1. SciReasoner: Laying the Scientific Reasoning Ground Across Disciplines
Authors: Yizhou Wang, Chen Tang, Han Deng, Jiabei Xiao, Jiaqi Liu, Jianyu Wu, Jun Yao, Pengze Li, Encheng Su, Lintao Wang, Guohang Zhuang, Yuchen Ren, Ben Fei, Ming Hu, Xin Chen, Dongzhan Zhou, Junjun He, Xiangyu Yue, Zhenfei Yin, Jiamin Wu, Qihao Zheng, Yuhao Zhou, Huihui Xu, Chenglong Ma, Yan Lu, Wenlong Zhang, Chunfeng Song, Philip Torr, Shixiang Tang, Xinzhu Ma, Wanli Ouyang, Lei Bai •
Published: 2025-09-25 •
Source: arXiv
We present a scientific reasoning foundation model that aligns natural language with heterogeneous scientific representations. The model is pretrained on a 206B-token corpus spanning scientific text, pure sequences, and sequence-text pairs, then aligned via SFT on 40M instructions, annealed cold-start bootstrapping to elicit long-form chain-of-thought, and reinforcement learning with task-specific reward shaping, which instills deliberate scientific reasoning. It supports four capability families, covering up to 103 tasks across workflows: (i) faithful translation between text and scientific formats, (ii) text/knowledge extraction, (iii) property prediction, (iv) property classification, (v) unconditional and conditional sequence generation and design. Compared with specialist systems, our approach broadens instruction coverage, improves cross-domain generalization, and enhances fidelity. We detail data curation and training and show that cross-discipline learning strengthens transfer and downstream reliability. The model, instruct tuning datasets and the evaluation code are open-sourced at https://huggingface.co/SciReason and https://github.com/open-sciencelab/SciReason.
2. Nova: Real-Time Agentic Vision-Language Model Serving with Adaptive Cross-Stage Parallelization
Authors: Yuhang Xu, Shengzhong Liu, Dong Zhang, Bingheng Yan, Fan Wu, Guihai Chen •
Published: 2025-09-25 •
Source: arXiv
This paper presents Nova, a real-time scheduling framework for serving agentic vision-language models (VLMs) on a single GPU with balanced per-request latency and overall request process throughput. Our design begins by enabling effective pipelining across vision encode, LLM prefill, and LLM decode stages of VLMs, by exploiting their heterogeneous resource demands during execution and incorporating elastic GPU spatial partitioning among stages to maximally utilize the compute and memory resources. Building on this, we introduce a real-time scheduling algorithm that adaptively calibrates resource allocation among stages based on a Pareto-optimal analysis of the latency-throughput trade-off, allowing the system to sustain responsiveness and resource efficiency under dynamic request loads. To further alleviate GPU memory pressure, we design a lightweight weight offloading strategy for vision encoders that preserves inference efficiency with minimized memory overhead. Extensive evaluations on both synthetic and real-world agent workloads demonstrate that Nova consistently outperforms the state-of-the-art baselines, improving the maximum latency by up to 23.3%, while keeping competitive throughput.
3. Strong cohomological integrality for symmetric stacks
Authors: Lucien Hennecart, Tasuki Kinjo •
Published: 2025-09-25 •
Source: arXiv
We prove a strong form of the cohomological integrality theorem, decomposing the cohomology of smooth symmetric stacks as the cohomological Hall induction of the intersection cohomology of the good moduli spaces of stacks of graded points. This generalizes the previous result by the second author together with Bu--Davison--Ib\'a\~nez Nu\~nez--P\u{a}durariu to non-orthogonal stacks, and confirms a conjecture of the first author that the algebraic BPS cohomology of the quotient stack of a symmetric representation matches the intersection cohomology group whenever it is nonzero. As a consequence, we obtain a version of the cohomological integrality theorem for general 0-shifted symplectic stacks with good moduli spaces, as well as for the character stacks of general compact oriented $3$-manifolds with reductive gauge groups. As an application, we prove Halpern-Leistner's conjecture on the purity of the Borel--Moore homology of $0$-shifted symplectic stacks admitting proper good moduli spaces. Our proof combines a cohomological bound for the algebraic BPS cohomology, due to the first author and based on Efimov's lemma, with a vanishing-cycle argument due to the second author in collaboration with Bu--Davison--Ib\'a\~nez Nu\~nez--P\u{a}durariu.
4. Vision-Intelligence-Enabled Beam Tracking for Cross-Interface Water-Air Optical Wireless Communications
Authors: Tianqi Mao, Jiayue Liu, Weijie Liu, Dezhi Zheng, Zhaocheng Wang •
Published: 2025-09-25 •
Source: arXiv
The escalating development of oceanic applications like underwater surveillance and mineral exploration, is motivating real-time wireless backhaul of the considerable observation data. Such prospects can be hardly realized by the narrowband acoustic approach. Alternatively, optical wireless communication (OWC) has emerged as a promising solution for maritime and underwater applications due to its great potential for broadband underwater transmission. However, the implementations of water-air OWC can be rather challenging, especially when penetrating the fluctuating interface, where the direction of refracted signals changes dynamically, causing severe beam misalignment with airborne stations. This has necessitated real-time transceiver alignment adaptable to the sophisticated oceanic environment, which has yet to be addressed. Against this background, this paper establishes a mathematical channel model for water-air optical wireless transmission across the fluctuating sea surface. Based on the model, we propose a vision-based beam tracking algorithm that leverages artificial intelligence (AI) methods for dynamic channel prediction. The proposed algorithm integrates a convolutional neural network (CNN) with bi-directional long short-term memory (Bi-LSTM), which further incorporates the attention mechanism to effectively extract critical spatio-temporal features from the vision data. The numerical simulation results show that the proposed algorithm can outperform its classical counterparts in maintaining receiving signal strength and supressing the vision noises, which demonstrates its robustness against the the harsh conditions of water-air OWC systems.
5. Rings of non-commutative functions and their fields of fractions
Authors: Méric L. Augat, Robert T. W. Martin, Eli Shamovich •
Published: 2025-09-25 •
Source: arXiv
Semi-free ideal rings, or semifirs, were introduced by Paul M. Cohn to study universal localizations in the non-commutative setting. We provide new examples of semifirs consisting of analytic functions in several non-commuting variables. These examples arise canonically in free analysis by completing the free algebra in the topology of ``uniform convergence on operator-space balls'' in the non-commutative universe of tuples of square matrices of any finite size. We show, in particular, that the ring of (uniformly) entire non-commutative (NC) functions in $d \in \mathbb{N}$ non-commuting variables, $\scr{O}_d$, is a semifir. Every finitely--generated right (or left) ideal in $\scr{O}_d$ is closed, which yields an analytic extension of G. Bergman's nullstellensatz for the free algebra. Any semifir admits a universal skew field of fractions; applying this to $\scr{O}_d$ yields the universal skew field of ``NC meromorphic expressions", $\scr{M} _d$. We show that any $f \in \scr{M} _d$ has a well-defined domain and evaluations in a large class of stably-finite topological algebras, including finite $C^*$-algebras, extending a result of Cohn for NC rational functions. As an application, we extend the almost sure convergence result of Haagerup and Thorbj\"ornsen for free polynomials evaluated on tuples of random matrices to the setting of NC meromorphic expressions.
6. LLMTrace: A Corpus for Classification and Fine-Grained Localization of AI-Written Text
Authors: Irina Tolstykh, Aleksandra Tsybina, Sergey Yakubson, Maksim Kuprashevich •
Published: 2025-09-25 •
Source: arXiv
The widespread use of human-like text from Large Language Models (LLMs) necessitates the development of robust detection systems. However, progress is limited by a critical lack of suitable training data; existing datasets are often generated with outdated models, are predominantly in English, and fail to address the increasingly common scenario of mixed human-AI authorship. Crucially, while some datasets address mixed authorship, none provide the character-level annotations required for the precise localization of AI-generated segments within a text. To address these gaps, we introduce LLMTrace, a new large-scale, bilingual (English and Russian) corpus for AI-generated text detection. Constructed using a diverse range of modern proprietary and open-source LLMs, our dataset is designed to support two key tasks: traditional full-text binary classification (human vs. AI) and the novel task of AI-generated interval detection, facilitated by character-level annotations. We believe LLMTrace will serve as a vital resource for training and evaluating the next generation of more nuanced and practical AI detection models. The project page is available at \href{https://sweetdream779.github.io/LLMTrace-info/}{iitolstykh/LLMTrace}.
7. Grounding AI Explanations in Experience: A Reflective Cognitive Architecture for Clinical Decision Support
Authors: Zijian Shao, Haiyang Shen, Mugeng Liu, Gecheng Fu, Yaoqi Guo, Yanfeng Wang, Yun Ma •
Published: 2025-09-25 •
Source: arXiv
Effective disease prediction in modern healthcare demands the twin goals of high accuracy and transparent, clinically meaningful explanations. Existing machine learning and large language model (LLM) based approaches often struggle to balance these goals. Many models yield accurate but unclear statistical outputs, while others generate fluent but statistically unsupported narratives, often undermining both the validity of the explanation and the predictive accuracy itself. This shortcoming comes from a shallow interaction with the data, preventing the development of a deep, detailed understanding similar to a human expert's. We argue that high accuracy and high-quality explanations are not separate objectives but are mutually reinforcing outcomes of a model that develops a deep, direct understanding of the data. To achieve this, we propose the Reflective Cognitive Architecture (RCA), a novel framework that coordinates multiple LLMs to learn from direct experience. RCA features an iterative rule refinement mechanism that improves its logic from prediction errors and a distribution-aware rules check mechanism that bases its reasoning in the dataset's global statistics. By using predictive accuracy as a signal to drive deeper comprehension, RCA builds a strong internal model of the data. We evaluated RCA on one private and two public datasets against 22 baselines. The results demonstrate that RCA not only achieves state-of-the-art accuracy and robustness with a relative improvement of up to 40\% over the baseline but, more importantly, leverages this deep understanding to excel in generating explanations that are clear, logical, evidence-based, and balanced, highlighting its potential for creating genuinely trustworthy clinical decision support systems. The code is available at \https://github.com/ssssszj/RCA.
8. \LARGE GMP$^{3}$: Learning-Driven, Bellman-Guided Trajectory Planning for UAVs in Real-Time on SE(3)
Authors: Babak Salamat, Dominik Mattern, Sebastian-Sven Olzem, Gerhard Elsbacher, Christian Seidel, Andrea M. Tonello •
Published: 2025-09-25 •
Source: arXiv
We propose $\text{GMP}^{3}$, a multiphase global path planning framework that generates dynamically feasible three-dimensional trajectories for unmanned aerial vehicles (UAVs) operating in cluttered environments. The framework extends traditional path planning from Euclidean position spaces to the Lie group $\mathrm{SE}(3)$, allowing joint learning of translational motion and rotational dynamics. A modified Bellman-based operator is introduced to support reinforcement learning (RL) policy updates while leveraging prior trajectory information for improved convergence. $\text{GMP}^{3}$ is designed as a distributed framework in which agents influence each other and share policy information along the trajectory: each agent refines its assigned segment and shares with its neighbors via a consensus-based scheme, enabling cooperative policy updates and convergence toward a path shaped globally even under kinematic constraints. We also propose DroneManager, a modular ground control software that interfaces the planner with real UAV platforms via the MAVLink protocol, supporting real-time deployment and feedback. Simulation studies and indoor flight experiments validate the effectiveness of the proposed method in constrained 3D environments, demonstrating reliable obstacle avoidance and smooth, feasible trajectories across both position and orientation. The open-source implementation is available at https://github.com/Domattee/DroneManager
9. Dense Semantic Matching with VGGT Prior
Authors: Songlin Yang, Tianyi Wei, Yushi Lan, Zeqi Xiao, Anyi Rao, Xingang Pan •
Published: 2025-09-25 •
Source: arXiv
Semantic matching aims to establish pixel-level correspondences between instances of the same category and represents a fundamental task in computer vision. Existing approaches suffer from two limitations: (i) Geometric Ambiguity: Their reliance on 2D foundation model features (e.g., Stable Diffusion, DINO) often fails to disambiguate symmetric structures, requiring extra fine-tuning yet lacking generalization; (ii) Nearest-Neighbor Rule: Their pixel-wise matching ignores cross-image invisibility and neglects manifold preservation. These challenges call for geometry-aware pixel descriptors and holistic dense correspondence mechanisms. Inspired by recent advances in 3D geometric foundation models, we turn to VGGT, which provides geometry-grounded features and holistic dense matching capabilities well aligned with these needs. However, directly transferring VGGT is challenging, as it was originally designed for geometry matching within cross views of a single instance, misaligned with cross-instance semantic matching, and further hindered by the scarcity of dense semantic annotations. To address this, we propose an approach that (i) retains VGGT's intrinsic strengths by reusing early feature stages, fine-tuning later ones, and adding a semantic head for bidirectional correspondences; and (ii) adapts VGGT to the semantic matching scenario under data scarcity through cycle-consistent training strategy, synthetic data augmentation, and progressive training recipe with aliasing artifact mitigation. Extensive experiments demonstrate that our approach achieves superior geometry awareness, matching reliability, and manifold preservation, outperforming previous baselines.
10. A Causality-Aware Spatiotemporal Model for Multi-Region and Multi-Pollutant Air Quality Forecasting
Authors: Junxin Lu, Shiliang Sun •
Published: 2025-09-25 •
Source: arXiv
Air pollution, a pressing global problem, threatens public health, environmental sustainability, and climate stability. Achieving accurate and scalable forecasting across spatially distributed monitoring stations is challenging due to intricate multi-pollutant interactions, evolving meteorological conditions, and region specific spatial heterogeneity. To address this challenge, we propose AirPCM, a novel deep spatiotemporal forecasting model that integrates multi-region, multi-pollutant dynamics with explicit meteorology-pollutant causality modeling. Unlike existing methods limited to single pollutants or localized regions, AirPCM employs a unified architecture to jointly capture cross-station spatial correlations, temporal auto-correlations, and meteorology-pollutant dynamic causality. This empowers fine-grained, interpretable multi-pollutant forecasting across varying geographic and temporal scales, including sudden pollution episodes. Extensive evaluations on multi-scale real-world datasets demonstrate that AirPCM consistently surpasses state-of-the-art baselines in both predictive accuracy and generalization capability. Moreover, the long-term forecasting capability of AirPCM provides actionable insights into future air quality trends and potential high-risk windows, offering timely support for evidence-based environmental governance and carbon mitigation planning.
11. BiNoMaP: Learning Category-Level Bimanual Non-Prehensile Manipulation Primitives
Authors: Huayi Zhou, Kui Jia •
Published: 2025-09-25 •
Source: arXiv
Non-prehensile manipulation, encompassing ungraspable actions such as pushing, poking, and pivoting, represents a critical yet underexplored domain in robotics due to its contact-rich and analytically intractable nature. In this work, we revisit this problem from two novel perspectives. First, we move beyond the usual single-arm setup and the strong assumption of favorable external dexterity such as walls, ramps, or edges. Instead, we advocate a generalizable dual-arm configuration and establish a suite of Bimanual Non-prehensile Manipulation Primitives (BiNoMaP). Second, we depart from the prevailing RL-based paradigm and propose a three-stage, RL-free framework to learn non-prehensile skills. Specifically, we begin by extracting bimanual hand motion trajectories from video demonstrations. Due to visual inaccuracies and morphological gaps, these coarse trajectories are difficult to transfer directly to robotic end-effectors. To address this, we propose a geometry-aware post-optimization algorithm that refines raw motions into executable manipulation primitives that conform to specific motion patterns. Beyond instance-level reproduction, we further enable category-level generalization by parameterizing the learned primitives with object-relevant geometric attributes, particularly size, resulting in adaptable and general parameterized manipulation primitives. We validate BiNoMaP across a range of representative bimanual tasks and diverse object categories, demonstrating its effectiveness, efficiency, versatility, and superior generalization capability.
12. Federated Flow Matching
Authors: Zifan Wang, Anqi Dong, Mahmoud Selim, Michael M. Zavlanos, Karl H. Johansson •
Published: 2025-09-25 •
Source: arXiv
Data today is decentralized, generated and stored across devices and institutions where privacy, ownership, and regulation prevent centralization. This motivates the need to train generative models directly from distributed data locally without central aggregation. In this paper, we introduce Federated Flow Matching (FFM), a framework for training flow matching models under privacy constraints. Specifically, we first examine FFM-vanilla, where each client trains locally with independent source and target couplings, preserving privacy but yielding curved flows that slow inference. We then develop FFM-LOT, which employs local optimal transport couplings to improve straightness within each client but lacks global consistency under heterogeneous data. Finally, we propose FFM-GOT, a federated strategy based on the semi-dual formulation of optimal transport, where a shared global potential function coordinates couplings across clients. Experiments on synthetic and image datasets show that FFM enables privacy-preserving training while enhancing both the flow straightness and sample quality in federated settings, with performance comparable to the centralized baseline.
13. Decipher-MR: A Vision-Language Foundation Model for 3D MRI Representations
Authors: Zhijian Yang, Noel DSouza, Istvan Megyeri, Xiaojian Xu, Amin Honarmandi Shandiz, Farzin Haddadpour, Krisztian Koos, Laszlo Rusko, Emanuele Valeriano, Bharadwaj Swaninathan, Lei Wu, Parminder Bhatia, Taha Kass-Hout, Erhan Bas •
Published: 2025-09-25 •
Source: arXiv
Magnetic Resonance Imaging (MRI) is a critical medical imaging modality in clinical diagnosis and research, yet its complexity and heterogeneity pose challenges for automated analysis, particularly in scalable and generalizable machine learning applications. While foundation models have revolutionized natural language and vision tasks, their application to MRI remains limited due to data scarcity and narrow anatomical focus. In this work, we present Decipher-MR, a 3D MRI-specific vision-language foundation model trained on a large-scale dataset comprising 200,000 MRI series from over 22,000 studies spanning diverse anatomical regions, sequences, and pathologies. Decipher-MR integrates self-supervised vision learning with report-guided text supervision to build robust, generalizable representations, enabling effective adaptation across broad applications. To enable robust and diverse clinical tasks with minimal computational overhead, Decipher-MR supports a modular design that enables tuning of lightweight, task-specific decoders attached to a frozen pretrained encoder. Following this setting, we evaluate Decipher-MR across diverse benchmarks including disease classification, demographic prediction, anatomical localization, and cross-modal retrieval, demonstrating consistent performance gains over existing foundation models and task-specific approaches. Our results establish Decipher-MR as a scalable and versatile foundation for MRI-based AI, facilitating efficient development across clinical and research domains.
14. Hunyuan3D-Omni: A Unified Framework for Controllable Generation of 3D Assets
Authors: Team Hunyuan3D, :, Bowen Zhang, Chunchao Guo, Haolin Liu, Hongyu Yan, Huiwen Shi, Jingwei Huang, Junlin Yu, Kunhong Li, Linus, Penghao Wang, Qingxiang Lin, Sicong Liu, Xianghui Yang, Yixuan Tang, Yunfei Zhao, Zeqiang Lai, Zhihao Liang, Zibo Zhao •
Published: 2025-09-25 •
Source: arXiv
Recent advances in 3D-native generative models have accelerated asset creation for games, film, and design. However, most methods still rely primarily on image or text conditioning and lack fine-grained, cross-modal controls, which limits controllability and practical adoption. To address this gap, we present Hunyuan3D-Omni, a unified framework for fine-grained, controllable 3D asset generation built on Hunyuan3D 2.1. In addition to images, Hunyuan3D-Omni accepts point clouds, voxels, bounding boxes, and skeletal pose priors as conditioning signals, enabling precise control over geometry, topology, and pose. Instead of separate heads for each modality, our model unifies all signals in a single cross-modal architecture. We train with a progressive, difficulty-aware sampling strategy that selects one control modality per example and biases sampling toward harder signals (e.g., skeletal pose) while downweighting easier ones (e.g., point clouds), encouraging robust multi-modal fusion and graceful handling of missing inputs. Experiments show that these additional controls improve generation accuracy, enable geometry-aware transformations, and increase robustness for production workflows.
15. Tree Search for LLM Agent Reinforcement Learning
Authors: Yuxiang Ji, Ziyu Ma, Yong Wang, Guanhua Chen, Xiangxiang Chu, Liaoni Wu •
Published: 2025-09-25 •
Source: arXiv
Recent advances in reinforcement learning (RL) have significantly enhanced the agentic capabilities of large language models (LLMs). In long-term and multi-turn agent tasks, existing approaches driven solely by outcome rewards often suffer from the problem of sparse supervision. To address the challenge, we propose Tree-based Group Relative Policy Optimization (Tree-GRPO), a grouped agent RL method based on tree search, where each tree node represents the complete agent interaction step. By sharing common prefixes, the tree search sampling increases the number of rollouts achievable within a fixed budget of tokens or tool calls. Moreover, we find that the tree-structured trajectory naturally allows the construction of step-wise process supervised signals even using only the outcome reward. Based on this, Tree-GRPO estimates the grouped relative advantages both on intra-tree and inter-tree levels. Through theoretical analysis, we demonstrate that the objective of intra-tree level group relative policy optimization is equivalent to that of step-level direct preference learning. Experiments across 11 datasets and 3 types of QA tasks demonstrate the superiority of the proposed tree-based RL over the chain-based RL method.
16. Double Poisson (vertex) algebra cohomology
Authors: Maxime Fairon, Daniele Valeri •
Published: 2025-09-25 •
Source: arXiv
A noncommutative (NC) version of Poisson geometry was initiated by Van den Bergh by introducing at the level of associative algebras the formalism of double Poisson brackets. Their key property is to induce (standard) Poisson brackets under each representation functor. Then, Pichereau and Van de Weyer developed and studied the corresponding cohomology theory under the assumption that there exists a NC bivector defining the double Poisson bracket. Our first main result is that one can remove this assumption by constructing a completed double Poisson cohomology valid in any situation, hence generalizing the approach of Pichereau-Van de Weyer. As an application, we show that the double Poisson cohomology complex associated to the path algebra of a quiver is acyclic. Furthermore, we show that this new double Poisson cohomology theory can be adapted to weaker forms of double Poisson brackets (called quasi-Poisson and gauged Poisson), and that it is compatible with representation functors. A second focus of this memoir concerns the formalism of double Poisson vertex algebras. These were introduced by De Sole, Kac and the second author, as NC versions of Poisson vertex algebras, which induce the latter structures under each representation functor. Our second main result is the development of cohomology theories for double Poisson vertex algebras. These are NC analogues of the basic, reduced and variational Poisson vertex algebra cohomologies. More importantly, we prove that under each representation functor these cohomology theories are compatible with their commutative counterparts. As an application, we compute the double Poisson vertex algebra cohomology of the generalized NC de Rham complex and of the generalized NC variational complex. Finally, we describe the relation between the double Poisson algebra and double Poisson vertex algebra cohomologies using jet and quotient functors.
17. SEEC: Stable End-Effector Control with Model-Enhanced Residual Learning for Humanoid Loco-Manipulation
Authors: Jaehwi Jang, Zhuoheng Wang, Ziyi Zhou, Feiyang Wu, Ye Zhao •
Published: 2025-09-25 •
Source: arXiv
Arm end-effector stabilization is essential for humanoid loco-manipulation tasks, yet it remains challenging due to the high degrees of freedom and inherent dynamic instability of bipedal robot structures. Previous model-based controllers achieve precise end-effector control but rely on precise dynamics modeling and estimation, which often struggle to capture real-world factors (e.g., friction and backlash) and thus degrade in practice. On the other hand, learning-based methods can better mitigate these factors via exploration and domain randomization, and have shown potential in real-world use. However, they often overfit to training conditions, requiring retraining with the entire body, and still struggle to adapt to unseen scenarios. To address these challenges, we propose a novel stable end-effector control (SEEC) framework with model-enhanced residual learning that learns to achieve precise and robust end-effector compensation for lower-body induced disturbances through model-guided reinforcement learning (RL) with a perturbation generator. This design allows the upper-body policy to achieve accurate end-effector stabilization as well as adapt to unseen locomotion controllers with no additional training. We validate our framework in different simulators and transfer trained policies to the Booster T1 humanoid robot. Experiments demonstrate that our method consistently outperforms baselines and robustly handles diverse and demanding loco-manipulation tasks.
18. From Physics to Machine Learning and Back: Part II - Learning and Observational Bias in PHM
Authors: Olga Fink, Ismail Nejjar, Vinay Sharma, Keivan Faghih Niresi, Han Sun, Hao Dong, Chenghao Xu, Amaury Wei, Arthur Bizzi, Raffael Theiler, Yuan Tian, Leandro Von Krannichfeldt, Zhan Ma, Sergei Garmaev, Zepeng Zhang, Mengjie Zhao •
Published: 2025-09-25 •
Source: arXiv
Prognostics and Health Management ensures the reliability, safety, and efficiency of complex engineered systems by enabling fault detection, anticipating equipment failures, and optimizing maintenance activities throughout an asset lifecycle. However, real-world PHM presents persistent challenges: sensor data is often noisy or incomplete, available labels are limited, and degradation behaviors and system interdependencies can be highly complex and nonlinear. Physics-informed machine learning has emerged as a promising approach to address these limitations by embedding physical knowledge into data-driven models. This review examines how incorporating learning and observational biases through physics-informed modeling and data strategies can guide models toward physically consistent and reliable predictions. Learning biases embed physical constraints into model training through physics-informed loss functions and governing equations, or by incorporating properties like monotonicity. Observational biases influence data selection and synthesis to ensure models capture realistic system behavior through virtual sensing for estimating unmeasured states, physics-based simulation for data augmentation, and multi-sensor fusion strategies. The review then examines how these approaches enable the transition from passive prediction to active decision-making through reinforcement learning, which allows agents to learn maintenance policies that respect physical constraints while optimizing operational objectives. This closes the loop between model-based predictions, simulation, and actual system operation, empowering adaptive decision-making. Finally, the review addresses the critical challenge of scaling PHM solutions from individual assets to fleet-wide deployment. Fast adaptation methods including meta-learning and few-shot learning are reviewed alongside domain generalization techniques ...
19. AI-Enhanced Multi-Dimensional Measurement of Technological Convergence through Heterogeneous Graph and Semantic Learning
Authors: Siming Deng, Runsong Jia, Chunjuan Luan, Mengjia Wu, Yi Zhang •
Published: 2025-09-25 •
Source: arXiv
Technological convergence refers to the phenomenon where boundaries between technological areas and disciplines are increasingly blurred. It enables the integration of previously distinct domains and has become a mainstream trend in today's innovation process. However, accurately measuring technological convergence remains a persistent challenge due to its inherently multidimensional and evolving nature. This study designs an Technological Convergence Index (TCI) that comprehensively measures convergence along two fundamental dimensions: depth and breadth. For depth calculation, we use IPC textual descriptions as the analytical foundation and enhance this assessment by incorporating supplementary patent metadata into a heterogeneous graph structure. This graph is then modeled using Heterogeneous Graph Transformers in combination with Sentence-BERT, enabling a precise representation of knowledge integration across technological boundaries. Complementing this, the breadth dimension captures the diversity of technological fields involved, quantified through the Shannon Diversity Index to measure the variety of technological combinations within patents. Our final TCI is constructed using the Entropy Weight Method, which objectively assigns weights to both dimensions based on their information entropy. To validate our approach, we compare the proposed TCI against established convergence measures, demonstrating its comparative advantages. We further establish empirical reliability through a novel robustness test that regresses TCI against indicators of patent quality. These findings are further substantiated through comprehensive robustness checks. Our multidimensional approach provides valuable practical insights for innovation policy and industry strategies in managing emerging cross-domain technologies.
20. Fine-Tuning LLMs to Analyze Multiple Dimensions of Code Review: A Maximum Entropy Regulated Long Chain-of-Thought Approach
Authors: Yongda Yu, Guohao Shi, Xianwei Wu, Haochuan He, XueMing Gu, Qianqian Zhao, Kui Liu, Qiushi Wang, Zhao Tian, Haifeng Shen, Guoping Rong •
Published: 2025-09-25 •
Source: arXiv
Large Language Models (LLMs) have shown great potential in supporting automated code review due to their impressive capabilities in context understanding and reasoning. However, these capabilities are still limited compared to human-level cognition because they are heavily influenced by the training data. Recent research has demonstrated significantly improved performance through fine-tuning LLMs with code review data. However, compared to human reviewers who often simultaneously analyze multiple dimensions of code review to better identify issues, the full potential of these methods is hampered by the limited or vague information used to fine-tune the models. This paper contributes MelcotCR, a chain-of-thought (COT) fine-tuning approach that trains LLMs with an impressive reasoning ability to analyze multiple dimensions of code review by harnessing long COT techniques to provide rich structured information. To address context loss and reasoning logic loss issues that frequently occur when LLMs process long COT prompts, we propose a solution that combines the Maximum Entropy (ME) modeling principle with pre-defined reasoning pathways in MelcotCR to enable more effective utilization of in-context knowledge within long COT prompts while strengthening the logical tightness of the reasoning process. Empirical evaluations on our curated MelcotCR dataset and the public CodeReviewer dataset reveal that a low-parameter base model, such as 14B Qwen2.5, fine-tuned with MelcotCR can surpass state-of-the-art methods in terms of the accuracy of detecting and describing code issues, with its performance remarkably on par with that of the 671B DeepSeek-R1 model.
21. LAVA: Explainability for Unsupervised Latent Embeddings
Authors: Ivan Stresec, Joana P. Gonçalves •
Published: 2025-09-25 •
Source: arXiv
Unsupervised black-box models can be drivers of scientific discovery, but remain difficult to interpret. Crucially, discovery hinges on understanding the model output, which is often a multi-dimensional latent embedding rather than a well-defined target. While explainability for supervised learning usually seeks to uncover how input features are used to predict a target, its unsupervised counterpart should relate input features to the structure of the learned latent space. Adaptations of supervised model explainability for unsupervised learning provide either single-sample or dataset-wide summary explanations. However, without automated strategies of relating similar samples to one another guided by their latent proximity, explanations remain either too fine-grained or too reductive to be meaningful. This is especially relevant for manifold learning methods that produce no mapping function, leaving us only with the relative spatial organization of their embeddings. We introduce Locality-Aware Variable Associations (LAVA), a post-hoc model-agnostic method designed to explain local embedding organization through its relationship with the input features. To achieve this, LAVA represents the latent space as a series of localities (neighborhoods) described in terms of correlations between the original features, and then reveals reoccurring patterns of correlations across the entire latent space. Based on UMAP embeddings of MNIST and a single-cell kidney dataset, we show that LAVA captures relevant feature associations, with visually and biologically relevant local patterns shared among seemingly distant regions of the latent spaces.
22. Automotive-ENV: Benchmarking Multimodal Agents in Vehicle Interface Systems
Authors: Junfeng Yan, Biao Wu, Meng Fang, Ling Chen •
Published: 2025-09-25 •
Source: arXiv
Multimodal agents have demonstrated strong performance in general GUI interactions, but their application in automotive systems has been largely unexplored. In-vehicle GUIs present distinct challenges: drivers' limited attention, strict safety requirements, and complex location-based interaction patterns. To address these challenges, we introduce Automotive-ENV, the first high-fidelity benchmark and interaction environment tailored for vehicle GUIs. This platform defines 185 parameterized tasks spanning explicit control, implicit intent understanding, and safety-aware tasks, and provides structured multimodal observations with precise programmatic checks for reproducible evaluation. Building on this benchmark, we propose ASURADA, a geo-aware multimodal agent that integrates GPS-informed context to dynamically adjust actions based on location, environmental conditions, and regional driving norms. Experiments show that geo-aware information significantly improves success on safety-aware tasks, highlighting the importance of location-based context in automotive environments. We will release Automotive-ENV, complete with all tasks and benchmarking tools, to further the development of safe and adaptive in-vehicle agents.
23. EvoMail: Self-Evolving Cognitive Agents for Adaptive Spam and Phishing Email Defense
Authors: Wei Huang, De-Tian Chu, Lin-Yuan Bai, Wei Kang, Hai-Tao Zhang, Bo Li, Zhi-Mo Han, Jing Ge, Hai-Feng Lin •
Published: 2025-09-25 •
Source: arXiv
Modern email spam and phishing attacks have evolved far beyond keyword blacklists or simple heuristics. Adversaries now craft multi-modal campaigns that combine natural-language text with obfuscated URLs, forged headers, and malicious attachments, adapting their strategies within days to bypass filters. Traditional spam detection systems, which rely on static rules or single-modality models, struggle to integrate heterogeneous signals or to continuously adapt, leading to rapid performance degradation. We propose EvoMail, a self-evolving cognitive agent framework for robust detection of spam and phishing. EvoMail first constructs a unified heterogeneous email graph that fuses textual content, metadata (headers, senders, domains), and embedded resources (URLs, attachments). A Cognitive Graph Neural Network enhanced by a Large Language Model (LLM) performs context-aware reasoning across these sources to identify coordinated spam campaigns. Most critically, EvoMail engages in an adversarial self-evolution loop: a ''red-team'' agent generates novel evasion tactics -- such as character obfuscation or AI-generated phishing text -- while the ''blue-team'' detector learns from failures, compresses experiences into a memory module, and reuses them for future reasoning. Extensive experiments on real-world datasets (Enron-Spam, Ling-Spam, SpamAssassin, and TREC) and synthetic adversarial variants demonstrate that EvoMail consistently outperforms state-of-the-art baselines in detection accuracy, adaptability to evolving spam tactics, and interpretability of reasoning traces. These results highlight EvoMail's potential as a resilient and explainable defense framework against next-generation spam and phishing threats.
24. Teaching RL Agents to Act Better: VLM as Action Advisor for Online Reinforcement Learning
Authors: Xiefeng Wu, Jing Zhao, Shu Zhang, Mingyu Hu •
Published: 2025-09-25 •
Source: arXiv
Online reinforcement learning in complex tasks is time-consuming, as massive interaction steps are needed to learn the optimal Q-function.Vision-language action (VLA) policies represent a promising direction for solving diverse tasks; however, their performance on low-level control remains limited, and effective deployment often requires task-specific expert demonstrations for fine-tuning. In this paper, we propose \textbf{VARL} (\textbf{V}LM as \textbf{A}ction advisor for online \textbf{R}einforcement \textbf{L}earning), a framework that leverages the domain knowledge of vision-language models (VLMs) to provide action suggestions for reinforcement learning agents. Unlike previous methods, VARL provides action suggestions rather than designing heuristic rewards, thereby guaranteeing unchanged optimality and convergence. The suggested actions increase sample diversity and ultimately improve sample efficiency, especially in sparse-reward tasks. To validate the effectiveness of VARL, we evaluate it across diverse environments and agent settings. Results show that VARL greatly improves sample efficiency without introducing significant computational overhead. These advantages make VARL a general framework for online reinforcement learning and make it feasible to directly apply reinforcement learning from scratch in real-world environments.
25. TrustJudge: Inconsistencies of LLM-as-a-Judge and How to Alleviate Them
Authors: Yidong Wang, Yunze Song, Tingyuan Zhu, Xuanwang Zhang, Zhuohao Yu, Hao Chen, Chiyu Song, Qiufeng Wang, Cunxiang Wang, Zhen Wu, Xinyu Dai, Yue Zhang, Wei Ye, Shikun Zhang •
Published: 2025-09-25 •
Source: arXiv
The adoption of Large Language Models (LLMs) as automated evaluators (LLM-as-a-judge) has revealed critical inconsistencies in current evaluation frameworks. We identify two fundamental types of inconsistencies: (1) Score-Comparison Inconsistency, where lower-rated responses outperform higher-scored ones in pairwise comparisons, and (2) Pairwise Transitivity Inconsistency, manifested through circular preference chains (A>B>C>A) and equivalence contradictions (A=B=C\neq A). We argue that these issues come from information loss in discrete rating systems and ambiguous tie judgments during pairwise evaluation. We propose TrustJudge, a probabilistic framework that addresses these limitations through two key innovations: 1) distribution-sensitive scoring that computes continuous expectations from discrete rating probabilities, preserving information entropy for more precise scoring, and 2) likelihood-aware aggregation that resolves transitivity violations using bidirectional preference probabilities or perplexity. We also formalize the theoretical limitations of current LLM-as-a-judge frameworks and demonstrate how TrustJudge's components overcome them. When evaluated with Llama-3.1-70B-Instruct as judge using our dataset, TrustJudge reduces Score-Comparison inconsistency by 8.43% (from 23.32% to 14.89%) and Pairwise Transitivity inconsistency by 10.82% (from 15.22% to 4.40%), while maintaining higher evaluation accuracy. Our work provides the first systematic analysis of evaluation framework inconsistencies in LLM-as-a-judge paradigms, offering both theoretical insights and practical solutions for reliable automated assessment. The framework demonstrates consistent improvements across various model architectures and scales, enabling more trustworthy LLM evaluation without requiring additional training or human annotations. The codes can be found at https://github.com/TrustJudge/TrustJudge.
26. Mammo-CLIP Dissect: A Framework for Analysing Mammography Concepts in Vision-Language Models
Authors: Suaiba Amina Salahuddin, Teresa Dorszewski, Marit Almenning Martiniussen, Tone Hovda, Antonio Portaluri, Solveig Thrun, Michael Kampffmeyer, Elisabeth Wetzer, Kristoffer Wickstrøm, Robert Jenssen •
Published: 2025-09-25 •
Source: arXiv
Understanding what deep learning (DL) models learn is essential for the safe deployment of artificial intelligence (AI) in clinical settings. While previous work has focused on pixel-based explainability methods, less attention has been paid to the textual concepts learned by these models, which may better reflect the reasoning used by clinicians. We introduce Mammo-CLIP Dissect, the first concept-based explainability framework for systematically dissecting DL vision models trained for mammography. Leveraging a mammography-specific vision-language model (Mammo-CLIP) as a "dissector," our approach labels neurons at specified layers with human-interpretable textual concepts and quantifies their alignment to domain knowledge. Using Mammo-CLIP Dissect, we investigate three key questions: (1) how concept learning differs between DL vision models trained on general image datasets versus mammography-specific datasets; (2) how fine-tuning for downstream mammography tasks affects concept specialisation; and (3) which mammography-relevant concepts remain underrepresented. We show that models trained on mammography data capture more clinically relevant concepts and align more closely with radiologists' workflows than models not trained on mammography data. Fine-tuning for task-specific classification enhances the capture of certain concept categories (e.g., benign calcifications) but can reduce coverage of others (e.g., density-related features), indicating a trade-off between specialisation and generalisation. Our findings show that Mammo-CLIP Dissect provides insights into how convolutional neural networks (CNNs) capture mammography-specific knowledge. By comparing models across training data and fine-tuning regimes, we reveal how domain-specific training and task-specific adaptation shape concept learning. Code and concept set are available: https://github.com/Suaiba/Mammo-CLIP-Dissect.
27. Vision Transformers: the threat of realistic adversarial patches
Authors: Kasper Cools, Clara Maathuis, Alexander M. van Oers, Claudia S. Hübner, Nikos Deligiannis, Marijke Vandewal, Geert De Cubber •
Published: 2025-09-25 •
Source: arXiv
The increasing reliance on machine learning systems has made their security a critical concern. Evasion attacks enable adversaries to manipulate the decision-making processes of AI systems, potentially causing security breaches or misclassification of targets. Vision Transformers (ViTs) have gained significant traction in modern machine learning due to increased 1) performance compared to Convolutional Neural Networks (CNNs) and 2) robustness against adversarial perturbations. However, ViTs remain vulnerable to evasion attacks, particularly to adversarial patches, unique patterns designed to manipulate AI classification systems. These vulnerabilities are investigated by designing realistic adversarial patches to cause misclassification in person vs. non-person classification tasks using the Creases Transformation (CT) technique, which adds subtle geometric distortions similar to those occurring naturally when wearing clothing. This study investigates the transferability of adversarial attack techniques used in CNNs when applied to ViT classification models. Experimental evaluation across four fine-tuned ViT models on a binary person classification task reveals significant vulnerability variations: attack success rates ranged from 40.04% (google/vit-base-patch16-224-in21k) to 99.97% (facebook/dino-vitb16), with google/vit-base-patch16-224 achieving 66.40% and facebook/dinov3-vitb16 reaching 65.17%. These results confirm the cross-architectural transferability of adversarial patches from CNNs to ViTs, with pre-training dataset scale and methodology strongly influencing model resilience to adversarial attacks.
28. An Improved Quantum Software Challenges Classification Approach using Transfer Learning and Explainable AI
Authors: Nek Dil Khan, Javed Ali Khan, Mobashir Husain, Muhammad Sohail Khan, Arif Ali Khan, Muhammad Azeem Akbar, Shahid Hussain •
Published: 2025-09-25 •
Source: arXiv
Quantum Software Engineering (QSE) is a research area practiced by tech firms. Quantum developers face challenges in optimizing quantum computing and QSE concepts. They use Stack Overflow (SO) to discuss challenges and label posts with specialized quantum tags, which often refer to technical aspects rather than developer posts. Categorizing questions based on quantum concepts can help identify frequent QSE challenges. We conducted studies to classify questions into various challenges. We extracted 2829 questions from Q&A platforms using quantum-related tags. Posts were analyzed to identify frequent challenges and develop a novel grounded theory. Challenges include Tooling, Theoretical, Learning, Conceptual, Errors, and API Usage. Through content analysis and grounded theory, discussions were annotated with common challenges to develop a ground truth dataset. ChatGPT validated human annotations and resolved disagreements. Fine-tuned transformer algorithms, including BERT, DistilBERT, and RoBERTa, classified discussions into common challenges. We achieved an average accuracy of 95% with BERT DistilBERT, compared to fine-tuned Deep and Machine Learning (D&ML) classifiers, including Feedforward Neural Networks (FNN), Convolutional Neural Networks (CNN), and Long Short-Term Memory networks (LSTM), which achieved accuracies of 89%, 86%, and 84%, respectively. The Transformer-based approach outperforms the D&ML-based approach with a 6\% increase in accuracy by processing actual discussions, i.e., without data augmentation. We applied SHAP (SHapley Additive exPlanations) for model interpretability, revealing how linguistic features drive predictions and enhancing transparency in classification. These findings can help quantum vendors and forums better organize discussions for improved access and readability. However,empirical evaluation studies with actual developers and vendors are needed.
29. CLAUSE: Agentic Neuro-Symbolic Knowledge Graph Reasoning via Dynamic Learnable Context Engineering
Authors: Yang Zhao, Chengxiao Dai, Wei Zhuo, Yue Xiu, Dusit Niyato •
Published: 2025-09-25 •
Source: arXiv
Knowledge graphs provide structured context for multi-hop question answering, but deployed systems must balance answer accuracy with strict latency and cost targets while preserving provenance. Static k-hop expansions and "think-longer" prompting often over-retrieve, inflate context, and yield unpredictable runtime. We introduce CLAUSE, an agentic three-agent neuro-symbolic framework that treats context construction as a sequential decision process over knowledge graphs, deciding what to expand, which paths to follow or backtrack, what evidence to keep, and when to stop. Latency (interaction steps) and prompt cost (selected tokens) are exposed as user-specified budgets or prices, allowing per-query adaptation to trade-offs among accuracy, latency, and cost without retraining. CLAUSE employs the proposed Lagrangian-Constrained Multi-Agent Proximal Policy Optimization (LC-MAPPO) algorithm to coordinate three agents: Subgraph Architect, Path Navigator, and Context Curator, so that subgraph construction, reasoning-path discovery, and evidence selection are jointly optimized under per-query resource budgets on edge edits, interaction steps, and selected tokens. Across HotpotQA, MetaQA, and FactKG, CLAUSE yields higher EM@1 while reducing subgraph growth and end-to-end latency at equal or lower token budgets. On MetaQA-2-hop, relative to the strongest RAG baseline (GraphRAG), CLAUSE achieves +39.3 EM@1 with 18.6% lower latency and 40.9% lower edge growth. The resulting contexts are compact, provenance-preserving, and deliver predictable performance under deployment constraints.
30. Automatic Red Teaming LLM-based Agents with Model Context Protocol Tools
Authors: Ping He, Changjiang Li, Binbin Zhao, Tianyu Du, Shouling Ji •
Published: 2025-09-25 •
Source: arXiv
The remarkable capability of large language models (LLMs) has led to the wide application of LLM-based agents in various domains. To standardize interactions between LLM-based agents and their environments, model context protocol (MCP) tools have become the de facto standard and are now widely integrated into these agents. However, the incorporation of MCP tools introduces the risk of tool poisoning attacks, which can manipulate the behavior of LLM-based agents. Although previous studies have identified such vulnerabilities, their red teaming approaches have largely remained at the proof-of-concept stage, leaving the automatic and systematic red teaming of LLM-based agents under the MCP tool poisoning paradigm an open question. To bridge this gap, we propose AutoMalTool, an automated red teaming framework for LLM-based agents by generating malicious MCP tools. Our extensive evaluation shows that AutoMalTool effectively generates malicious MCP tools capable of manipulating the behavior of mainstream LLM-based agents while evading current detection mechanisms, thereby revealing new security risks in these agents.
31. Decoding the Surgical Scene: A Scoping Review of Scene Graphs in Surgery
Authors: Angelo Henriques, Korab Hoxha, Daniel Zapp, Peter C. Issa, Nassir Navab, M. Ali Nasseri •
Published: 2025-09-25 •
Source: arXiv
Scene graphs (SGs) provide structured relational representations crucial for decoding complex, dynamic surgical environments. This PRISMA-ScR-guided scoping review systematically maps the evolving landscape of SG research in surgery, charting its applications, methodological advancements, and future directions. Our analysis reveals rapid growth, yet uncovers a critical 'data divide': internal-view research (e.g., triplet recognition) almost exclusively uses real-world 2D video, while external-view 4D modeling relies heavily on simulated data, exposing a key translational research gap. Methodologically, the field has advanced from foundational graph neural networks to specialized foundation models that now significantly outperform generalist large vision-language models in surgical contexts. This progress has established SGs as a cornerstone technology for both analysis, such as workflow recognition and automated safety monitoring, and generative tasks like controllable surgical simulation. Although challenges in data annotation and real-time implementation persist, they are actively being addressed through emerging techniques. Surgical SGs are maturing into an essential semantic bridge, enabling a new generation of intelligent systems to improve surgical safety, efficiency, and training.
32. Concise and Sufficient Sub-Sentence Citations for Retrieval-Augmented Generation
Authors: Guo Chen, Qiuyuan Li, Qiuxian Li, Hongliang Dai, Xiang Chen, Piji Li •
Published: 2025-09-25 •
Source: arXiv
In retrieval-augmented generation (RAG) question answering systems, generating citations for large language model (LLM) outputs enhances verifiability and helps users identify potential hallucinations. However, we observe two problems in the citations produced by existing attribution methods. First, the citations are typically provided at the sentence or even paragraph level. Long sentences or paragraphs may include a substantial amount of irrelevant content. Second, sentence-level citations may omit information that is essential for verifying the output, forcing users to read the surrounding context. In this paper, we propose generating sub-sentence citations that are both concise and sufficient, thereby reducing the effort required by users to confirm the correctness of the generated output. To this end, we first develop annotation guidelines for such citations and construct a corresponding dataset. Then, we propose an attribution framework for generating citations that adhere to our standards. This framework leverages LLMs to automatically generate fine-tuning data for our task and employs a credit model to filter out low-quality examples. Our experiments on the constructed dataset demonstrate that the propose approach can generate high-quality and more readable citations.
33. ArchGPT: Understanding the World's Architectures with Large Multimodal Models
Authors: Yuze Wang, Luo Yang, Junyi Wang, Yue Qi •
Published: 2025-09-25 •
Source: arXiv
Architecture embodies aesthetic, cultural, and historical values, standing as a tangible testament to human civilization. Researchers have long leveraged virtual reality (VR), mixed reality (MR), and augmented reality (AR) to enable immersive exploration and interpretation of architecture, enhancing accessibility, public understanding, and creative workflows around architecture in education, heritage preservation, and professional design practice. However, existing VR/MR/AR systems are often developed case-by-case, relying on hard-coded annotations and task-specific interactions that do not scale across diverse built environments. In this work, we present ArchGPT, a multimodal architectural visual question answering (VQA) model, together with a scalable data-construction pipeline for curating high-quality, architecture-specific VQA annotations. This pipeline yields Arch-300K, a domain-specialized dataset of approximately 315,000 image-question-answer triplets. Arch-300K is built via a multi-stage process: first, we curate architectural scenes from Wikimedia Commons and filter unconstrained tourist photo collections using a novel coarse-to-fine strategy that integrates 3D reconstruction and semantic segmentation to select occlusion-free, structurally consistent architectural images. To mitigate noise and inconsistency in raw textual metadata, we propose an LLM-guided text verification and knowledge-distillation pipeline to generate reliable, architecture-specific question-answer pairs. Using these curated images and refined metadata, we further synthesize formal analysis annotations-including detailed descriptions and aspect-guided conversations-to provide richer semantic variety while remaining faithful to the data. We perform supervised fine-tuning of an open-source multimodal backbone ,ShareGPT4V-7B, on Arch-300K, yielding ArchGPT.
34. Plant identification based on noisy web data: the amazing performance of deep learning (LifeCLEF 2017)
Authors: Herve Goeau, Pierre Bonnet, Alexis Joly •
Published: 2025-09-25 •
Source: arXiv
The 2017-th edition of the LifeCLEF plant identification challenge is an important milestone towards automated plant identification systems working at the scale of continental floras with 10.000 plant species living mainly in Europe and North America illustrated by a total of 1.1M images. Nowadays, such ambitious systems are enabled thanks to the conjunction of the dazzling recent progress in image classification with deep learning and several outstanding international initiatives, such as the Encyclopedia of Life (EOL), aggregating the visual knowledge on plant species coming from the main national botany institutes. However, despite all these efforts the majority of the plant species still remain without pictures or are poorly illustrated. Outside the institutional channels, a much larger number of plant pictures are available and spread on the web through botanist blogs, plant lovers web-pages, image hosting websites and on-line plant retailers. The LifeCLEF 2017 plant challenge presented in this paper aimed at evaluating to what extent a large noisy training dataset collected through the web and containing a lot of labelling errors can compete with a smaller but trusted training dataset checked by experts. To fairly compare both training strategies, the test dataset was created from a third data source, i.e. the Pl@ntNet mobile application that collects millions of plant image queries all over the world. This paper presents more precisely the resources and assessments of the challenge, summarizes the approaches and systems employed by the participating research groups, and provides an analysis of the main outcomes.
35. An Automated Retrieval-Augmented Generation LLaMA-4 109B-based System for Evaluating Radiotherapy Treatment Plans
Authors: Junjie Cui, Peilong Wang, Jason Holmes, Leshan Sun, Michael L. Hinni, Barbara A. Pockaj, Sujay A. Vora, Terence T. Sio, William W. Wong, Nathan Y. Yu, Steven E. Schild, Joshua R. Niska, Sameer R. Keole, Jean-Claude M. Rwigema, Samir H. Patel, Lisa A. McGee, Carlos A. Vargas, Wei Liu •
Published: 2025-09-25 •
Source: arXiv
Purpose: To develop a retrieval-augmented generation (RAG) system powered by LLaMA-4 109B for automated, protocol-aware, and interpretable evaluation of radiotherapy treatment plans. Methods and Materials: We curated a multi-protocol dataset of 614 radiotherapy plans across four disease sites and constructed a knowledge base containing normalized dose metrics and protocol-defined constraints. The RAG system integrates three core modules: a retrieval engine optimized across five SentenceTransformer backbones, a percentile prediction component based on cohort similarity, and a clinical constraint checker. These tools are directed by a large language model (LLM) using a multi-step prompt-driven reasoning pipeline to produce concise, grounded evaluations. Results: Retrieval hyperparameters were optimized using Gaussian Process on a scalarized loss function combining root mean squared error (RMSE), mean absolute error (MAE), and clinically motivated accuracy thresholds. The best configuration, based on all-MiniLM-L6-v2, achieved perfect nearest-neighbor accuracy within a 5-percentile-point margin and a sub-2pt MAE. When tested end-to-end, the RAG system achieved 100% agreement with the computed values by standalone retrieval and constraint-checking modules on both percentile estimates and constraint identification, confirming reliable execution of all retrieval, prediction and checking steps. Conclusion: Our findings highlight the feasibility of combining structured population-based scoring with modular tool-augmented reasoning for transparent, scalable plan evaluation in radiation therapy. The system offers traceable outputs, minimizes hallucination, and demonstrates robustness across protocols. Future directions include clinician-led validation, and improved domain-adapted retrieval models to enhance real-world integration.