论文分享 | 智能体相关研究进展

我们从2024-11-18到2024-11-22的45篇文章中精选出5篇优秀的工作分享给读者。

1. Exploring the Impact of Non-Verbal Virtual Agent Behavior on User Engagement in Argumentative Dialogues
2. Syllabus: Portable Curricula for Reinforcement Learning Agents
3. Learning to Cooperate with Humans using Generative Agents
4. Physics-Informed LLM-Agent for Automated Modulation Design in Power Electronics Systems
5. Towards Full Delegation: Designing Ideal Agentic Behaviors for Travel Planning

1.Exploring the Impact of Non-Verbal Virtual Agent Behavior on User Engagement in Argumentative Dialogues

Authors: Annalena Bea Aicher, Yuki Matsuda, Keichii Yasumoto, Wolfgang Minker, Elisabeth André, Stefan Ultes

https://arxiv.org/abs/2411.11102

论文摘要

Engaging in discussions that involve diverse perspectives and exchanging arguments on a controversial issue is a natural way for humans to form opinions. In this process, the way arguments are presented plays a crucial role in determining how engaged users are, whether the interaction takes place solely among humans or within human-agent teams. This is of great importance as user engagement plays a crucial role in determining the success or failure of cooperative argumentative discussions. One main goal is to maintain the user’s motivation to participate in a reflective opinion-building process, even when addressing contradicting viewpoints. This work investigates how non-verbal agent behavior, specifically co-speech gestures, influ ences the user’s engagement and interest during an ongoing argu mentative interaction. The results of a laboratory study conducted with 56 participants demonstrate that the agent’s co-speech ges tures have a substantial impact on user engagement and interest and the overall perception of the system. Therefore, this research offers valuable insights for the design of future cooperative argumentative virtual agents.

论文简评

这篇关于虚拟助手与用户互动的研究论文，主要探讨了语音交互中手势交流对用户参与度的影响。通过一项实验研究，研究人员发现，在辩论对话中，这些非言语交流方式能够显著提高用户的参与感和兴趣。这项研究成果对于设计具有合作性的虚拟对话助手具有重要意义，为未来的研究提供了宝贵的视角。

首先，该研究聚焦于一个重要的主题——人机交互中的用户体验。随着技术的发展，越来越多的人工智能（AI）系统开始与人类进行交互，而如何提升这种交互体验成为了研究的重要方向之一。本研究的成果表明，利用合适的非语言沟通方式，可以有效增强用户与虚拟助手之间的互动，从而改善他们的满意度和参与度。

其次，该研究采用了严谨的设计方法，并取得了积极的结果。通过实验室中的参与者研究，研究人员成功地证明了手势交流确实能显著影响用户的情绪反应和参与程度。这一发现不仅丰富了对虚拟助手的理解，也为我们设计更高效、人性化的AI系统提供了参考。

最后，该研究强调了非语言沟通的重要性。尽管人们通常忽视非言语交流的作用，但它们在促进人机界面的友好性方面发挥着不可替代的角色。因此，未来的AI研发应更加重视非语言元素的运用，以更好地满足用户的需求和期望。

总的来说，这篇论文提供了一种全面且实用的方法来评估虚拟助手的用户体验，特别是在与用户的对话时。它展示了如何通过合理的非语言沟通策略提高用户参与度，以及这一过程是如何被实证研究证实的。因此，这篇论文不仅是对现有理论的补充，也是对未来研究的启示。

2.Syllabus: Portable Curricula for Reinforcement Learning Agents

Authors: Ryan Sullivan, Ryan Pégoud, Ameen Ur Rahmen, Xinchen Yang, Junyun Huang, Aayush Verma, Nistha Mitra, John P. Dickerson

https://arxiv.org/abs/2411.11318

论文摘要

Curriculum learning has been a quiet yet crucial component of many of the high-profile successes of reinforcement learning. Despite this, none of the major reinforcement learning libraries directly support curriculum learning or include implementations. These methods can improve the capabilities and robustness of RL agents but often require significant, complex changes to agent training code. We introduce Syllabus, a library for training RL agents with curriculum learning as a solution to this problem. Syllabus provides a universal API for curriculum learning algorithms, implementations of popular methods, and infrastructure for easily integrating them with distributed training code written in nearly any RL library. Syllabus provides a minimal API for each of the core components of curriculum learning, dramatically simplifying the process of designing new algorithms and applying existing algorithms to new environments. We demonstrate that the same Syllabus code can be used to train agents written in multiple different RL libraries across various domains. In doing so, we present the first examples of curriculum learning in NetHack and Neural MMO, two premier challenges for single-agent and multi-agent RL respectively, achieving strong results compared to state-of-the-art baselines.

论文简评

这篇论文深入探讨了Syllabus这一创新工具，其旨在通过一个通用API来简化各种基于强化学习（RL）代理中的课程学习集成。它的目标是在各个环境中使研究人员能够轻松应用和设计不同的算法。此外，该论文还展示了框架的多样性和灵活性，通过在多个具有挑战性的环境中的实验，证明了其强大的功能。

综上所述，Syllabus是一项重要的研究成果，它解决了主流RL库中关于课程学习支持的重大缺口，并提供了一个通用的API，使得不同类型的课程学习方法可以被整合到各种RL库中。这种多功能性以及对多种环境的适应能力，使其成为研究者们进行复杂任务时的理想选择。

3.Learning to Cooperate with Humans using Generative Agents

Authors: Yancheng Liang, Daphne Chen, Abhishek Gupta, Simon S. Du, Natasha Jaques

https://arxiv.org/abs/2411.13934

论文摘要

Training agents that can coordinate zero-shot with humans is a key mission in multi-agent reinforcement learning (MARL). Current algorithms focus on training simulated human partner policies which are then used to train a Cooperator agent. The simulated human is produced either through behavior cloning over a dataset of human cooperation behavior or by using MARL to create a population of simulated agents. However, these approaches often struggle to produce a Cooperator that can coordinate well with real humans, since the simulated humans fail to cover the diverse strategies and styles employed by people in the real world. We show that learning a generative model of human partners can effectively address this issue. Our model learns a latent variable representation of the human that encodes the human’s unique strategy, intention, experience, or style. This generative model can be flexibly trained from any human or neural policy agent interaction data. By sampling from the latent space, we can use the generative model to produce different partners to train Cooperator agents. We evaluate our method—Generative Agent Modeling for Multi-agent Adaptation (GAMMA)—on Overcooked, a challenging cooperative cooking game that has become a standard benchmark for zero-shot coordination. We conduct evaluations with real human teammates, and the results show that GAMMA consistently improves performance, whether the generative model is trained on simulated populations or human datasets. Further, we propose a method for posterior sampling from the generative model that is biased toward the human data, enabling us to efficiently improve performance with only a small amount of expensive human interaction data.

论文简评

总体来说，《Gamma：一种生成模型以改善AI与人类伙伴协同合作能力的多代理强化学习设置》这篇论文是一项重要研究，旨在解决在多代理强化学习中人工智能（AI）与人类合作伙伴协调的问题。通过从学习到的隐空间提取人类行为样本，Gamma能够使AI灵活地适应和优化合作策略。实验结果显示，基于Gamma的改进方法在Overcooked游戏中的表现优于传统方法，且传统方法多依赖模拟或有限的人类数据。此外，论文还提供了详细的评估，包括对真实人类参与者的实验证明了GAMMA的实际应用性。总的来说，Gamma的研究成果对于提升人工智能与人类合作的有效性和效率具有重要意义。

4.Physics-Informed LLM-Agent for Automated Modulation Design in Power Electronics Systems

Authors: Junhua Liu, Fanfan Lin, Xinze Li, Kwan Hui Lim, Shuai Zhao

https://arxiv.org/abs/2411.14214

论文摘要

Large language models (LLM) have demonstrated superb performance in natural language generation and reasoning tasks. In the quest for carbon neutrality and a high-performance renewable energy power system, existing AI-assisted design automation systems have severe limitations in explainability, scalability, and usability. In this work, we explore the use of LLM with an intuitive interface to automate the design of power converter modulation strategy. We introduce LLM-controlled Physics-informed Power Converter Modulation Design Automation (LP-COMDA), a design system that employs a frontend of LLM-powered chat interface to facilitate the collection and validation of design specifications, and a backend of hierarchical Physics-Informed Neural Network (PINN) surrogate models that take valid specifications to generate optimal power converter modulation design. Experiments show that LP-COMDA outperforms all baseline methods, achieving a 63.2% reduction in error compared to the second-best benchmark method regarding the standard mean absolute error. Furthermore, empirical studies with 20 experts show that design time with LP-COMDA is over 33 times faster than conventional methods.

论文简评

本文探讨了一种名为LP-COMDA的物理指导语言模型控制自适应代理的设计，旨在自动化电力电子系统中功率转换的调制设计。这一目标是通过解决现代系统复杂性所带来的重大挑战来实现，聚焦于提高系统的解释性、可扩展性和易用性。通过与物理启发方法相结合，LP-COMDA展示了显著的误差降低和设计时间减少，证明了其有效性。此外，它还强调了LLMs在工程与设计中的潜在应用价值。总的来说，本文展示了一个强大的解决方案，旨在为复杂的电力电子系统提供高效的调制设计自动化，具有重要的理论与实践意义。

5.Towards Full Delegation: Designing Ideal Agentic Behaviors for Travel Planning

Authors: Song Jiang, Da JU, Andrew Cohen, Sasha Mitts, Aaron Foss, Justine T Kao, Xian Li, Yuandong Tian

https://arxiv.org/abs/2411.13904

论文摘要

How are LLM-based agents used in the future? While many of the existing work on agents has focused on improving the performance of a specific family of ob jective and challenging tasks, in this work, we take a different perspective by thinking about full delegation: agents take over humans’ routine decision-making processes and are trusted by humans to find solutions that fit people’s personal ized needs and are adaptive to ever-changing context. In order to achieve such a goal, the behavior of the agents, i.e., agentic behaviors, should be evaluated not only on their achievements (i.e., outcome evaluation), but also how they achieved that (i.e., procedure evaluation). For this, we propose APEC Agent Constitution, a list of criteria that an agent should follow for good agentic behaviors, including Accuracy, Proactivity, Efficiency and Credibility. To verify whether APEC aligns with human preferences, we develop APEC-Travel, a travel planning agent that proactively extracts hidden personalized needs via multi-round dialog with trav elers. APEC-Travel is constructed purely from synthetic data generated by Llama3.1-405B-Instruct with a diverse set of travelers’ persona to simulate rich distribution of dialogs. Iteratively fine-tuned to follow APEC Agent Constitution, APEC-Travelsurpasses baselines by 20.7% on rule based metrics and 9.1% on LLM-as-a-Judge scores across the constitution axes.

论文简评

这篇论文提出了一个名为APEC-Travel的旅行规划代理系统，旨在通过一种新的提议的APEC代理人宪法来增强代理行为，并展示其优于基准模型的能力。该系统通过对多轮对话和使用合成数据迭代训练，能够理解旅客的偏好。此外，作者还展示了全权委托决策任务给代理机构的实际问题，表明了该系统的实用性。总的来说，这篇文章提出了一种新的评估框架，强调AI领域的现实应用，特别是在旅行计划方面。

我们欢迎您在评论区中留下宝贵的建议！包括但不限于：

• 可以提出推文中论文简评的不足！
• 可以分享最近更值得推荐的论文并给出理由！

推荐阅读