Agent-based systems, coordination, cooperation
Jizhan Fang, Buqiang Xu, Zhixian Wang et al.
For LLM agents operating in dynamic environments, FluxMem addresses the brittleness of static memory by enabling adaptive connectivity evolution, leading to consistent SOTA results across diverse benchmarks.
Fanqi Kong, Ruijie Zhang, Huaxiao Yin et al.
This work addresses the reliability and interpretability of multi-agent systems for developers and researchers, though it is incremental as it builds on existing error attribution methods with automated data generation.
Qitong Sun, Jun Han, Tianlin Li et al.
This addresses GPU kernel optimization for AI systems, offering a more interpretable and efficient approach compared to prior LLM-based methods.
Yiwen Song, Tomas Pfister, Yale Song
This addresses the problem of inefficient or inaccessible video quality evaluation for users of video generation models, offering a novel black-box optimization method.
Yichang Xu, Gaowen Liu, Ramana Rao Kompella et al. · gatech
This addresses the challenge of scaling video reasoning to real-world long videos for applications like video question-answering, though it appears incremental as an optimization of existing multi-agent and VLM approaches.
Wenhao Wang, Haoting Shi, Mengying Yuan et al.
For researchers developing privacy-preserving GUI agents, this benchmark enables systematic study of cross-platform heterogeneity, filling a gap in existing federated learning benchmarks.
Yuchen Ma, Yue Huang, Han Bao et al.
For LLM agent developers, SkillGen automates the creation of high-quality, reusable skills that improve agent performance without retraining, addressing the bottleneck of manual skill authoring.
Taiyi Wang, Sian Gooding, Florian Hartmann et al.
This addresses the challenge of autonomous control in dynamic digital environments for AI developers, offering a novel approach to enhance agent robustness, though it is incremental in combining planning and RL techniques.
Chenyu Zhou, Huacan Chai, Wenteng Chen et al.
This provides a unified review framework for researchers and practitioners building LLM agents, though it is primarily conceptual rather than presenting new experimental results.
Yihao Zhang, Zeming Wei, Xiaokun Luan et al.
This addresses critical security risks for users of interconnected multi-agent systems, exposing vulnerabilities that could lead to autonomous attacks without attacker intervention.
Xiaoyu Ma, Lianyu Hu, Wenbing Tang et al.
This addresses the challenge of persistent memory for embodied agents in complex 3D environments, offering a scalable solution for generalizable embodied intelligence.
Rui Meng, Bhavana Dalvi Mishra, Jiefeng Chen et al.
For researchers and developers of autonomous AI systems, this work addresses the critical problem of output verifiability, providing a framework and system that eliminates common failure modes like hallucinated references and unreproducible results.
Yi Zhong, Buqiang Xu, Yijun Wang et al.
For developers and practitioners in industrial automation, this benchmark addresses the costly and error-prone manual construction of executable visual workflows by evaluating LLMs' ability to automate this process.
Minh V. T. Thai, Tue Le, Dung Nguyen Manh et al.
This addresses the need for better benchmarks in AI-assisted software engineering to reflect real-world development challenges, though it is incremental as it builds on existing benchmarking efforts.
Kai Wang, Biaojie Zeng, Zeming Wei et al.
This addresses safety risks for developers and users of multi-agent systems, though it appears incremental as it builds on existing standards like OWASP.
Jing Yu Koh, Ruslan Salakhutdinov, Daniel Fried
For researchers and practitioners building computer use agents, this work addresses the limitations of single-agent systems for complex, long-horizon tasks by introducing a multi-agent coordination framework.
Qingyang Shao, Xin Zhang, Zhouyang Yuan et al.
This work addresses the challenge of maintaining data, computations, and evidence connections in long-horizon bioinformatics tasks for researchers and practitioners, offering a substantial improvement over existing LLM agent systems.
Zhaopeng Feng, Liangcai Su, Zhen Zhang et al.
Addresses the critical bottleneck of finite context capacity in LLM-based autonomous agents for long-horizon tasks, offering a principled framework for future context management strategies.
Xiaohang Nie, Zihan Guo, Kezhuo Yang et al.
This addresses the problem of agent interoperability and social integration for developers and users in decentralized digital ecosystems, representing a novel architectural approach rather than an incremental improvement.
Jatin Ganhotra, Sami Serhan, Antonio Abu Nassar et al.
This addresses the need for better automated issue resolution in enterprise software development, where Java is widely used, but it is incremental as it builds on existing agent-based methods with a focus on a specific language.