❌

Reading view

↗️

From Garbage to Gold: A Data-Architectural Theory of Predictive Robustness

✇cs.AI, q-bio.NC updates on arXiv.org
By: Terrence J. Lee-St. John · Jordan L. Lawson · Bartlomiej Piechowski-Jozwiak
arXiv:2603.12288v1 Announce Type: cross Abstract: Tabular machine learning presents a paradox: modern models achieve state-of-the-art performance using high-dimensional (high-D), collinear, error-prone data, defying the "Garbage In, Garbage Out" mantra. To help resolve this, we synthesize principles from Information Theory, Latent Factor Models, and Psychometrics, clarifying that predictive robustness arises not solely from data cleanliness, but from the synergy between data architecture and model capacity. Partitioning predictor-space "noise" into "Predictor Error" and "Structural Uncertainty" (informational deficits from stochastic generative mappings), we prove that leveraging high-D sets of error-prone predictors asymptotically overcomes both types of noise, whereas cleaning a low-D set is fundamentally bounded by Structural Uncertainty. We demonstrate why "Informative Collinearity" (dependencies from shared latent causes) enhances reliability and convergence efficiency, and explain why increased dimensionality reduces the latent inference burden, enabling feasibility with finite samples. To address practical constraints, we propose "Proactive Data-Centric AI" to identify predictors that enable robustness efficiently. We also derive boundaries for Systematic Error Regimes and show why models that absorb "rogue" dependencies can mitigate assumption violations. Linking latent architecture to Benign Overfitting, we offer a first step towards a unified view of robustness to Outcome Error and predictor-space noise, while also delineating when traditional DCAI's focus on label cleaning remains powerful. By redefining data quality from item-level perfection to portfolio-level architecture, we provide a theoretical rationale for "Local Factories" -- learning from live, uncurated enterprise "data swamps" -- supporting a deployment paradigm shift from "Model Transfer" to "Methodology Transfer'' to overcome static generalizability limitations.
  •  
  • ↗️
↗️

Reinforcement Learning for Diffusion LLMs with Entropy-Guided Step Selection and Stepwise Advantages

✇cs.AI, q-bio.NC updates on arXiv.org
By: Vishnu Teja Kunde · Fatemeh Doudi · Mahdi Farahbakhsh · Dileep Kalathil · Krishna Narayanan · Jean-Francois Chamberland
arXiv:2603.12554v1 Announce Type: cross Abstract: Reinforcement learning (RL) has been effective for post-training autoregressive (AR) language models, but extending these methods to diffusion language models (DLMs) is challenging due to intractable sequence-level likelihoods. Existing approaches therefore rely on surrogate likelihoods or heuristic approximations, which can introduce bias and obscure the sequential structure of denoising. We formulate diffusion-based sequence generation as a finite-horizon Markov decision process over the denoising trajectory and derive an exact, unbiased policy gradient that decomposes over denoising steps and is expressed in terms of intermediate advantages, without requiring explicit evaluation of the sequence likelihood. To obtain a practical and compute-efficient estimator, we (i) select denoising steps for policy updates via an entropy-guided approximation bound, and (ii) estimate intermediate advantages using a one-step denoising reward naturally provided by the diffusion model, avoiding costly multi-step rollouts. Experiments on coding and logical reasoning benchmarks demonstrate state-of-the-art results, with strong competitive performance on mathematical reasoning, outperforming existing RL post-training approaches for DLMs. Code is available at https://github.com/vishnutez/egspo-dllm-rl.
  •  
  • ↗️
↗️

Optimize Wider, Not Deeper: Consensus Aggregation for Policy Optimization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Zelal Su (Lain) · Mustafaoglu · Sungyoung Lee · Eshan Balachandar · Risto Miikkulainen · Keshav Pingali
arXiv:2603.12596v1 Announce Type: cross Abstract: Proximal policy optimization (PPO) approximates the trust region update using multiple epochs of clipped SGD. Each epoch may drift further from the natural gradient direction, creating path-dependent noise. To understand this drift, we can use Fisher information geometry to decompose policy updates into signal (the natural gradient projection) and waste (the Fisher-orthogonal residual that consumes trust region budget without first-order surrogate improvement). Empirically, signal saturates but waste grows with additional epochs, creating an optimization-depth dilemma. We propose Consensus Aggregation for Policy Optimization (CAPO), which redirects compute from depth to width: $K$ PPO replicates are optimized on the same batch, differing only in minibatch shuffling order, and then aggregated into a consensus. We study aggregation in two spaces: Euclidean parameter space, and the natural parameter space of the policy distribution via the logarithmic opinion pool. In natural parameter space, the consensus provably achieves higher KL-penalized surrogate and tighter trust region compliance than the mean expert; parameter averaging inherits these guarantees approximately. On continuous control tasks, CAPO outperforms PPO and compute-matched deeper baselines under fixed sample budgets by up to 8.6x. CAPO demonstrates that policy optimization can be improved by optimizing wider, rather than deeper, without additional environment interactions.
  •  
  • ↗️
↗️

Feynman: Knowledge-Infused Diagramming Agent for Scalable Visual Designs

✇cs.AI, q-bio.NC updates on arXiv.org
By: Zixin Wen · Yifu Cai · Kyle Lee · Sam Estep · Josh Sunshine · Aarti Singh · Yuejie Chi · Wode Ni
arXiv:2603.12597v1 Announce Type: cross Abstract: Visual design is an essential application of state-of-the-art multi-modal AI systems. Improving these systems requires high-quality vision-language data at scale. Despite the abundance of internet image and text data, knowledge-rich and well-aligned image-text pairs are rare. In this paper, we present a scalable diagram generation pipeline built with our agent, Feynman. To create diagrams, Feynman first enumerates domain-specific knowledge components (''ideas'') and performs code planning based on the ideas. Given the plan, Feynman translates ideas into simple declarative programs and iterates to receives feedback and visually refine diagrams. Finally, the declarative programs are rendered by the Penrose diagramming system. The optimization-based rendering of Penrose preserves the visual semantics while injecting fresh randomness into the layout, thereby producing diagrams with visual consistency and diversity. As a result, Feynman can author diagrams along with grounded captions with very little cost and time. Using Feynman, we synthesized a dataset with more than 100k well-aligned diagram-caption pairs. We also curate a visual-language benchmark, Diagramma, from freshly generated data. Diagramma can be used for evaluating the visual reasoning capabilities of vision-language models. We plan to release the dataset, benchmark, and the full agent pipeline as an open-source project.
  •  
  • ↗️
↗️

CarPLAN: Context-Adaptive and Robust Planning with Dynamic Scene Awareness for Autonomous Driving

✇cs.AI, q-bio.NC updates on arXiv.org
By: Junyong Yun · Jungho Kim · ByungHyun Lee · Dongyoung Lee · Sehwan Choi · Seunghyeop Nam · Kichun Jo · Jun Won Choi
arXiv:2603.12607v1 Announce Type: cross Abstract: Imitation learning (IL) is widely used for motion planning in autonomous driving due to its data efficiency and access to real-world driving data. For safe and robust real-world driving, IL-based planning requires capturing the complex driving contexts inherent in real-world data and enabling context-adaptive decision-making, rather than relying solely on expert trajectory imitation. In this paper, we propose CarPLAN, a novel IL-based motion planning framework that explicitly enhances driving context understanding and enables adaptive planning across diverse traffic scenarios. Our contributions are twofold: We introduce Displacement-Aware Predictive Encoding (DPE) to improve the model's spatial awareness by predicting future displacement vectors between the Autonomous Vehicle (AV) and surrounding scene elements. This allows the planner to account for relational spacing when generating trajectories. In addition to the standard imitation loss, we incorporate an augmented loss term that captures displacement prediction errors, ensuring planning decisions consider relative distances from other agents. To improve the model's ability to handle diverse driving contexts, we propose Context-Adaptive Multi-Expert Decoder (CMD), which leverages the Mixture of Experts (MoE) framework. CMD dynamically selects the most suitable expert decoders based on scene structure at each Transformer layer, enabling adaptive and context-aware planning in dynamic environments. We evaluate CarPLAN on the nuPlan benchmark and demonstrate state-of-the-art performance across all closed-loop simulation metrics. In particular, CarPLAN exhibits robust performance on challenging scenarios such as Test14-Hard, validating its effectiveness in complex driving conditions. Additional experiments on the Waymax benchmark further demonstrate its generalization capability across different benchmark settings.
  •  
  • ↗️
↗️

RetroReasoner: A Reasoning LLM for Strategic Retrosynthesis Prediction

✇cs.AI, q-bio.NC updates on arXiv.org
By: Hanbum Ko · Chanhui Lee · Ye Rin Kim · Rodrigo Hormazabal · Sehui Han · Sungbin Lim · Sungwoong Kim
arXiv:2603.12666v1 Announce Type: cross Abstract: Retrosynthesis prediction is a core task in organic synthesis that aims to predict reactants for a given product molecule. Traditionally, chemists select a plausible bond disconnection and derive corresponding reactants, which is time-consuming and requires substantial expertise. While recent advancements in molecular large language models (LLMs) have made progress, many methods either predict reactants without strategic reasoning or conduct only a generic product analysis, rather than reason explicitly about bond-disconnection strategies that logically lead to the choice of specific reactants. To overcome these limitations, we propose RetroReasoner, a retrosynthetic reasoning model that leverages chemists' strategic thinking. RetroReasoner is trained using both supervised fine-tuning (SFT) and reinforcement learning (RL). For SFT, we introduce SyntheticRetro, a framework that generates structured disconnection rationales alongside reactant predictions. In the case of RL, we apply a round-trip accuracy as reward, where predicted reactants are passed through a forward synthesis model, and predictions are rewarded when the forward-predicted product matches the original input product. Experimental results show that RetroReasoner not only outperforms prior baselines but also generates a broader range of feasible reactant proposals, particularly in handling more challenging reaction instances.
  •  
  • ↗️
↗️

Interrogating Design Homogenization in Web Vibe Coding

✇cs.AI, q-bio.NC updates on arXiv.org
By: Donghoon Shin · Alice Gao · Rock Yuren Pang · Jaewook Lee · Katharina Reinecke · Emily Tseng
arXiv:2603.13036v1 Announce Type: cross Abstract: Generative AI is known for its tendency to homogenize, often reproducing dominant style conventions found in training data. However, it remains unclear how these homogenizing effects extend to complex structural tasks like web design. As lay creators increasingly turn to LLMs to 'vibe-code' websites -- prompting for aesthetic and functional goals rather than writing code -- they may inadvertently narrow the diversity of their designs, and limit creative expression throughout the internet. In this paper, we interrogate the possibility of design homogenization in web vibe coding. We first characterize the vibe coding lifecycle, pinpointing stages where homogenization risks may arise. We then conduct a sociotechnical risk analysis unpacking the potential harms of web vibe coding and their interaction with design homogenization. We identify that the push for frictionless generation can exacerbate homogenization and its harms. Finally, we propose a mitigation framework centered on the idea of productive friction. Through case studies at the micro, meso, and macro levels, we show how centering productive friction can empower creators to challenge default outputs and preserve diverse expression in AI-mediated web design.
  •  
  • ↗️
↗️

PhysMoDPO: Physically-Plausible Humanoid Motion with Preference Optimization

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yangsong Zhang, Anujith Muraleedharan, Rikhat Akizhanov, Abdul Ahad Butt, G\"ul Varol, Pascal Fua, Fabio Pizzati, Ivan Laptev
arXiv:2603.13228v1 Announce Type: cross Abstract: Recent progress in text-conditioned human motion generation has been largely driven by diffusion models trained on large-scale human motion data. Building on this progress, recent methods attempt to transfer such models for character animation and real robot control by applying a Whole-Body Controller (WBC) that converts diffusion-generated motions into executable trajectories. While WBC trajectories become compliant with physics, they may expose substantial deviations from original motion. To address this issue, we here propose PhysMoDPO, a Direct Preference Optimization framework. Unlike prior work that relies on hand-crafted physics-aware heuristics such as foot-sliding penalties, we integrate WBC into our training pipeline and optimize diffusion model such that the output of WBC becomes compliant both with physics and original text instructions. To train PhysMoDPO we deploy physics-based and task-specific rewards and use them to assign preference to synthesized trajectories. Our extensive experiments on text-to-motion and spatial control tasks demonstrate consistent improvements of PhysMoDPO in both physical realism and task-related metrics on simulated robots. Moreover, we demonstrate that PhysMoDPO results in significant improvements when applied to zero-shot motion transfer in simulation and for real-world deployment on a G1 humanoid robot.
  •  
  • ↗️
↗️

SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks

✇cs.AI, q-bio.NC updates on arXiv.org
By: Xiangyi Li · Wenbo Chen · Yimin Liu · Shenghan Zheng · Xiaokun Chen · Yifeng He · Yubo Li · Bingran You · Haotian Shen · Jiankai Sun · Shuyi Wang · Binxu Li · Qunhong Zeng · Di Wang · Xuandong Zhao · Yuanli Wang · Roey Ben Chaim · Zonglin Di · Yipeng Gao · Junwei He · Yizhuo He · Liqiang Jing · Luyang Kong · Xin Lan · Jiachen Li · Songlin Li · Yijiang Li · Yueqian Lin · Xinyi Liu · Xuanqing Liu · Haoran Lyu · Ze Ma · Bowei Wang · Runhui Wang · Tianyu Wang · Wengao Ye · Yue Zhang · Hanwen Xing · Yiqi Xue · Steven Dillmann · Han-chung Lee
arXiv:2602.12670v3 Announce Type: replace Abstract: Agent Skills are structured packages of procedural knowledge that augment LLM agents at inference time. Despite rapid adoption, there is no standard way to measure whether they actually help. We present SkillsBench, a benchmark of 86 tasks across 11 domains paired with curated Skills and deterministic verifiers. Each task is evaluated under three conditions: no Skills, curated Skills, and self-generated Skills. We test 7 agent-model configurations over 7,308 trajectories. Curated Skills raise average pass rate by 16.2 percentage points(pp), but effects vary widely by domain (+4.5pp for Software Engineering to +51.9pp for Healthcare) and 16 of 84 tasks show negative deltas. Self-generated Skills provide no benefit on average, showing that models cannot reliably author the procedural knowledge they benefit from consuming. Focused Skills with 2--3 modules outperform comprehensive documentation, and smaller models with Skills can match larger models without them.
  •  
  • ↗️
↗️

Integration of TinyML and LargeML: A Survey of 6G and Beyond

✇cs.AI, q-bio.NC updates on arXiv.org
By: Thai-Hoc Vu · Ngo Hoang Tu · Thien Huynh-The · Kyungchun Lee · Sunghwan Kim · Miroslav Voznak · Quoc-Viet Pham
arXiv:2505.15854v2 Announce Type: replace-cross Abstract: The evolution from fifth-generation (5G) to sixth-generation (6G) networks is driving an unprecedented demand for advanced machine learning (ML) solutions. Deep learning has already demonstrated significant impact across mobile networking and communication systems, enabling intelligent services such as smart healthcare, smart grids, autonomous vehicles, aerial platforms, digital twins, and the metaverse. At the same time, the rapid proliferation of resource-constrained Internet-of-Things (IoT) devices has accelerated the adoption of tiny machine learning (TinyML) for efficient on-device intelligence, while large machine learning (LargeML) models continue to require substantial computational resources to support large-scale IoT services and ML-generated content. These trends highlight the need for a unified framework that integrates TinyML and LargeML to achieve seamless connectivity, scalable intelligence, and efficient resource management in future 6G systems. This survey provides a comprehensive review of recent advances enabling the integration of TinyML and LargeML in next-generation wireless networks. In particular, we (i) provide an overview of TinyML and LargeML, (ii) analyze the motivations and requirements for unifying these paradigms within the 6G context, (iii) examine efficient bidirectional integration approaches, (iv) review state-of-the-art solutions and their applicability to emerging 6G services, and (v) identify key challenges related to performance optimization, deployment feasibility, resource orchestration, and security. Finally, we outline promising research directions to guide the holistic integration of TinyML and LargeML for intelligent, scalable, and energy-efficient 6G networks and beyond.
  •  
  • ↗️
↗️

Neural-Quantum-States Impurity Solver for Quantum Embedding Problems

✇cs.AI, q-bio.NC updates on arXiv.org
By: Yinzhanghao Zhou · Tsung-Han Lee · Ao Chen · Nicola Lanat\`a · Hong Guo
arXiv:2509.12431v2 Announce Type: replace-cross Abstract: Neural quantum states (NQS) have emerged as a promising approach to solve second-quantized Hamiltonians, because of their scalability and flexibility. In this work, we design and benchmark an NQS impurity solver for the quantum embedding (QE) methods, focusing on the ghost Gutzwiller Approximation (gGA) framework. We introduce a graph transformer-based NQS framework able to represent arbitrarily connected impurity orbitals of the embedding Hamiltonian (EH) and develop an error control mechanism to stabilize iterative updates throughout the QE loops. We validate the accuracy of our approach with benchmark gGA calculations of the Anderson Lattice Model, yielding results in excellent agreement with the exact diagonalisation impurity solver. Finally, our analysis of the computational budget reveals the method's principal bottleneck to be the high-accuracy sampling of physical observables required by the embedding loop, rather than the NQS variational optimization, directly highlighting the critical need for more efficient inference techniques.
  •  
  • ↗️
↗️

When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling

✇cs.AI, q-bio.NC updates on arXiv.org
By: Heecheol Yun · Kwangmin Ki · Junghyun Lee · Eunho Yang
arXiv:2510.15346v2 Announce Type: replace-cross Abstract: Ensembling Large Language Models (LLMs) has gained attention as a promising approach to surpass the performance of individual models by leveraging their complementary strengths. In particular, aggregating models' next-token probability distributions to select the next token has been shown to be effective in various tasks. However, while successful for short-form answers, its application to long-form generation remains underexplored. In this paper, we show that using existing ensemble methods in long-form generation requires a careful choice of ensembling positions, since the standard practice of ensembling at every token often degrades performance. We identify two key factors for determining the ensembling positions: tokenization mismatch across models and consensus in their next-token probability distributions. Based on this, we propose SAFE, (Stable And Fast LLM Ensembling), a framework that selectively ensembles by jointly considering these factors. To further improve stability, we apply a probability sharpening strategy when the ensemble distribution becomes overly smooth, enabling the selection of more confident tokens during ensembling. Our experiments on diverse benchmarks, including MATH500 and BBH, demonstrate that SAFE outperforms existing methods in both accuracy and efficiency, with gains achieved even when ensembling fewer than 1% of tokens.
  •  
  • ↗️
↗️

First-line zolbetuximab plus mFOLFOX6 and nivolumab in unresectable CLDN18.2-positive gastric or gastroesophageal junction adenocarcinoma: a phase 2 trial

✇Nature Medicine
By: Kohei Shitara · Hirokazu Shoji · Nicola Fazio · Sara Lonardi · Keun-Wook Lee · Li-Yuan Bai · Kensei Yamaguchi · Jean-Philippe Metges · Gianluca Masi · Denis Smith · Tae-Yong Kim · Maria Matsangou · Archita Shrivastava · Miaomai Zhou · Jason Hill · Abraham Guerrero · Xuewei Wang · Aziz Zaanan · Samuel J. Klempner

Nature Medicine, Published online: 16 March 2026; doi:10.1038/s41591-026-04306-9

In cohort 4 of the ILUSTRO trial, combination of anti-CLDN18.2 zolbetuximab plus mFOLFOX6 and nivolumab in patients with CLDN18.2-positive, HER2-negative metastatic gastric or gastroesophageal junction adenocarcinoma led to encouraging clinical efficacy, supporting the testing of this combination in a phase 3 trial.
  •  
  • ↗️
↗️

Noninvasive biomarkers in thymic epithelial tumors: a systematic review of cfDNA/ctDNA detection, molecular profiling, and organoid-based monitoring

✇MRD
By: Raffaele Rocco · Ani Antaramian · Vikram Krishna · Philicia Moonsamy · Andrew R Brownlee · Harmik J Soukiasian

J Thorac Dis. 2026 Feb 28;18(2):171. doi: 10.21037/jtd-2025-1-2467. Epub 2026 Feb 26.

ABSTRACT

BACKGROUND: Thymic epithelial tumors (TETs), including thymomas and thymic carcinomas, are rare malignancies with limited treatment options and no established biomarkers for surveillance. Circulating cell-free DNA (cfDNA) and circulating tumor DNA (ctDNA) provide a non-invasive method for understanding tumor biology, detecting minimal residual disease (MRD), and possibly identifying recurrence. While this approach has added to the management of other solid tumors, its role in TETs remains poorly defined. The objective of this review was to evaluate the feasibility, molecular insights, and clinical utility of cfDNA and ctDNA for diagnosis, molecular profiling, and recurrence monitoring in TETs.

METHODS: This systematic review summarizes the current evidence on cfDNA and ctDNA in TETs. Studies were identifies through systematic searches of PubMed, Embase, Web of Science, MEDLINE, Cochrane Library, and American Society of Clinical Oncology (ASCO) meeting abstracts from inception through July 2025. Eligible studies reported cfDNA or ctDNA analysis in patients with histologically confirmed thymoma or thymic carcinoma, and excluded reviews, commentaries, abstracts without full text, and non-blood based liquid biopsy studies. Data extraction included patient characteristics, assay platforms, mutational findings, and clinical applications. Data were synthesized narratively due to methodological heterogeneity. No formal risk of bias assessment was performed because of the small number of included studies.

RESULTS: Six studies involving 289 patients met inclusion criteria. ctDNA detection was feasible across all studies, with detection rates ranging from 46% to 80%. Recurrent alterations included TP53, CDKN2A/B, KIT, and other variants. Liquid biopsy enabled genomic profiling at diagnosis and dynamic monitoring during treatment. Notably, several studies have suggested that disease recurrence may be detectable through liquid biopsy prior to the appearance of radiographic changes on conventional imaging. Despite these promising observations, evidence remains limited by small sample size, variability in assay methods, and short follow up duration.

CONCLUSIONS: Liquid biopsy approaches based on cfDNA and ctDNA have shown applicability in TETs and provide clinically relevant molecular information in settings where tissue-based analysis is limited. Tumor informed ctDNA strategies show particular promise for postoperative monitoring and longitudinal disease assessment, whereas broader clinical adoption remains investigational. Further prospective, multicenter studies are needed to establish standardized workflows and clarify the role of liquid biopsy across diagnostic, therapeutic, and surveillance contexts in TETs.

PMID:41816481 | PMC:PMC12972770 | DOI:10.21037/jtd-2025-1-2467

  •  
  • ↗️
↗️

Lense–Thirring precessing magnetar engine drives a superluminous supernova

✇Nature - Issue - nature.com science feeds
By: Joseph R. Farah · Logan J. Prust · D. Andrew Howell · Yuan Qi Ni · Curtis McCully · Moira Andrews · Harsh Kumar · Daichi Hiramatsu · Sebastian Gomez · Kathryn Wynn · Alexei V. Filippenko · K. Azalee Bostroem · Edo Berger · Peter Blanchard

Nature, Published online: 11 March 2026; doi:10.1038/s41586-026-10151-0

Observations of a nearby type I superluminous supernova showing oscillating light-curve bumps provide evidence of a centrally located magnetar in the wake of the explosion, surrounded by an infalling accretion disk undergoing Lense–Thirring precession.
  •  
  • ↗️
↗️

A sorghum pangenome reference improves global crop trait discovery

✇Nature - Issue - nature.com science feeds
By: Geoffrey P. Morris · Avril M. Harder · Adam L. Healey · Chloee M. McLaughlin · Joanna L. Rifkin · Clara Cruet-Burgos · Jerry W. Jenkins · Shengqiang Shu · John J. Spiekerman · Carl J. VanGessel · Erica Agnew · Alain Audebert · Kerrie Barry · Ivan Baxter · Gregory Beurier · Lori Beth Boston · Richard E. Boyles · Siobhan M. Brady · Victoria Bunting · Jacqueline M. Chaparro · Chaney Courtney · Joseph Sékou B. Dembele · Santosh Deshpande · Cyril Diatta · Nathaniel Eck · Andrea L. Eveland · Jacques M. Faye · Dave Flowers · Daniel Fonceka · Boubacar Gano · Marie de Gracia Coquerel · David Goodstein · Jane Grimwood · Matthew E. Hudson · Jana Kholova · Katherine Johnson · Kristen K. Johnson · Dorota Kawa · Mamoutou Kouressy · Stephen Kresovich · Scott Lee · Peggy G. Lemaux · Robert Lowery · Delphine Luquet · Fanna Maina · Sujan Mamidi · John K. McKay · Todd P. Michael · Taye T. Mindaye · John Mullet · Philip Ozersky · Christopher Plott · Jessica E. Prenni · Gael Pressoir · Jean-François Rami · Trevor W. Rife · Jocelyn Saxton · Bassirou Sine · Avinash Sreedasyam · Jayson T

Nature, Published online: 11 March 2026; doi:10.1038/s41586-026-10229-9

A pangenome reference for the phenotypically diverse crop sorghum aims to help accelerate future efforts to breed crops that are better adapted to changing environments.
  •  
  • ↗️
↗️

Nanophotonic waveguide chip-to-world beam scanning

✇Nature - Issue - nature.com science feeds
By: Matt Saha · Y. Henry Wen · Andrew S. Greenspon · Matthew Zimmermann · Kevin J. Palm · Alex Witte · Yin Min Goh · Chao Li · Jonathan Bumstead · Kevin Schädler · Ryan Fortin · Mark Dong · Andrew J. Leenheer · Genevieve Clark · Gerald Gilbert · Matt Eichenfield · Dirk Englund

Nature, Published online: 11 March 2026; doi:10.1038/s41586-025-10038-6

A monolithically integrated photonic ski-jump enables scalable, diffraction-limited 2D beam scanning from photonic chips, achieving ultrahigh spot rates, compact footprints and applications spanning displays, sensing and quantum photonics.
  •  
  • ↗️
↗️

Immune evasive DNA donors and recombinases license kilobase-scale writing

✇Nature - Issue - nature.com science feeds
By: Connor J. Tou · Keqiang Xie · Joana Ferreira da Silva · Pazhanichamy Kalailingam · Eliz Amar-Lewis · David Rufino-Ramos · William Sawyer · Madeline L. Eller · Jakob Starzyk · Ishita Majumdar · Jiao Wang · Danna Lee · Shaobo Yang · Ronald J. Meis · Gary A. Dahl · Jiahe Li · Richard Shan · Natalie Artzi · Patricia L. Musolino · Hao Wu · Benjamin P. Kleinstiver

Nature, Published online: 11 March 2026; doi:10.1038/s41586-026-10241-z

INSTALL overcomes fundamental challenges for DNA delivery and integration methods by synergizing immune-stealth nucleic acids with recombinases to enable kilobase-scale integration strategies without viral vectors.
  •  
  • ↗️
↗️

Hospitality-VQA: Decision-Oriented Informativeness Evaluation for Vision-Language Models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Jeongwoo Lee · Baek Duhyeong · Eungyeol Han · Soyeon Shin · Gukin han · Seungduk Kim · Jaehyun Jeon · Taewoo Jeong
arXiv:2603.07868v1 Announce Type: new Abstract: Recent advances in Vision-Language Models (VLMs) have demonstrated impressive multimodal understanding in general domains. However, their applicability to decision-oriented domains such as hospitality remains largely unexplored. In this work, we investigate how well VLMs can perform visual question answering (VQA) about hotel and facility images that are central to consumer decision-making. While many existing VQA benchmarks focus on factual correctness, they rarely capture what information users actually find useful. To address this, we first introduce Informativeness as a formal framework to quantify how much hospitality-relevant information an image-question pair provides. Guided by this framework, we construct a new hospitality-specific VQA dataset that covers various facility types, where questions are specifically designed to reflect key user information needs. Using this benchmark, we conduct experiments with several state-of-the-art VLMs, revealing that VLMs are not intrinsically decision-aware-key visual signals remain underutilized, and reliable informativeness reasoning emerges only after modest domain-specific finetuning.
  •  
  • ↗️
❌