中 EN
登录 注册账户

Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达，利用大语言模型自动生成交叉领域文献分析简报。

登录 注册账户
02.
arXiv (CS.CV) 2026-06-18

The Reward Was in Your Data All Along: Correcting Flow Matching with Discriminator-Guided RL

作者:
Nicolas Beltran-Velez↗Felix Friedrich↗Zhang Xiaofeng↗Reyhane Askari-Hemmat↗Xiaochuang Han↗Adriana Romero-Soriano↗Michal Drozdzal↗

Score- and flow-matching models often rely on preference-based reinforcement learning for two purposes: aligning with subjective preferences and, surprisingly, recovering properties such as visual realism and coherent object structure that matching-based training is intended to learn from the data itself. We argue that this reflects a structural mismatch. Matching losses measure $\ell_2$ regression error on the velocity or score field under training-time marginals, a proxy poorly aligned with the visual and semantic properties that determine sample quality at inference. Given a reward aligned with these properties, RL sidesteps the mismatch by evaluating the model on its own samples and following the reward landscape directly. The challenge is to obtain such a reward without relying on human preferences, which are expensive and conflate data realism with annotator inclinations. We propose Discriminator-Guided RL (DRL). DRL trains a discriminator to separate data from base-model samples in a pretrained representation space and uses its logit as the reward in KL-regularized RL. The pretrained space restricts the discriminator to perceptually meaningful directions, and the logit estimates the log-likelihood ratio between data and model, which is the optimal reward for targeting the data distribution. Across SiT, JiT, REPA, and RAE, DRL reduces guidance-free FID (e.g., $9.38 \to 2.62$ on SiT) and semantic-space FD (e.g., $88.2 \to 19.3$ on DINOv3 for SiT), with consistent gains across all backbones, and improves human-preference rewards without training on them. It also yields a better Pareto frontier between preference reward and image fidelity under subsequent preference-based post-training, increasing alignment while reducing low-level artifacts such as oversaturation and excessive brightness.

阅读与讨论 → 访问原文 →
03.
arXiv (CS.AI) 2026-06-24

SEAL: Searching Expandable Architectures for Incremental Learning

作者:
Matteo Gambella↗Manuel Roveri↗

arXiv:2505.10457v3 Announce Type: replace-cross Abstract: Incremental learning is a machine learning paradigm where a model learns from a sequential stream of tasks. This setting poses a key challenge: balancing plasticity (learning new tasks) and stability (preserving past knowledge). Neural Architecture Search (NAS), a branch of AutoML, automates the design of the architecture of Deep Neural Networks and has shown success in static settings. However, existing NAS-based approaches to incremental learning often rely on expanding the model at every task, making them impractical in resource-constrained environments. In this work, we introduce SEAL, a NAS-based framework tailored for data-incremental learning, a scenario where disjoint data samples arrive sequentially and are not stored for future access. SEAL adapts the model structure dynamically by expanding it only when necessary, based on a capacity estimation metric. Stability is preserved through cross-distillation training after each expansion step. The NAS component jointly searches for both the architecture and the optimal expansion policy. Experiments across multiple benchmarks demonstrate that SEAL effectively reduces forgetting and enhances accuracy while allocating additional capacity only when required. These results highlight the promise of combining NAS and selective expansion for efficient, adaptive learning in incremental scenarios.

阅读与讨论 → 访问原文 →
04.
arXiv (CS.CL) 2026-06-18

JetFlow: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting

作者:
Lanxiang Hu↗Zhaoxiang Feng↗Yulun Wu↗Haoran Yuan↗Yujie Zhao↗Yu-Yang Qian↗Bojun Wang↗Daxin Jiang↗Yibo Zhu↗Tajana Rosing↗Hao Zhang↗

Speculative decoding (SD) accelerates autoregressive Large Language Models (LLMs) by drafting multiple tokens and verifying them in parallel, but it faces a scaling limitation: increasing the draft budget improves speed only when acceptance remains high and drafting overhead stays low. This ceiling has been difficult to break because prior head-based SD methods face a causality-efficiency dilemma. Autoregressive drafters produce path-conditioned candidates that are effective for tree speculative decoding with higher acceptance length, but their drafting cost grows with tree depth. Bidirectional block-diffusion drafters generate all positions in one pass, but their branch-agnostic marginals can form individually plausible yet mutually inconsistent trees, wasting budget and reducing acceptance. We propose JetFlow, a head-based SD framework that combines one-forward drafting efficiency with branch-wise causal conditioning. JetFlow trains a causal parallel draft head over fused hidden states from the frozen target model, producing candidate trees whose scores align with the target model's autoregressive factorization. This enables JetFlow to convert larger draft budgets into longer accepted prefixes and higher end-to-end speedup. Across math, coding, and chat benchmarks on dense and MoE Qwen3 models, JetFlow consistently outperforms bidirectional-head and tree-based SD baselines. On H100 GPUs, JetFlow achieves up to 9.64x speedup on MATH-500 and 4.58x on open-ended conversational workloads, with further latency gains demonstrated through vLLM integration under realistic serving loads. Our code and models are available at https://github.com/hao-ai-lab/JetFlow.

阅读与讨论 → 访问原文 →
05.
arXiv (math.PR) 2026-06-17

Convergence Analysis of the Random Bisection Method

作者:
Ludovick Bouthat↗Philippe-Andr\'e Luneau↗Philippe Petitclerc↗

arXiv:2603.20483v2 Announce Type: replace-cross Abstract: We propose a generalized version of the bisection method where the cutting point between the two subintervals is chosen at random following an arbitrary distribution. We compute expected convergence rates with respect to any arbitrary a priori distribution for the position of the root in the initial interval and proved that it depends only on the the expectation $\mathbb{E}[c(1-c)]$ of the cut $c$. We also provide a generalization of the method for $K$ random cuts and study its convergence properties. Most probabilistic derivations are kept fairly simple for the ease of understanding of a larger audience. Our theoretical results are then validated numerically using statistical simulation.

阅读与讨论 → 访问原文 →
06.
medRxiv (Medicine) 2026-06-22

Assessment of adaptive functioning in Angelman syndrome using the Vineland Adaptive Behavior Scales, Third Edition

作者:
Potter↗S. N↗Zhang↗Friedman↗Gable↗Ali↗Barbieri-Welge↗R. L↗Ben-Tall↗Caravella↗K. E↗DeRamus↗…

Purpose: This study examined longitudinal trajectories of adaptive functioning in 331 individuals with Angelman syndrome (AS) using the Vineland Adaptive Behavior Scales, Third Edition (Vineland-3) and examined differences by molecular subtype. Methods: A total of 331 individuals (156 females, 47%) with genetically confirmed AS (ages 6 months to 52 years) were assessed between 2018 and 2025, including 207 with a deletion subtype, 63 with uniparental disomy or imprinting defect, and 61 with a UBE3A point mutation. Growth scale values were analyzed using linear mixed-effects models with log2-transformed age. Results: Individuals with deletion subtypes demonstrated significantly lower adaptive functioning across domains compared to those with non-deletion subtypes. Adaptive skills across all Vineland-3 subdomains increased nonlinearly with age, showing faster growth early in life that slowed over time, with largely parallel trajectories across subtypes. Conclusion: Individuals with AS demonstrate slow but steady growth in adaptive functioning that continues into adulthood, with progress varying by molecular subtype. These findings provide updated natural history benchmarks and demonstrate the utility of the Vineland-3 for clinical trials.

阅读与讨论 → 访问原文 →
07.
arXiv (CS.CL) 2026-06-16

Generative AI and the future of scientometrics: current topics and future questions

作者:
Benedetto Lepori↗Jens Peter Andersen↗Karsten Donnay↗

In this paper, we contribute to the debate on generative artificial intelligence (GenAI) in scientometrics. We argue that moving from a trial-and-error approach to an explainable and actionable use requires a principled understanding of strengths and weaknesses of GenAI as compared with other techniques and with human judgment. To this end, we introduce a conceptual framework based on the distinction between the semantic dimensions of texts, i.e. the meanings attributed to words, and their pragmatic dimension, i.e. their embedding within communicative situations. We leverage this framework to interpret the results of applications of GenAI in scientometrics and to provide guidance to users. Specifically, we conclude that key parameters to be considered are the nature of the task, the level of granularity of the analysis and whether the goal was descriptive, inferential or evaluative. These parameters lead to different strategies for using GenAI and human-machine integration. Finally, we suggest that, by generating large amounts of scientific language, GenAI might affect textual characteristics used to measure science, such as authors, words, and references. We argue that careful empirical work and theoretical reflection will be essential to remain capable of interpreting the evolving patterns of knowledge production in the age of AI.

阅读与讨论 → 访问原文 →
08.
arXiv (quant-ph) 2026-06-19

String dynamics of a (2+1)D U(1) quantum link model on a digital quantum computer

作者:
Anthony Gandon↗Alessandro Mariani↗Debasish Banerjee↗Emilie Huffman↗Gurtej Kanwar↗Francesco Tacchino↗Uwe-Jens Wiese↗Ivano Tavernelli↗

arXiv:2606.19601v1 Announce Type: new Abstract: The (2+1)D U(1) pure gauge theory always exists in the confining phase, with strings of non-zero string tension giving a characteristic linear potential between static charges. This makes it a useful testing ground for quantum computing methods designed to study string dynamics of confining gauge theories. Here we implement a minimal U(1) quantum link model on a quantum computer with qubit degrees of freedom representing the dual height variables of the model. This facilitates an efficient realization of plaquette interactions and enables effective calculations of real-time dynamics that are inaccessible to traditional quantum Monte Carlo. A specifically tailored lattice geometry is chosen to match the heavy-hexagonal geometry of the IBM quantum hardware used here, minimizing non-adjacent qubit interactions. By performing quantum quenches from a simple initial string state, we probe the transverse quantum fluctuations of the string before it thermalizes. Our experimental results from digital quantum simulations, with up to 112 qubits, show good agreement with reference tensor-network calculations at short times and with thermal averages at long times. Near the phase transition, the quench dynamics exhibit large fluctuations of the initial string that extend across both spatial dimensions of the lattice. Nonetheless, our error-mitigated estimators from the quantum hardware also give accurate predictions in that regime, with noise-induced violations of local gauge symmetries comparable to finite-bond-dimension tensor-network results.

阅读与讨论 → 访问原文 →
09.
arXiv (CS.LG) 2026-06-17

On Randomized Algorithms in Online Strategic Classification

作者:
Chase Hutton↗Adam Melrod↗Han Shao↗

arXiv:2602.06257v2 Announce Type: replace Abstract: Online strategic classification studies settings in which agents strategically modify their features to obtain favorable predictions. For example, given a classifier that determines loan approval based on credit scores, applicants may open or close credit cards and bank accounts to obtain a positive prediction. The learning goal is to achieve low mistake or regret bounds despite such behavior. While randomized algorithms have the potential to offer advantages to the learner in strategic settings, they have been largely underexplored. In the realizable setting, no lower bound is known for randomized algorithms, and existing lower bound constructions for deterministic learners can be circumvented by randomization. In the agnostic setting, the best known regret upper bound is $O(T^{3/4}\log^{1/4}T|\mathcal H|)$, which is far from the standard online learning rate of $O(\sqrt{T\log|\mathcal H|})$. In this work, we provide refined bounds for online strategic classification in both settings; our bounds depend on the Littlestone dimension $\mathrm{Ldim}(\mathcal H)$ of the hypothesis class $\mathcal H$ and the maximum degree $\Delta$ of the manipulation graph. In the realizable setting, we extend, for $T > \mathrm{Ldim}(\mathcal H) \Delta^2$, the existing lower bound $\Omega(\mathrm{Ldim}(\mathcal H) \Delta)$ for deterministic learners to all learners. This yields the first lower bound that applies to randomized learners. We then provide the first randomized learner that improves the known (deterministic) upper bound of $O(\mathrm{Ldim}(\mathcal H) \cdot \Delta \log \Delta)$. In the agnostic setting, we give an improper randomized learner that improves the regret upper bound to $O(\sqrt{T\log|\mathcal H|})$, matching the standard online learning rate. We also show a larger lower bound for all proper learning rules, demonstrating that improperness is necessary to achieve the optimal rate.

阅读与讨论 → 访问原文 →
10.
arXiv (CS.CL) 2026-06-12

EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments

作者:
Jundong Xu↗Qingchuan Li↗Jiaying Wu↗Yihuai Lan↗Shuyue Stella Li↗Huichi Zhou↗Bowen Jiang↗Lei Wang↗Jun Wang↗Anh Tuan Luu↗Caiming Xiong↗Hae Won Park↗…

Large language model (LLM) agents have achieved strong performance on a wide range of benchmarks, yet most evaluations assume static environments. In contrast, real-world deployment is inherently dynamic, requiring agents to continually align their knowledge, skills, and behavior with changing environments and updated task conditions. To address this gap, we introduce EvoArena, a benchmark suite that models environment changes as sequences of progressive updates across terminal, software, and social domains. We further propose EvoMem, a patch-based memory paradigm that records memory evolution as structured update histories, enabling agents to reason about environmental evolution through changes in their memory. Experiments show that current agents struggle on EvoArena, achieving an average accuracy of 39.6% across evolving terminal, software, and social-preference domains. EvoMem consistently improves performance, yielding an average gain of 1.5% on EvoArena and also improving standard benchmarks such as GAIA and LoCoMo by 6.1% and 4.8%. Beyond individual tasks, EvoMem further improves chain-level accuracy by 3.7% on EvoArena, where success requires completing a consecutive sequence of related evolutionary subtasks. Mechanistic analysis shows that EvoMem improves evidence capture in the memory, indicating better preservation of complete evolving environment states. Our results highlight the importance of modeling evolution in both evaluation and memory for reliable agent deployment.

阅读与讨论 → 访问原文 →
11.
arXiv (math.PR) 2026-06-11

On the structure of the sandpile identity element on Sierpinski gasket graphs

作者:
Robin Kaiser↗Ecaterina Sava-Huss↗Julia \"Uberbacher↗

arXiv:2603.12006v2 Announce Type: replace-cross Abstract: We consider the identity of the abelian sandpile group of finite approximation graphs of the Sierpinski gasket, and we show that the second-order term in the scaling limit converges to the path distance to the nearest corner on the Sierpinski gasket. The proof relies on a decomposition of the identity of the sandpile group into the sum of a constant function and the Laplacian of the graph distance on the approximating graphs.

阅读与讨论 → 访问原文 →
12.
arXiv (CS.CL) 2026-06-15

Can Post-Training Turn LLMs into Good Medical Coders? An Empirical Study of Generative ICD Coding

作者:
Ziqing Wang↗Weihao Li↗Shijie Chen↗Yuan Luo↗Kaize Ding↗

Automated International Classification of Diseases (ICD) coding is a core medical-coding task for billing, epidemiology, and clinical decision support. Generative large language models (LLMs) are often reported as weak medical coders, but this finding mainly comes from inference-time settings such as prompting, retrieval, reranking, or tool use, leaving the role of task-specific post-training underexplored. We present a controlled empirical study of post-training for generative ICD coding, comparing discriminative baselines with LLM coders across prompting, supervised fine-tuning, and reinforcement learning under a common protocol and metric set. To our knowledge, this is the first study to evaluate RL-based post-training for generative LLM coders in ICD coding. We further introduce PHI, a diagnostic curriculum that extends GRPO to refine missed-code cases. Our results show that prompting-only evaluation substantially underestimates the potential of LLMs for ICD coding. SFT provides the main capability jump, GRPO further improves code-set prediction beyond SFT, and PHI provides targeted gains on macro-level performance. These findings suggest that the main bottleneck is not the generative formulation alone, but how the model is adapted and optimized for full-taxonomy recall. We release our code, data splits, and checkpoints at https://github.com/AlexandreWANG915/LLM4ICD.

阅读与讨论 → 访问原文 →
13.
arXiv (CS.AI) 2026-06-12

ToolSense: A Diagnostic Framework for Auditing Parametric Tool Knowledge in LLMs

作者:
Ashutosh Hathidara↗Sai Shruthi Sistla↗Sebastian Schreiber↗Sahil Bansal↗

arXiv:2606.12451v1 Announce Type: new Abstract: Large language models deployed as agents over large tool catalogs face a critical tool-retrieval bottleneck. As embedding-based retrieval approaches rely on compact encoders that may under-capture specialized tool semantics, parametric tool retrieval addresses this by encoding each tool as a virtual token appended to the LLM vocabulary, fine-tuned in two stages (memorization then retrieval SFT) to use the LLM as a retriever, achieving strong performance on standard ToolBench retrieval benchmarks. Yet these benchmarks use verbose, fully-specified queries, and their evaluation applies constrained decoding that restricts outputs to valid token paths, neither reveals whether the model actually understands its tools. We introduce ToolSense, an open-source LLM-powered diagnostic framework that takes any tool catalog as input and automatically generates three benchmarks: a Realistic Retrieval Benchmark (RRB) with queries at three ambiguity tiers, an MCQ probing benchmark, and a QA probing benchmark. Applying ToolSense to ToolBench (~47k tools) and evaluating five parametric model training configurations reveals a knowledge-retrieval dissociation: on RRB queries, several configurations collapse by ~50-64 percentage points compared to fully-specified ToolBench benchmarks, falling below the embedding-model baseline. Additionally, despite strong retrieval performance, some models score near-random on factual probes, suggesting a knowledge-retrieval dissociation. We open-source the ToolSense framework and the ToolBench diagnostic benchmarks at https://github.com/SAP/toolsense.

阅读与讨论 → 访问原文 →
14.
medRxiv (Medicine) 2026-06-18

Instantaneous-Frequency EEG Microstate Dynamics Stratify Motor Subtypes in Parkinson's Disease

作者:
Nobukawa↗Watanabe↗

Parkinson's disease (PD) is clinically heterogeneous, yet objective electrophysiological markers of its postural-instability/gait-difficulty (PIGD) and tremor-dominant (TD) motor subtypes are lacking. We tested whether the temporal dynamics of instantaneous-frequency (IF) microstates in resting-state electroencephalography (EEG) distinguish these subtypes from each other and from healthy controls (HC). In a publicly available cohort (OpenNeuro ds007526) comprising 28 HC and 97 PD patients classified as PIGD (n=50) or TD (n=47), the spatial distribution of the IF was reduced by principal component analysis and modeled with a Gaussian hidden Markov model, yielding three recurrent microstates. Per-participant mean dwell time, occupancy, and state-transition probabilities were compared across the three groups and, within PD, correlated with clinical scores. We found that the dynamics of one microstate varied systematically across groups: its dwell time, occupancy, and self-transition probability increased monotonically from HC through TD to PIGD, while outgoing transitions decreased, so that the state became an increasingly persistent attractor. For dwell time, all three pairwise contrasts survived correction (HC versus PIGD, Hedges' g=1.06; HC versus TD, g=0.59; PIGD versus TD, g=0.40). None of the dynamic indices was associated with clinical severity, disease duration, or medication dose within PD. IF-microstate dynamics thus stratify the PD motor subtypes along a graded continuum without tracking continuous disease severity. The approach offers a candidate objective EEG marker for motor-subtype stratification, complementing spectral characterizations of PD.

阅读与讨论 → 访问原文 →
15.
arXiv (CS.AI) 2026-06-12

The Containment Gap: How Deployed Agentic AI Frameworks Fail Public-Facing Safety Requirements

作者:
Md Jafrin Hossain↗Mohammad Arif Hossain↗Weiqi Liu↗Nirwan Ansari↗

arXiv:2606.12797v1 Announce Type: new Abstract: Agentic large language model systems that autonomously invoke tools, maintain persistent memory, and execute multi-step plans are increasingly deployed in public-facing domains, including government services, healthcare triage, and financial advising. We ask whether the frameworks used to build these systems provide architectural-level structural safety guarantees. Applying six containment principles derived from a compositional model of agentic architectures, we audit three dominant frameworks (LangChain, AutoGPT, and OpenAI Agents SDK) and find no native compliance in any of them. Memory integrity, a defense against one of the most prevalent vulnerability classes, is not observed in any of the three evaluated frameworks. We validate these findings empirically: in a simulated government benefits agent built on LangChain, a single memory-poisoning write induces persistent targeted corruption across all tested seeds and backends, increasing the wrongful denial rate for targeted applicants to 88.9%. Under a complex five-factor policy, the same attack preserves aggregate accuracy while increasing targeted wrongful denials by 3.5x, rendering the corruption difficult to detect through standard monitoring. We then introduce two lightweight containment mechanisms: a memory integrity validator and a policy gate, which eliminate both attack vectors with sub-millisecond overhead (

阅读与讨论 → 访问原文 →
16.
arXiv (CS.CL) 2026-06-16

Metacognitive Myopia in Large Language Models

作者:
Florian Scholten↗Tobias R. Rebholz↗Mandy H\"utter↗

Large Language Models (LLMs) exhibit potentially harmful biases that reinforce culturally embedded stereotypes, influence moral judgments, or amplify positive evaluations of majority groups. We propose metacognitive myopia as a cognitive-ecological framework accounting for a conglomerate of established and emerging LLM biases. Our theoretical framework posits that biased samples in the information environment cause five symptoms of metacognitive myopia in LLMs: integration of invalid embeddings, susceptibility to redundant information, neglect of base rates in conditional computation, decision rules based on frequency, and inappropriate higher-order statistical inference for nested data structures. Moreover, it posits that the two main components of metacognition, monitoring and control, could account for these five symptoms. Accordingly, we further outline how monitoring and control could be approximated technically, for instance, through hidden parallel reasoning histories that allow interactive LLMs to evaluate risks of myopic inference before generating overt responses. Our theoretical framework provides a novel perspective on flawed human-machine interactions and agentic AI and raises significant ethical concerns regarding the implementation of LLMs in organizational structures and high-stakes decisions.

阅读与讨论 → 访问原文 →
17.
arXiv (CS.CV) 2026-06-16

Shift-and-Sum Quantization for Visual Autoregressive Models

作者:
Jaehyeon Moon↗Bumsub Ham↗

Post-training quantization (PTQ) enables efficient deployment of deep networks using a small set of data. Its application to visual autoregressive models (VAR), however, remains relatively unexplored. We identify two key challenges for applying PTQ to VAR: (i) large reconstruction errors in attention-value products, especially at coarse scales where high attention scores occur more frequently; and (ii) a discrepancy between the sampling frequencies of codebook entries and their predicted probabilities due to limited calibration data. To address these challenges, we propose a PTQ framework tailored for VAR. First, we introduce a shift-and-sum quantization method that reduces reconstruction errors by aggregating quantized results from symmetrically shifted duplicates of value tokens. Second, we present a resampling strategy for calibration data that aligns sampling frequencies of codebook entries with their predicted probabilities. Experiments on class-conditional image generation, inpainting, outpainting, and class-conditional editing show consistent improvements across VAR architectures, establishing a new state of the art in PTQ for VAR.

阅读与讨论 → 访问原文 →
18.
arXiv (CS.CL) 2026-06-16

Deep Temporal Modeling and Ensemble Fusion for Multimodal Emotion Recognition from Physiological Signals

作者:
Desta Haileselassie Hagos↗Saurav Keshari Aryal↗Patrick Ymele-Leki↗Anietie Andy↗Legand L. Burge↗

Physiological stress and emotion recognition are important for health monitoring and affective computing. In this work, we present a comprehensive evaluation of deep learning models such as Long Short-Term Memory (LSTM), Temporal Convolutional Networks (TCN), and Transformer on the WESAD dataset for multimodal affect recognition using wrist and chest sensor signals. We perform ablation studies to assess the individual contributions of each modality by training models on wrist-only and chest-only inputs. In addition, we implement a late-fusion ensemble strategy that combines predictions from all three architectures trained on multimodal input. We also employ early fusion at the sensor level by concatenating wrist and chest signals before feeding them into each model. Our results show that Transformer models consistently achieve the highest accuracy in multimodal settings, while TCN models perform best in the wrist-only configuration. The ensemble method yields the highest overall accuracy (98.91 +/- 0.13%) and macro-F1 score (98.56 +/- 0.17%). These findings demonstrate the effectiveness of sensor fusion and ensemble-based fusion in developing robust systems for physiological emotion recognition.

阅读与讨论 → 访问原文 →
19.
arXiv (CS.CL) 2026-06-16

Building Customer Support AI Agents at 100M-User Scale: An Evaluation-Driven Framework

作者:
Aman Gupta↗Kevin Rossell↗Edesio Alcoba\c{c}a↗Jose Chrystian Lima Pacheco↗Carolina Baptista de Lima↗Shao Tang↗Luiz Paulo Rabachini↗Luis Moneda↗Herbert Fei↗Daniel Silva↗Rohan Ramanath↗

The rapid rise in LLM capabilities has made AI agents increasingly viable across a broad range of tasks. Among the most promising applications is building production-ready customer-facing agents, a challenge that demands coordinated excellence in evaluation methodology, context engineering, training, and online measurement. Yet these critical pillars are typically developed in isolation, creating blind spots that only surface after deployment. In this paper, we present a unified framework that bridges offline development with online impact for customer support AI agents at Nubank, a company with 100M+ users. Our approach integrates several key components: (1) structured context engineering tailored to customer support agents, (2) systematic human-in-the-loop prompt iteration, (3) rigorous LLM judge evaluation with measured inter-rater agreement and GEPA optimization for consistency, and (4) ideation-to-production validation. A central insight is that evaluation-pipeline quality directly determines iteration velocity. We present results from five production deployments spanning distinct domains: card delivery, debt management, credit-limit support, card management, and product explanation. These deployments deliver consistent customer-satisfaction gains while substantially accelerating iteration. In our card-delivery deployment, large-scale A/B testing yields a 37 percentage-point improvement in AI transactional Net Promoter Score and a 29 percentage-point gain in self-service rate over prior agent variants, alongside a strong correlation between offline simulation metrics and online outcomes, demonstrating that eval-driven development reliably predicts production impact. On most use cases, AI satisfaction reaches within a few percentage points of expert human agents.

阅读与讨论 → 访问原文 →
20.
arXiv (CS.AI) 2026-06-18

A Technical Taxonomy of LLM Agent Communication Protocols

作者:
Linus Sander↗Habtom Kahsay Gidey↗Alexander Lenz↗Alois Knoll↗

arXiv:2606.19135v1 Announce Type: cross Abstract: As large language models (LLMs) advance and multi-agent systems aim to overcome the limits of standalone agents, robust communication protocols are becoming essential infrastructure for distributed agent networks. Nonetheless, the fragmented protocol landscape presents a significant interoperability challenge. This study develops a technical taxonomy to classify and analyze LLM agent communication protocols. Following an established iterative method, we defined the taxonomy's purpose, meta-characteristic, and ending conditions, then performed five iterations, three empirical-to-conceptual and two conceptual-to-empirical, on nine actively maintained open-source protocols with demonstrable adoption. The taxonomy comprises five dimensions: counterparty, payload, interaction state, discovery mechanism, and schema flexibility. Classification reveals recurring architectural patterns: all sampled agent-to-agent protocols combine hybrid payloads with session-state persistence; most protocols support multiple predefined schemas, and two negotiate schemas at runtime, indicating a trend toward schema flexibility; decentralized discovery remains rare. Analysis suggests short-term convergence pressure toward protocols unifying agent-to-agent and agent-to-context (tool and data) communication. Long-term, however, no single protocol is likely to maximize versatility, efficiency, and portability simultaneously. The field will more likely evolve toward a federated, layered protocol stack. The framework guides protocol selection and highlights open research gaps such as privacy and policy enforcement.}

阅读与讨论 → 访问原文 →
21.
arXiv (CS.AI) 2026-06-15

An interpretable unsupervised representation learning for high precision measurement in particle physics

作者:
Xing-Jian Lv↗De-Xing Miao↗Zi-Jun Xu↗Jian-Chun Wang↗

arXiv:2511.22246v2 Announce Type: replace-cross Abstract: Unsupervised learning has been widely applied to various tasks in particle physics. However, existing models lack precise control over their learned representations, limiting physical interpretability and hindering their use for accurate measurements. We propose the Histogram AutoEncoder (HistoAE), an unsupervised representation learning network featuring a custom histogram-based loss that enforces a physically structured latent space. Applied to silicon microstrip detectors, HistoAE learns an interpretable two-dimensional latent space corresponding to the particle's charge and impact position. After simple post-processing, it achieves a charge resolution of $0.25\,e$ and a position resolution of $3\,\mu\mathrm{m}$ on beam-test data, comparable to the conventional approach. These results demonstrate that unsupervised deep learning models can enable physically meaningful and quantitatively precise measurements. Moreover, the generative capacity of HistoAE enables straightforward extensions to fast detector simulations.

阅读与讨论 → 访问原文 →
22.
arXiv (CS.LG) 2026-06-24

SkillHone: A Harness for Continual Agent Skill Evolution Through Persistent Decision History

作者:
Zhiwei Li↗Yong Hu↗

arXiv:2606.08671v2 Announce Type: replace Abstract: Agent skills extend language-model agents with task-specific procedures, scripts, and references, but the tasks and environments they target continually change. Existing methods improve skills in bounded runs and retain only the final artifact, discarding the decision history that later agents need to interpret prior revisions, evaluations, and rejected alternatives. We introduce SkillHone, a harness for continual agent skill evolution grounded in persistent decision history. SkillHone pairs skill revisions with evaluation-side evidence that supplies practice feedback, recording structured histories of diagnoses, revisions, evidence, and outcomes. Role-separated subagents run candidate skills on practice probes with redacted reporting and propose revisions informed by prior decisions, enabling cross-session refinement without rediscovering past rationale. On deep-research benchmarks, SkillHone runs without a pre-integrated search stack and outperforms the commercially backed deep-research agent by 15.8 points on GAIA and 3.2 points on WebWalkerQA-EN, while also exceeding prior skill-evolution methods. We further deploy SkillHone on internal tool-mediated analysis scenarios, where it improves accuracy by an average of 18.8 points across seven settings.

阅读与讨论 → 访问原文 →
23.
arXiv (CS.LG) 2026-06-19

A Model-Driven Approach for Developing Families of Reinforcement Learning Environments

作者:
Xiaoran Liu↗Istvan David↗

arXiv:2606.20324v1 Announce Type: cross Abstract: Virtual training environments are software-intensive systems in which reinforcement learning (RL) agents learn, adapt, and demonstrate meaningful behavior. Virtual training environments offer a safe and cost-efficient alternative to training agents in real-world settings. However, to converge, most realistic RL problems require training in multiple, mostly similar but slightly different environments - i.e., families of environment variants. The typical development process of environment families is a labor-intensive and error-prone manual endeavor that does not scale well. To alleviate these issues, in this paper, we propose a model-driven approach for developing families of RL training environments. To obtain the family of environments, we develop an approach and prototype tool. In our approach, a hybrid genetic algorithm - a combination of population-based global search and heuristic local search - generates environment families. Mutations and constraints are expressed as model transformations and are operationalized into a search process by a state-of-the-art model transformation engine. We demonstrate the soundness of our approach in a wildfire mitigation scenario and curriculum learning - a particular learning paradigm that relies on environment families.

阅读与讨论 → 访问原文 →
25.
arXiv (CS.AI) 2026-06-11

Sparse probes and murky physics: a case study of interpretability challenges in a foundation model for continuum dynamics

作者:
Katherine Rosenfeld↗Maike Sonnewald↗

arXiv:2606.11657v1 Announce Type: cross Abstract: Generative AI emulators are increasingly used in scientific domains where we already have strong theory, benchmarks, and physical intuition. This raises a central evaluation and interpretability question: when a foundation-style model can reproduce known continuum dynamics, what internal mechanism supports that behavior, is the internal behaviour consistent with known physics, and how does it relate to where the emulator succeeds or fails? We investigate a cross-domain foundation model for continuum dynamics, Walrus by Polymathic, using mechanistic interpretability guided by physical principles. We apply a sparse autoencoder (SAE) to probe a selected layer, and address the practical challenge of triaging a large feature set (over 20,000) using enstrophy as a physically grounded metric. As a deliberately simple testbed, we focus on shear flow and compare feature recruitment across multiple shear-flow setups, i.e. parameter values in the numerical simulation. Across setups we find evidence of piecewise consistency, with subsets of features recurring in similar roles, but this structure is intermittent and does not map cleanly onto standard physical decompositions. In parallel, direct comparisons between numerical simulation and the emulator reveal systematic output-level discrepancies, including regimes where energy/structures become too diffuse or too localized. We connect parts of these discrepancies to changes in specific SAE feature usage. Our work highlights open questions for scientific foundation models: how to robustly prioritize mechanistically meaningful features, how to separate stable structure from analysis artifacts (including single-layer and SAE limitations), and how to use established benchmarks to decide when "different" internal representations are genuinely informative rather than merely effective.

阅读与讨论 → 访问原文 →