The paper presents a framework called SNPE‑TTT for adapting pre‑trained neural posterior estimators, specifically conditional normalizing flows, to new inference tasks in agent‑based models (ABMs). It explores four adaptation strategies—full‑parameter adaptation, low‑rank adaptation, and two gradient‑based update spaces—and evaluates them on economic simulation models with both tractable and intractable likelihoods. Results show that the framework can improve posterior recovery with a limited number of target simulations, though performance varies by task and adaptation method.
By Junyao Zhang, Jinglai Li, Junqi Tang
The paper presents Variational Doob-Wasserstein Flow (VDWF), a simulation-free method for learning unbalanced population dynamics on the Riemannian Wasserstein-Fisher-Rao manifold. VDWF first learns a geometry model with a data-dependent metric to capture intrinsic dynamics, then uses Doob's h-transform to incorporate boundary information for trajectory inference. It employs a growth model to learn dynamics and construct intermediate mass samples in WFR space, analyzes convergence via a uniform Log-Sobolev inequality, and demonstrates competitive performance on real-world datasets for both balanced and unbalanced dynamics.
By Weizhu Qian
The paper investigates when inexpensive low‑fidelity (LF) physics models can reduce the number of costly high‑fidelity (HF) simulations needed for accurate solutions of parametric partial differential equations. It derives a sample‑complexity bound showing that learning the HF field via its discrepancy from the LF model can cut HF samples by a factor of κ^{2γ}, where κ is the complexity ratio and γ the effective‑dimension exponent, and confirms this scaling in a kernel setting. Guided by this theory, the authors introduce GaMF‑PINN, a gated multi‑fidelity physics‑informed neural network that selectively trusts the LF model and falls back to physics‑only learning when necessary, coupled with a conformal uncertainty procedure that guarantees coverage of the HF solution. The method is evaluated on the newly released MFPDE‑Bench, achieving HF‑PINN accuracy with up to eight times fewer HF solves, robustness to corrupted LF models, and reliable uncertainty estimates compared to Bayesian PINNs and deep ensembles.
By Olaf Yunus Laitinen Imanov, Rana Irem Turhan, Isgandar Jafarli
AmbientFM is a foundation model designed for ambient sensing using ubiquitous WiFi signals. It is pre‑trained on 9.2 million unlabeled Channel State Information samples collected over 439 days from 20 commercial devices in real homes, making it the largest real‑world WiFi CSI pre‑training corpus reported to date. The model learns transferable wireless representations through contrastive learning, masked reconstruction, and physics‑informed objectives, and with lightweight adaptation it can transfer to nine downstream tasks, achieving above 0.90 AUROC on all classification benchmarks and supporting dense spatial reconstruction.
By Guozhen Zhu, Yuqian Hu, Sakila Jayaweera, Weihang Gao, Wei-Hsiang Wang, Jiaxuan Zhang, Beibei Wang, Chenshu Wu, K. J. Ray Liu
Vib2ECG introduces a new dataset of synchronized twelve‑lead ECG and chest‑wall vibration recordings collected at the six standard precordial sites (V1–V6) from 17 subjects, using inertial measurement units. The accompanying benchmark demonstrates that a lightweight 364K‑parameter U‑Net can reconstruct electrical cardiac signals at various locations from these low‑cost vibrational signals, though a hallucination effect—where ECG waveforms appear without corresponding electrical activity—was observed and analyzed. This work expands the use of cardiac vibrational signals for mobile ECG monitoring and offers insights into the spatial relationship between cardiac electrical and mechanical activities.
By Guorui Lu, Xiaohui Cai, Todor Stefanov, Qinyu Chen
SymADiT is a symmetry‑aware variant of the All‑atom Diffusion Transformer (ADiT) designed for crystal generation. It uses Wyckoff‑based crystal representations and enforces symmetry constraints from the crystal’s space group and atom positions during latent‑space generative modeling. The approach yields stable, symmetric, and physically plausible crystals while maintaining competitive performance against both symmetry‑aware and symmetry‑agnostic models.
By Anmar Karmush, Cedric Mathieu Brandenburg, Soheil Ershadrad, Johanna Rosen, Michael Felsberg, Filip Ekstr\"om Kelvinius
DisCoMBO introduces the Distributional Conformance Score (DisCo), a bounded metric that blends the flexibility of Probabilistic Circuits with formal uncertainty quantification for Sequential Model-Based Optimization. DisCo measures model surprise in a [0,1] range, retaining linear-time inference while enabling accurate assessment of external knowledge conformance. The resulting DisCoMBO framework is proven to be a zero-regret algorithm and shows strong performance on benchmarks in AutoML, material optimization, and wind park optimization.
By Jonas Seng, Bennet Wittelsbach, Kristian Kersting
The study investigates how large language models retrieve network information, such as domain names and IP addresses, by identifying specific attention heads and neurons responsible for this capability. Using causal ablation and correlational analyses across five LLMs, the authors find that a small subset of heads can achieve near-perfect detection of network relationships, while single neurons show variable performance. The results demonstrate that the granularity of analysis—heads versus individual neurons—significantly affects the ability to localize retrieval mechanisms.
By Md Abdul Kadir (German Research Center for Artificial Intelligence, University of Oldenburg, Oldenburg, Germany), Md Mohasin Hossain (German Research Center for Artificial Intelligence, Saarland University, Saarbr\"ucken, Germany), Omair Shahzad Bhatti (German Research Center for Artificial Intelligence), Abdulrahman Mohamed Selim (German Research Center for Artificial Intelligence, University of Oldenburg, Oldenburg, Germany), Daniel Sonntag (German Research Center for Artificial Intelligence, University of Oldenburg, Oldenburg, Germany)
The paper introduces a continuous‑time actor‑critic framework for jump‑diffusion control problems, using a time‑inhomogeneous q‑function and occupation measure to derive a policy‑gradient representation that handles time‑dependent drift, volatility, and jump terms. The actor is modeled with conditional normalizing flows, allowing flexible non‑Gaussian continuous‑action policies while maintaining exact likelihoods for entropy regularization. Experiments on linear‑quadratic control, Merton portfolio optimization, and a multi‑agent portfolio game show stable learning, accurate policy approximation, and good scaling with dimension and agent count.
By Liya Guo, Ruimeng Hu, Xu Yang, Yi Zhu
CrystalREPA is a plug‑and‑play framework that aligns the atom‑wise hidden states of crystal generative encoders with frozen universal machine‑learning interatomic potential (MLIP) representations using an element‑aware contrastive objective. By doing so, it transfers stability‑aware atomistic priors to generative models with minimal training overhead and no extra inference cost, consistently improving thermodynamic stability, structural validity, and fidelity across multiple generative frameworks and benchmark datasets. The study also shows that an MLIP’s suitability as a teacher is better predicted by the distinguishability of its atom‑wise representation space than by its leaderboard accuracy, providing a practical, accuracy‑independent selection criterion.
whyItMatters":"The approach demonstrates a practical method to enhance crystal generative models’ ability to produce stable, realistic structures by leveraging existing MLIP knowledge without incurring additional inference costs."
By Chengqian Zhang, Yucheng Jin, Duo Zhang, Tiejun Li, Han Wang
The paper introduces PhysECD, a parity-aware E(3)-equivariant machine learning framework that predicts electronic circular dichroism (ECD) spectra from conformer geometries in milliseconds. It builds spectra from underlying quantum‑chemical quantities to enforce correct chiroptical symmetry. The authors also present SCECD, a large conformer‑resolved dataset of 578,104 computed spectra, and demonstrate that PhysECD achieves state‑of‑the‑art accuracy on benchmarks and matches experimental spectra, offering a fast alternative to TDDFT for absolute configuration assignment.
By Yi Jiang, Letian Chen, Runhan Shi, Lea Bae, Zheng Ling, Dong Ha Kim, Tong Zhu, Yang Yang
FUSION is a skill‑based research agent that automates the use of publicly available nuclear‑physics codes by retrieving the code, running verified inputs, parsing outputs, and ensuring results meet specified benchmarks. The current release supports twenty different codes across various nuclear physics domains and includes a searchable collection of literature references. Documentation and source code are freely available under an MIT license.
By Jin Lei
The paper introduces quantum Hamiltonian embeddings, a supervised learning framework that encodes a classical training dataset into a single parameterized Hamiltonian. Training optimizes the Hamiltonian’s parameters, and its approximate ground state defines the classification rule, enabling inference either classically or quantumly. The authors provide efficient gradient estimation, prove generalization guarantees, and show that, under cryptographic assumptions, the method can realize hypothesis functions beyond efficient classical reach, demonstrating scalability on IBM Heron processors.
By Riccardo Molteni, Minjun Jeon, Paolo Braccia, Abhinav Deshpande, Vedran Dunjko, Liang Jiang, Su-un Lee, Swarnadeep Majumder, Javier Robledo Moreno, Parthasarathy Suryanarayanan, Kunal Sharma
FinLLM-KG is a model‑agnostic framework that enhances financial sentiment analysis by integrating causal large language models with a financial knowledge graph derived from SEC 10‑K filings. It injects temporally grounded, company‑level context at inference time without updating model weights, allowing sentiment predictions to consider broader firm‑level information. The framework also attributes sentiment scores to individual sentences and retrieved knowledge‑graph facts, improving explainability and supporting human‑in‑the‑loop review, and simulations show it consistently outperforms state‑of‑the‑art baselines in portfolio performance.
By Giorgos Iacovides, Oliver Hannibal, Wuyang Zhou, Danilo Mandic
The paper introduces LOTUS, a low‑rank multilinear decomposition method for detecting hallucinations in large language models. LOTUS compresses variable‑length token representations into compact features using statistical descriptors of peak activations, distributions, and cross‑layer changes, then applies low‑rank decomposition to capture the underlying structure of hidden states. Experiments on several LLMs and QA benchmarks show that LOTUS outperforms recent state‑of‑the‑art hallucination detectors while requiring less labeled data.
By Devansh Sharma, Faizul Rakib Sayem, Shahana Ibrahim
The paper introduces HSC-Bench, a benchmark with 1,039 tasks designed to evaluate how well receiving AI models can take over tasks after handoffs. It measures two key aspects: the ability to recover relevant repository evidence and the ability to continue from established conversational, tool, and file-work states. Experiments with six models show that structured handoff information consistently improves evidence recovery and, for most models, continuation performance, while the best performers differ between the two tracks.
By Wenyuan Li, Guang Li, Dongyuan Li, Keisuke Maeda, Takahiro Ogawa, Miki Haseyama
The paper presents a Japanese discourse parsing system built on Segmented Discourse Representation Theory (SDRT) by translating existing English SDRT corpora with a large language model (LLM). Comparative experiments with prompt-based and fine-tuned LLMs show that Japanese and English parsers achieve similar performance, indicating cross‑lingual applicability of SDRT. Ablation studies reveal that Japanese discourse markers have little influence, while predictions rely mainly on contextual discourse relations rather than clause content, explaining the comparable results across typologically distinct languages.
By Satomu Tomisaki, Yosuke Mikami, Amane Watahiki, Hitomi Yanaka
The paper introduces Amplify, Locate, Estimate (ALE), a technique for efficiently estimating rare failure probabilities in safety‑fine‑tuned large language models. ALE amplifies failure rates by exploiting changes in token log‑probabilities, then uses a score‑function estimator and importance sampling to extrapolate back to the original rare rates. Experiments on 41 targets show ALE’s estimates closely match reference values, achieving accurate predictions for failure rates as low as 10⁻⁵ with far fewer labeled rollouts than direct sampling.
By Patrick Reichherzer, Jonathan Evertz, Thorsten Holz
DocXEval is the first comprehensive framework that adapts existing explainable AI methods and evaluation metrics to multimodal document understanding, covering text, layout, and image modalities. The authors systematically evaluate 13 XAI methods across 3 tasks, 8 datasets, 4 models, and 17 metrics, finding that Occlusion consistently performs best in faithfulness and interpretability. The study also provides meta‑evaluation insights, diagnostic analyses of modality contributions, spatial attribution patterns, and potential dataset biases, offering a benchmark and practical guide for future research.
By Saifullah Saifullah, David Dembinsky, Stefan Agne, Andreas Dengel, Sheraz Ahmed
ETAMem is a multi‑view memory framework designed to improve long‑term conversational agents by decoupling heterogeneous information and aligning memory representations with distinct characteristics. It parses conversational data into atomic units, identifies affective content, and organizes these units into three complementary hierarchical views—Event, Affect, and Time—using adaptively calibrated thresholds. Retrieval is performed via entropy‑based allocation of query budgets across views, enabling hierarchical search within each view. Experiments on LoCoMo and LongMemEval‑S show that ETAMem surpasses state‑of‑the‑art baselines with relative accuracy gains of 12.46% and 7.96%, while remaining token‑efficient, and analyses confirm the complementary benefits of the different memory views.
By Xinying Qian, Ying Zhang, Yu Zhao, Xubo Liu, Baohang Zhou, Jeff Z. Pan