural distinctions from traditional AI agents. We then adapt and extend the AI TRiSM framework for Agentic AI, structured around key pillars: <em>Explainability, ModelOps, Security, Privacy</em> and <em>their Lifecycle Governance</em>, each contextualized to the challenges of AMAS. A risk taxonomy is proposed to capture the unique threats and vulnerabilities of Agentic AI, ranging from coordination failures to prompt-based adversarial manipulation. To make coordination and tool use measurable in practice, we propose two metrics: the Component Synergy Score (CSS), which captures inter-agent enablement, and the Tool Utilization Efficacy (TUE), which evaluates whether tools are invoked correctly and efficiently. We further discuss strategies for improving explainability in Agentic AI, as well as approaches to enhancing security and privacy through encryption, adversarial robustness, and regulatory compliance. The review concludes with a research roadmap for the responsible development and deployment of Agentic AI, highlighting key directions to align emerging systems with TRiSM principles-ensuring safety, transparency, and accountability in their operation.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 71-95"},"PeriodicalIF":14.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147404227","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A survey on generative recommendation: Data, model, and tasks","authors":"Min Hou,&nbsp;Le Wu,&nbsp;Yuxin Liao,&nbsp;Yonghui Yang,&nbsp;Zhen Zhang,&nbsp;Yu Wang,&nbsp;Changlong Zheng,&nbsp;Han Wu,&nbsp;Richang Hong","doi":"10.1016/j.aiopen.2026.05.002","DOIUrl":"10.1016/j.aiopen.2026.05.002","url":null,"abstract":"<div><div>Recommender systems serve as a foundational infrastructure in modern information ecosystems, helping users navigate the expanding digital content space and discover items aligned with their preferences. At their core, recommender systems address a fundamental research problem: matching users with items. Over the past decades, the field has experienced successive technological paradigm shifts, from collaborative filtering and matrix factorization in the machine learning era to sophisticated neural architectures in the deep learning era. Recently, the emergence of generative models, especially large language models (LLMs) and diffusion models have sparked a new paradigm: generative recommendation, which reconceptualizes the recommendation problem as a generation task rather than a discriminative scoring procedure. This survey provides a comprehensive examination of this paradigm through a unified tripartite framework spanning data, model, and task dimensions. Rather than simply categorizing works, we systematically decompose approaches into operational stages—data augmentation and unification, model alignment and training, task formulation and execution. At the data level, generative models enable knowledge-infused augmentation and agent-based simulation while unifying heterogeneous signals. At the model level, we taxonomize LLM-based methods, large recommendation models, and diffusion approaches, analyzing their alignment mechanisms and innovations. At the task level, we illuminate new capabilities including conversational interaction, explainable reasoning, and personalized content generation. We identify five key advantages: world knowledge integration, natural language understanding, reasoning capabilities, scaling laws, and creative generation. We critically examine challenges in benchmark design, model robustness, and deployment efficiency, while charting a roadmap toward intelligent recommendation assistants that fundamentally reshape human-information interaction.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 169-193"},"PeriodicalIF":8.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148656761","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"KGMedQA: An automated and comprehensive KG-based benchmark for biomedical LLM assessment","authors":"Qirui Hao,&nbsp;Kewei Cheng,&nbsp;Xinyu Liu,&nbsp;Hongliang Wang,&nbsp;Junming Shao,&nbsp;Carl Yang","doi":"10.1016/j.aiopen.2026.07.001","DOIUrl":"10.1016/j.aiopen.2026.07.001","url":null,"abstract":"<div><div>The emergence of large language models (LLMs) has significantly advanced natural language understanding, generation, and reasoning across various domains, including the biomedical field. Despite these advancements, the evaluation of biomedical LLMs remains limited, primarily relying on manually crafted datasets which are insufficient for comprehensively assessing LLMs’ capabilities. To address this challenge and cater specifically to the requirements of biomedical LLMs, we propose KGMedQA, an innovative evaluation benchmark based on knowledge graphs (KGs) designed to assess the knowledge and reasoning abilities of LLMs. Through careful alignment between natural language and KG structures, KGMedQA can be applied to arbitrary KGs, enabling automated question/answer generation and LLM evaluation. By leveraging the advantages of KG structures, we design seven tasks of varying complexity and focus, accompanied by specialized evaluation metrics. Experiments conducted with KGMedQA involve ten different LLMs, including general and specialized biomedical models, tested across two KGs focusing on different types of biomedical knowledge and reasoning. Compared to traditional methods, our results uncover more novel insights. For instance, while specialized models exhibit strengths in knowledge, they have deficiencies in reasoning abilities compared to general models. Additionally, factors such as model scales and prompting methods also impact the performance of LLMs. Our benchmark represents advancements in the evaluation of domain-specific LLMs, offering an effective tool for future research and development. Our source code is available at: <span><span>https://github.com/PerseidsMeteorShower/KGMedQA</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 152-168"},"PeriodicalIF":8.9,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148656768","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Which type of students can LLMs act? Investigating authentic simulation with graph-based Human–AI collaborative system","authors":"Haoxuan Li ,&nbsp;Jifan Yu ,&nbsp;Xin Cong ,&nbsp;Yang Dang ,&nbsp;Daniel Zhang-Li ,&nbsp;Lu Mi ,&nbsp;Yisi Zhan ,&nbsp;Huiqin Liu ,&nbsp;Zhiyuan Liu","doi":"10.1016/j.aiopen.2026.05.001","DOIUrl":"10.1016/j.aiopen.2026.05.001","url":null,"abstract":"<div><div>While rapid advances in large language models (LLMs) are reshaping data-driven intelligent education, accurately simulating students remains an important but challenging bottleneck for scalable educational data collection, evaluation, and intervention design. However, current works are limited by scarce real interaction data, costly expert evaluation for realism, and a lack of large-scale, systematic analyses of LLMs ability in simulating students. We address this gap by presenting a three-stage LLM-human collaborative pipeline to automatically generate and filter high-quality student agents. We leverage a two-round automated scoring validated by human experts and deploy a score propagation module to obtain more consistent scores across the student similarity graph. Experiments show that combining automated scoring, expert calibration, and graph-based propagation yields simulated student that more closely track authentication by human judgments. We then analyze which profiles and behaviors are simulated more faithfully, supporting subsequent studies on personalized learning and educational assessment.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 142-151"},"PeriodicalIF":14.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148167742","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GHOST 2.0: Generative high-fidelity one shot transfer of heads","authors":"Alexander Groshev ,&nbsp;Anastasiia Iashchenko ,&nbsp;Pavel Paramonov ,&nbsp;Denis Dimitrov ,&nbsp;Andrey Kuznetsov","doi":"10.1016/j.aiopen.2026.02.003","DOIUrl":"10.1016/j.aiopen.2026.02.003","url":null,"abstract":"<div><div>While the task of face swapping has recently gained attention in the research community, a related problem of head swapping remains largely unexplored. In addition to skin color transfer, head swap poses extra challenges, such as the need to preserve structural information of the whole head during synthesis and inpaint gaps between swapped head and background. In this paper, we address these concerns with GHOST 2.0, which consists of two problem-specific modules. First, we introduce enhanced Aligner model for head reenactment, which preserves identity information at multiple scales and is robust to extreme pose variations. Secondly, we use a Blender module that seamlessly integrates the reenacted head into the target background by transferring skin color and inpainting mismatched regions. Both modules outperform the baselines on the corresponding tasks, allowing to achieve state-of-the-art results in head swapping. We also tackle complex cases, such as large difference in hair styles of source and target.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 45-61"},"PeriodicalIF":14.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147404237","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating cross-view multi-scale perception and RAG-enabled expert fusion for medical prediction","authors":"Cheng Wang,&nbsp;Yongbin Liu,&nbsp;Ying Yu,&nbsp;Chunping Ouyang,&nbsp;Yaping Wan","doi":"10.1016/j.aiopen.2026.02.004","DOIUrl":"10.1016/j.aiopen.2026.02.004","url":null,"abstract":"<div><div>Electronic Health Records (EHRs) continuously monitor patients’ health status in Intensive Care Units (ICUs), capturing irregular numerical time-series data and unstructured clinical text. While existing studies primarily focus on handling modality irregularities, they often overlook the complex intra- and inter-sequence interactions as well as the dependencies between short-term and long-term features. Moreover, clinical notes are typically semantically sparse and structurally noisy, making them difficult to interpret. To address these challenges, we propose a novel multimodal predictive model. For irregular numerical time-series data, we design a cross-view multi-scale framework that integrates cross-attention mechanisms with multi-scale convolutions. This enables dynamic modeling of diverse temporal embeddings while precisely capturing intrinsic inter-variable interactions and cross-temporal dependencies, all with reduced computational complexity. For clinical text, we adopt a retrieval-augmented technique that leverages external medical knowledge graphs (KGs) and large language models (LLMs) to enrich text representations related to medical codes. These enhanced embeddings are then fused with clinical notes via a gated mechanism, effectively alleviating semantic sparsity. We validate the effectiveness of the proposed approach on two critical clinical prediction tasks. Experimental results show maximum relative F1 score improvements of 3.3%, 6.0%, and 3.4% for MISTS, clinical notes, and multimodal fusion tasks, respectively, demonstrating our method’s excellent medical predictive capability.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"7 ","pages":"Pages 62-70"},"PeriodicalIF":14.8,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147404238","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}