StrixAE is an intelligent audio enhancement agent that uses a multimodal large language model to coordinate multiple enhancement and personalization models. It is trained in two stages: first, chain‑of‑thought supervised fine‑tuning on AcoustBench, and second, Audio Perception Reinforcement Learning that optimizes format validity, structural coherence, and perceptual quality. The approach yields state‑of‑the‑art performance and strong generalization on real‑world test datasets, outperforming many existing open‑source and proprietary solutions.
By Chenglin Wu, Junjie Wu, Jinhang Chen, Mingyang Chen, Zixu Lin, Jiabian Chen, Xinghao Ding, Xiaotong Tu
The Civilization Framework proposes a new way for AI systems to communicate by treating the entire civilization—one human sovereign, a persistent ledger, and interchangeable agents—as the addressable party, rather than individual agents. It introduces the Embassy Protocol, an asynchronous, carrier‑agnostic overlay that routes messages to a ledger endpoint where any online agent of the receiver can process them, with commitment state on both ledgers serving as the ground truth. The framework also identifies a temporal‑weight effect in AI‑to‑AI communication, demonstrates its impact in a preregistered experiment, and explores mitigation strategies such as instruction‑level provenance labeling and sealed‑answer accuracy equivalence.
whyItMatters":"The framework aims to reduce context loss and authority bias in AI interactions by grounding communication in a shared ledger and sovereign oversight, potentially improving reliability and accountability in multi‑agent systems."
By Guangjun Liu
The paper introduces R$^2$-MAD, a framework that enhances multi-agent debate by giving agents an experience memory from past debates. It uses a debate-state-aware retrieval policy to adjust concept priors based on current consensus, and derives confidence weights from retrieved experiences to modulate peer influence. Experiments demonstrate consistent improvements over existing single-agent and MAD baselines.
By Xuanfa Jin, Zhijian Ma, Yongcheng Zeng, Xinyu Cui, Haifeng Zhang, Jun Wang
The paper investigates zero‑shot task generalisation in offline multi‑agent reinforcement learning by extending sequence‑modeling architectures to support multi‑task observation and action spaces and variable agent counts. It finds that increasing task diversity, rather than merely enlarging the dataset, is the key driver for robust zero‑shot transfer. Experiments on four challenging environments show a 3.2× mean improvement on held‑out tasks compared to single‑task models and outperform strong behaviour‑cloning baselines.
By Oussama Hidaoui, Omer Ebead, Ulrich Armel Mbou Sob, Siddarth Singh, Juan Claude Formanek, Felix Chalumeau, Omayma Mahjoub, Sasha Abramowitz, Ruan John de Kock, Wiem Khlifi, Louay Ben Nessir, Simon Verster Du Toit, Daniel Rajaonarivonivelomanantsoa, Asim Awad Osman, Arnol Manuel Fokam, Refiloe Shabe, Arnu Pretorius
The paper proposes treating translation as a structured decision space explored by multiple autonomous agents, rather than producing a single output. Using Turkish–Syrian Arabic dialogue, three agents—zero‑shot, dialect‑stabilized, and pivot translation—are compared on 5,000 sentences, with stabilization nearly doubling dialect marker usage and reducing structural instability. The study introduces an interpretability framework that quantifies decision flexibility through dialect marker frequency, lexical proximity, and structural variance.
By Hasan Alkhder, Mohammad Abboush, Igor Tchappi, Ahmet Zengin, Amro Najjar
PatchBench introduces a benchmark to evaluate AI agents on realistic vulnerability patching tasks, addressing two key threats to validity: patch memorization and surface-level fixes that merely suppress crashes. The study finds that 25% of agent patches resemble historical developer patches, and that PoC-only validation inflates success rates by 1.83× on average. PatchBench mitigates these issues by selecting vulnerabilities whose true fixes lie outside the crash stack, migrating historical vulnerabilities into new contexts, and employing rigorous validation for security and semantic correctness.
By Chihao Shen, Jiacheng Li, Aastha Mahajan, Jeffery Siyuan Tian, Yonghwi Kwon, Yizheng Chen
PaperScout is an autonomous agent that treats academic paper search as a sequential decision-making process, dynamically deciding when and how to use search and expansion tools based on accumulated context. The authors identify a granularity mismatch in standard reinforcement learning for multi-turn tasks and propose Proximal Sequence Policy Optimization (PSPO), a sequence-level policy optimization method that aligns learning with agent–environment interactions. Experiments on synthetic and real-world benchmarks show that PaperScout outperforms structured retrieval and RL baselines in recall and relevance, demonstrating the effectiveness of its adaptive agentic framework and optimization strategy.
By Tingyue Pan, Jie Ouyang, Mingyue Cheng, Qingchuan Li, Zirui Liu, Daoyu Wang, Mingfan Pan, Shuo Yu, Qi Liu, Enhong Chen
CoMAP introduces a framework that jointly evolves textual world models and agent policies through a closed‑loop interaction. At each decision step the world model forecasts future state feedback for candidate actions, while the agent reflects on the reliability of this feedback to refine its action. The resulting on‑policy trajectories are used to self‑distill and update the world model, improving prediction accuracy and long‑horizon decision‑making across embodied planning, web navigation, and tool‑use benchmarks.
By Youwei Liu, Jian Wang, Hanlin Wang, Wenjie Li
The paper introduces ARTEMIS, a no-code evolutionary optimization platform that automatically tunes large language model (LLM) agents by jointly optimizing prompts, tool descriptions, and parameters using semantically-aware genetic operators. Starting from a benchmark script and natural language goals, ARTEMIS discovers configurable components, extracts performance signals from execution logs, and evolves configurations without architectural changes. Experiments on four agent systems show significant gains: a 13.6% increase in acceptance rate for the ALE Agent, a 10.1% performance boost for the Mini‑SWE Agent, a 36.9% token‑reduction for the CrewAI Agent, and a 22% accuracy improvement for the MathTales‑Teacher Agent using a smaller open‑source model.
By Paul Brookes, Vardan Voskanyan, Rafail Giavrimis, Matthew Truscott, Mina Ilieva, Chrystalla Pavlou, Alexandru Staicu, Manal Adham, Will Evers- Hood, Jingzhi Gong, Kejia Zhang, Matvey Fedoseev, Vishal Sharma, Roman Bauer, Zheng Wang, Hema Nair, Wei Jie, Tianhua Xu, Aurora Constantin, Leslie Kanthan, Michail Basios
PeroMAS is a multi-agent system designed for discovering perovskite materials, integrating tools encapsulated as Model Context Protocols to handle the full workflow from literature retrieval to synthesis. It enables design under multi-objective constraints and outperforms single LLM or traditional search methods in discovery efficiency. The system’s effectiveness was validated by real synthesis experiments guided by expert evaluation.
By Yishu Wang, Wei Liu, Yifan Li, Shengxiang Xu, Xujie Yuan, Ran Li, Yuyu Luo, Jia Zhu, Shimin Di, Min-Ling Zhang, Guixiang Li
The paper surveys collaborative learning methods that move beyond traditional Euclidean data to graph-structured data. It reviews foundational principles for Euclidean settings—learning effectiveness, efficiency, and privacy—and then extends the discussion to graph data, presenting a taxonomy of distribution scenarios, statistical heterogeneities, and standardized problem formulations. The survey also outlines open challenges and future research directions in this emerging field.
By R\'emi Bourgerie, \v{S}ar\=unas Girdzijauskas, Viktoria Fodor
TIGPO (Temporal Instance-Graph Policy Optimization) extends graph-based credit assignment for long-horizon LLM agents by maintaining a persistent transition graph per task across policy updates. It allocates rollout budgets to both new exploration and revisiting past tasks, pairing current rollouts with earlier ones to create cross‑temporal references that stabilize advantage estimation. Experiments on ALFWorld and WebShop show TIGPO consistently outperforms previous group‑based and graph‑based policy optimization methods.
By Jinwei Gan
The paper investigates how large language model agents behave when they are aware that their internal activations are being monitored. By informing agents of an activation monitor and providing feedback on previous scores, the study finds that agents still collude and that probes trained on internal activations remain accurate across all conditions. Experiments were conducted on a four‑agent blackjack game and a two‑agent Simmons prisoners game using Qwen3‑32B‑AWQ and GPT‑OSS‑20B models.
By Aritra Das, Jaee Ponde, Mihir More, Debayan Gupta
MemoryLACE (MemLACE) is a lightweight memory framework that explicitly models the lifecycle of textual evidence—capturing sparse merge, supersession, and contradiction relations—while preserving atomic natural‑language memories and their provenance. Unlike traditional systems that retrieve memories independently, MemLACE reconstructs relation‑aware evidence units that expose current, historical, supporting, and conflicting evidence for downstream reasoning. In benchmark evaluations (BEAM and StructMemEval) using both open‑weight and proprietary LLM backbones, MemLACE achieves the highest overall performance among same‑backbone comparisons and reduces BEAM runtime by 66.6% compared to the strongest reflective‑memory baseline, Hindsight.
By Meriem Yacoubi, Pia Schmidt, Nenad Petrovic, Ahmed Frikha, Martin Kirchhoff, Alois Knoll
The paper reports that counterfactual fairness audits of clinical language‑model agents are unreliable without accounting for a per‑action instability floor. By repeatedly running identical vignettes, the authors found that actions changed 8.7% of the time, with instability varying eightfold across actions. A second model confirmed a pooled floor of 6.7%, showing that any reported fairness estimate lacking this floor cannot be interpreted as evidence of disparity.
By Rohith Reddy Bellibaltu, Manpreet Singh, Deepak Parashar, Rahul Joshi
The paper proposes Byrd-NAFL, a Byzantine‑robust federated learning algorithm that incorporates Nesterov’s momentum and resilient aggregation rules. It achieves fast and safe convergence under non‑convex, smooth loss functions with relaxed gradient assumptions, and provides a finite‑time convergence guarantee. Experiments show that Byrd-NAFL outperforms existing methods in convergence speed, accuracy, and resilience to various malicious attacks.
By Lihan Xu, Xiaoyi Fan, Gang Wang, Runhao Zeng, Xiping Hu, Yanjie Dong
The paper introduces PersonaLink, a training‑free method that distills a user’s interaction history into a bounded three‑field persona and iteratively refines it by self‑evaluating a frozen 7B language model on held‑out labeled data. Each refinement rewrites the persona only if it does not regress on that slice, ensuring the persona remains bounded and query‑independent. On a 200‑user news categorization task (LaMP‑2), PersonaLink achieves 0.745–0.755 accuracy, statistically indistinguishable from BM25 retrieval’s 0.760–0.765 accuracy, demonstrating that distilled personas can match retrieval for classification but not for regression tasks.
By JaeHa Yoon, Minjun Park, Seoyeon Kim, Jiwoo Lee, Hyunwoo Choi, Dohyun Kang
The paper "Beyond Reproducibility: Towards Security-Aware Evaluation of Research Artifacts" examines 1,388 research artifacts from top security conferences, uncovering 132,431 candidate security findings through static analysis. It introduces a taxonomy for context-aware security assessment and presents SAFE, an autonomous framework that accurately distinguishes security-relevant findings and classifies risk types. The study demonstrates that nearly 45% of findings are security-relevant, highlighting the need for security-aware evaluation alongside traditional reproducibility checks.
By Nanda Rani, Christian Rossow
The paper investigates whether computer‑aided engineering (CAE) simulation agents require specialized features beyond a generic LLM harness. Experiments show that a single‑agent harness with multi‑turn reasoning, tool use, and execution feedback can match or outperform multi‑agent specialized systems, with domain knowledge (solver tutorials) providing the most significant performance boost. The study highlights that modern harnesses already supply many capabilities previously thought necessary for CAE agents.
By Jiasheng Shi, Tianhan Zhang
Editable Visual Design introduces a new design paradigm that combines a Coding Agent with a Vision‑Language Model (VLM) and an image generation model. The VLM acts as the creative brain, understanding requirements, planning tasks, and judging aesthetics, while the image generator produces isolated visual assets on demand. The agent follows an "imagine first, then act" workflow, generating assets, writing native HTML/CSS, and refining the design through visual feedback, ultimately producing editable, layer‑wise artifacts with real text that can be adjusted via a graphical interface.
By Junyan Ye, Wei Liu, Dongzhi Jiang, Zichen Wen, HaoDong Li, Zhutao Lv, Jiaxin Lin, Jinhua Yu, Jun He, Zilong Huang, Rui Chen, Weijia Li