This paper provides a comprehensive analysis of the challenges and controversies associated with blockchain technology. It identifies technical challenges such as scalability, security, privacy, and interoperability, as well as business and adoption challenges, and the social, economic, ethical, and environmental controversies present in current blockchain systems. We argue that responsible blockchain development is key to overcoming these challenges and achieving mass adoption. This paper defines Responsible Blockchain and introduces the STEADI principles (sustainable, transparent, ethical, adaptive, decentralized, and inclusive) for responsible blockchain development. Additionally, it presents the Actor-Network Theory-based Responsible Development Methodology (ANT-RDM) for blockchains, which includes the steps of problematization, interessement, enrollment, and mobilization.
{"title":"Responsible Blockchain: STEADI Principles and the Actor-Network Theory-based Development Methodology (ANT-RDM)","authors":"Yibai Li, Ahmed Gomaa, Xiaobing Li","doi":"arxiv-2409.06179","DOIUrl":"https://doi.org/arxiv-2409.06179","url":null,"abstract":"This paper provides a comprehensive analysis of the challenges and\u0000controversies associated with blockchain technology. It identifies technical\u0000challenges such as scalability, security, privacy, and interoperability, as\u0000well as business and adoption challenges, and the social, economic, ethical,\u0000and environmental controversies present in current blockchain systems. We argue\u0000that responsible blockchain development is key to overcoming these challenges\u0000and achieving mass adoption. This paper defines Responsible Blockchain and\u0000introduces the STEADI principles (sustainable, transparent, ethical, adaptive,\u0000decentralized, and inclusive) for responsible blockchain development.\u0000Additionally, it presents the Actor-Network Theory-based Responsible\u0000Development Methodology (ANT-RDM) for blockchains, which includes the steps of\u0000problematization, interessement, enrollment, and mobilization.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"42 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Siddharth Chaturvedi, Ahmed El-Gazzar, Marcel van Gerven
Foraging for resources is a ubiquitous activity conducted by living organisms in a shared environment to maintain their homeostasis. Modelling multi-agent foraging in-silico allows us to study both individual and collective emergent behaviour in a tractable manner. Agent-based modelling has proven to be effective in simulating such tasks, though scaling the simulations to accommodate large numbers of agents with complex dynamics remains challenging. In this work, we present Foragax, a general-purpose, scalable, hardware-accelerated, multi-agent foraging toolkit. Leveraging the JAX library, our toolkit can simulate thousands of agents foraging in a common environment, in an end-to-end vectorized and differentiable manner. The toolkit provides agent-based modelling tools to model various foraging tasks, including options to design custom spatial and temporal agent dynamics, control policies, sensor models, and boundary conditions. Further, the number of agents during such simulations can be increased or decreased based on custom rules. The toolkit can also be used to potentially model more general multi-agent scenarios.
{"title":"Foragax: An Agent Based Modelling framework based on JAX","authors":"Siddharth Chaturvedi, Ahmed El-Gazzar, Marcel van Gerven","doi":"arxiv-2409.06345","DOIUrl":"https://doi.org/arxiv-2409.06345","url":null,"abstract":"Foraging for resources is a ubiquitous activity conducted by living organisms\u0000in a shared environment to maintain their homeostasis. Modelling multi-agent\u0000foraging in-silico allows us to study both individual and collective emergent\u0000behaviour in a tractable manner. Agent-based modelling has proven to be\u0000effective in simulating such tasks, though scaling the simulations to\u0000accommodate large numbers of agents with complex dynamics remains challenging.\u0000In this work, we present Foragax, a general-purpose, scalable,\u0000hardware-accelerated, multi-agent foraging toolkit. Leveraging the JAX library,\u0000our toolkit can simulate thousands of agents foraging in a common environment,\u0000in an end-to-end vectorized and differentiable manner. The toolkit provides\u0000agent-based modelling tools to model various foraging tasks, including options\u0000to design custom spatial and temporal agent dynamics, control policies, sensor\u0000models, and boundary conditions. Further, the number of agents during such\u0000simulations can be increased or decreased based on custom rules. The toolkit\u0000can also be used to potentially model more general multi-agent scenarios.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"113 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190500","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Contemporary research in autonomous driving has demonstrated tremendous potential in emulating the traits of human driving. However, they primarily cater to areas with well built road infrastructure and appropriate traffic management systems. Therefore, in the absence of traffic signals or in unstructured environments, these self-driving algorithms are expected to fail. This paper proposes a strategy for autonomously navigating multiple vehicles in close proximity to their desired destinations without traffic rules in unstructured environments. Graphical Neural Networks (GNNs) have demonstrated good utility for this task of multi-vehicle control. Among the different alternatives of training GNNs, supervised methods have proven to be most data-efficient, albeit require ground truth labels. However, these labels may not always be available, particularly in unstructured environments without traffic regulations. Therefore, a tedious optimization process may be required to determine them while ensuring that the vehicles reach their desired destination and do not collide with each other or any obstacles. Therefore, in order to expedite the training process, it is essential to reduce the optimization time and select only those samples for labeling that add most value to the training. In this paper, we propose a warm start method that first uses a pre-trained model trained on a simpler subset of data. Inference is then done on more complicated scenarios, to determine the hard samples wherein the model faces the greatest predicament. This is measured by the difficulty vehicles encounter in reaching their desired destination without collision. Experimental results demonstrate that mining for hard samples in this manner reduces the requirement for supervised training data by 10 fold. Videos and code can be found here: url{https://yininghase.github.io/multiagent-collision-mining/}.
{"title":"Enhancing the Performance of Multi-Vehicle Navigation in Unstructured Environments using Hard Sample Mining","authors":"Yining Ma, Ang Li, Qadeer Khan, Daniel Cremers","doi":"arxiv-2409.05119","DOIUrl":"https://doi.org/arxiv-2409.05119","url":null,"abstract":"Contemporary research in autonomous driving has demonstrated tremendous\u0000potential in emulating the traits of human driving. However, they primarily\u0000cater to areas with well built road infrastructure and appropriate traffic\u0000management systems. Therefore, in the absence of traffic signals or in\u0000unstructured environments, these self-driving algorithms are expected to fail.\u0000This paper proposes a strategy for autonomously navigating multiple vehicles in\u0000close proximity to their desired destinations without traffic rules in\u0000unstructured environments. Graphical Neural Networks (GNNs) have demonstrated good utility for this task\u0000of multi-vehicle control. Among the different alternatives of training GNNs,\u0000supervised methods have proven to be most data-efficient, albeit require ground\u0000truth labels. However, these labels may not always be available, particularly\u0000in unstructured environments without traffic regulations. Therefore, a tedious\u0000optimization process may be required to determine them while ensuring that the\u0000vehicles reach their desired destination and do not collide with each other or\u0000any obstacles. Therefore, in order to expedite the training process, it is\u0000essential to reduce the optimization time and select only those samples for\u0000labeling that add most value to the training. In this paper, we propose a warm\u0000start method that first uses a pre-trained model trained on a simpler subset of\u0000data. Inference is then done on more complicated scenarios, to determine the\u0000hard samples wherein the model faces the greatest predicament. This is measured\u0000by the difficulty vehicles encounter in reaching their desired destination\u0000without collision. Experimental results demonstrate that mining for hard\u0000samples in this manner reduces the requirement for supervised training data by\u000010 fold. Videos and code can be found here:\u0000url{https://yininghase.github.io/multiagent-collision-mining/}.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"34 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190502","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kang Wang, Zhishu Shen, Zhenwei Wang, Tiehua Zhang
Deep reinforcement learning (DRL) methods that incorporate graph neural networks (GNNs) have been extensively studied for intelligent traffic signal control, which aims to coordinate traffic signals effectively across multiple intersections. Despite this progress, the standard graph learning used in these methods still struggles to capture higher-order correlations in real-world traffic flow. In this paper, we propose a multi-agent proximal policy optimization framework DHG-PPO, which incorporates PPO and directed hypergraph module to extract the spatio-temporal attributes of the road networks. DHG-PPO enables multiple agents to ingeniously interact through the dynamical construction of hypergraph. The effectiveness of DHG-PPO is validated in terms of average travel time and throughput against state-of-the-art baselines through extensive experiments.
{"title":"Towards Multi-agent Policy-based Directed Hypergraph Learning for Traffic Signal Control","authors":"Kang Wang, Zhishu Shen, Zhenwei Wang, Tiehua Zhang","doi":"arxiv-2409.05037","DOIUrl":"https://doi.org/arxiv-2409.05037","url":null,"abstract":"Deep reinforcement learning (DRL) methods that incorporate graph neural\u0000networks (GNNs) have been extensively studied for intelligent traffic signal\u0000control, which aims to coordinate traffic signals effectively across multiple\u0000intersections. Despite this progress, the standard graph learning used in these\u0000methods still struggles to capture higher-order correlations in real-world\u0000traffic flow. In this paper, we propose a multi-agent proximal policy\u0000optimization framework DHG-PPO, which incorporates PPO and directed hypergraph\u0000module to extract the spatio-temporal attributes of the road networks. DHG-PPO\u0000enables multiple agents to ingeniously interact through the dynamical\u0000construction of hypergraph. The effectiveness of DHG-PPO is validated in terms\u0000of average travel time and throughput against state-of-the-art baselines\u0000through extensive experiments.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190503","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We study interactions between agents in multi-agent systems, in which the agents are misinformed with regards to the game that they play, essentially having a subjective and incorrect understanding of the setting, without being aware of it. For that, we introduce a new game-theoretic concept, called misinformation games, that provides the necessary toolkit to study this situation. Subsequently, we enhance this framework by developing a time-discrete procedure (called the Adaptation Procedure) that captures iterative interactions in the above context. During the Adaptation Procedure, the agents update their information and reassess their behaviour in each step. We demonstrate our ideas through an implementation, which is used to study the efficiency and characteristics of the Adaptation Procedure.
{"title":"Adaptation Procedure in Misinformation Games","authors":"Konstantinos Varsos, Merkouris Papamichail, Giorgos Flouris, Marina Bitsaki","doi":"arxiv-2409.04854","DOIUrl":"https://doi.org/arxiv-2409.04854","url":null,"abstract":"We study interactions between agents in multi-agent systems, in which the\u0000agents are misinformed with regards to the game that they play, essentially\u0000having a subjective and incorrect understanding of the setting, without being\u0000aware of it. For that, we introduce a new game-theoretic concept, called\u0000misinformation games, that provides the necessary toolkit to study this\u0000situation. Subsequently, we enhance this framework by developing a\u0000time-discrete procedure (called the Adaptation Procedure) that captures\u0000iterative interactions in the above context. During the Adaptation Procedure,\u0000the agents update their information and reassess their behaviour in each step.\u0000We demonstrate our ideas through an implementation, which is used to study the\u0000efficiency and characteristics of the Adaptation Procedure.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"42 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Federico Berto, Chuanbo Hua, Laurin Luttmann, Jiwoo Son, Junyoung Park, Kyuree Ahn, Changhyun Kwon, Lin Xie, Jinkyoo Park
Multi-agent combinatorial optimization problems such as routing and scheduling have great practical relevance but present challenges due to their NP-hard combinatorial nature, hard constraints on the number of possible agents, and hard-to-optimize objective functions. This paper introduces PARCO (Parallel AutoRegressive Combinatorial Optimization), a novel approach that learns fast surrogate solvers for multi-agent combinatorial problems with reinforcement learning by employing parallel autoregressive decoding. We propose a model with a Multiple Pointer Mechanism to efficiently decode multiple decisions simultaneously by different agents, enhanced by a Priority-based Conflict Handling scheme. Moreover, we design specialized Communication Layers that enable effective agent collaboration, thus enriching decision-making. We evaluate PARCO in representative multi-agent combinatorial problems in routing and scheduling and demonstrate that our learned solvers offer competitive results against both classical and neural baselines in terms of both solution quality and speed. We make our code openly available at https://github.com/ai4co/parco.
{"title":"PARCO: Learning Parallel Autoregressive Policies for Efficient Multi-Agent Combinatorial Optimization","authors":"Federico Berto, Chuanbo Hua, Laurin Luttmann, Jiwoo Son, Junyoung Park, Kyuree Ahn, Changhyun Kwon, Lin Xie, Jinkyoo Park","doi":"arxiv-2409.03811","DOIUrl":"https://doi.org/arxiv-2409.03811","url":null,"abstract":"Multi-agent combinatorial optimization problems such as routing and\u0000scheduling have great practical relevance but present challenges due to their\u0000NP-hard combinatorial nature, hard constraints on the number of possible\u0000agents, and hard-to-optimize objective functions. This paper introduces PARCO\u0000(Parallel AutoRegressive Combinatorial Optimization), a novel approach that\u0000learns fast surrogate solvers for multi-agent combinatorial problems with\u0000reinforcement learning by employing parallel autoregressive decoding. We\u0000propose a model with a Multiple Pointer Mechanism to efficiently decode\u0000multiple decisions simultaneously by different agents, enhanced by a\u0000Priority-based Conflict Handling scheme. Moreover, we design specialized\u0000Communication Layers that enable effective agent collaboration, thus enriching\u0000decision-making. We evaluate PARCO in representative multi-agent combinatorial\u0000problems in routing and scheduling and demonstrate that our learned solvers\u0000offer competitive results against both classical and neural baselines in terms\u0000of both solution quality and speed. We make our code openly available at\u0000https://github.com/ai4co/parco.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"37 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190505","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jannik Peters, Constantin Waubert de Puiseau, Hasan Tercan, Arya Gopikrishnan, Gustavo Adolpho Lucas De Carvalho, Christian Bitter, Tobias Meisen
The field of emergent language represents a novel area of research within the domain of artificial intelligence, particularly within the context of multi-agent reinforcement learning. Although the concept of studying language emergence is not new, early approaches were primarily concerned with explaining human language formation, with little consideration given to its potential utility for artificial agents. In contrast, studies based on reinforcement learning aim to develop communicative capabilities in agents that are comparable to or even superior to human language. Thus, they extend beyond the learned statistical representations that are common in natural language processing research. This gives rise to a number of fundamental questions, from the prerequisites for language emergence to the criteria for measuring its success. This paper addresses these questions by providing a comprehensive review of 181 scientific publications on emergent language in artificial intelligence. Its objective is to serve as a reference for researchers interested in or proficient in the field. Consequently, the main contributions are the definition and overview of the prevailing terminology, the analysis of existing evaluation methods and metrics, and the description of the identified research gaps.
{"title":"A Survey on Emergent Language","authors":"Jannik Peters, Constantin Waubert de Puiseau, Hasan Tercan, Arya Gopikrishnan, Gustavo Adolpho Lucas De Carvalho, Christian Bitter, Tobias Meisen","doi":"arxiv-2409.02645","DOIUrl":"https://doi.org/arxiv-2409.02645","url":null,"abstract":"The field of emergent language represents a novel area of research within the\u0000domain of artificial intelligence, particularly within the context of\u0000multi-agent reinforcement learning. Although the concept of studying language\u0000emergence is not new, early approaches were primarily concerned with explaining\u0000human language formation, with little consideration given to its potential\u0000utility for artificial agents. In contrast, studies based on reinforcement\u0000learning aim to develop communicative capabilities in agents that are\u0000comparable to or even superior to human language. Thus, they extend beyond the\u0000learned statistical representations that are common in natural language\u0000processing research. This gives rise to a number of fundamental questions, from\u0000the prerequisites for language emergence to the criteria for measuring its\u0000success. This paper addresses these questions by providing a comprehensive\u0000review of 181 scientific publications on emergent language in artificial\u0000intelligence. Its objective is to serve as a reference for researchers\u0000interested in or proficient in the field. Consequently, the main contributions\u0000are the definition and overview of the prevailing terminology, the analysis of\u0000existing evaluation methods and metrics, and the description of the identified\u0000research gaps.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"34 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Multi-agent reinforcement learning (MARL) has exploded in popularity in recent years. Many approaches have been developed but they can be divided into three main types: centralized training and execution (CTE), centralized training for decentralized execution (CTDE), and Decentralized training and execution (DTE). CTDE methods are the most common as they can use centralized information during training but execute in a decentralized manner -- using only information available to that agent during execution. CTDE is the only paradigm that requires a separate training phase where any available information (e.g., other agent policies, underlying states) can be used. As a result, they can be more scalable than CTE methods, do not require communication during execution, and can often perform well. CTDE fits most naturally with the cooperative case, but can be potentially applied in competitive or mixed settings depending on what information is assumed to be observed. This text is an introduction to CTDE in cooperative MARL. It is meant to explain the setting, basic concepts, and common methods. It does not cover all work in CTDE MARL as the subarea is quite extensive. I have included work that I believe is important for understanding the main concepts in the subarea and apologize to those that I have omitted.
{"title":"An Introduction to Centralized Training for Decentralized Execution in Cooperative Multi-Agent Reinforcement Learning","authors":"Christopher Amato","doi":"arxiv-2409.03052","DOIUrl":"https://doi.org/arxiv-2409.03052","url":null,"abstract":"Multi-agent reinforcement learning (MARL) has exploded in popularity in\u0000recent years. Many approaches have been developed but they can be divided into\u0000three main types: centralized training and execution (CTE), centralized\u0000training for decentralized execution (CTDE), and Decentralized training and\u0000execution (DTE). CTDE methods are the most common as they can use centralized information\u0000during training but execute in a decentralized manner -- using only information\u0000available to that agent during execution. CTDE is the only paradigm that\u0000requires a separate training phase where any available information (e.g., other\u0000agent policies, underlying states) can be used. As a result, they can be more\u0000scalable than CTE methods, do not require communication during execution, and\u0000can often perform well. CTDE fits most naturally with the cooperative case, but\u0000can be potentially applied in competitive or mixed settings depending on what\u0000information is assumed to be observed. This text is an introduction to CTDE in cooperative MARL. It is meant to\u0000explain the setting, basic concepts, and common methods. It does not cover all\u0000work in CTDE MARL as the subarea is quite extensive. I have included work that\u0000I believe is important for understanding the main concepts in the subarea and\u0000apologize to those that I have omitted.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Long-distance transport plays a vital role in the economic growth of countries. However, there is a lack of systems being developed for monitoring and support of long-route vehicles (LRV). Sustainable and context-aware transport systems with modern technologies are needed. We model for long-distance vehicle transportation monitoring and support systems in a multi-agent environment. Our model incorporates the distance vehicle transport mechanism through agent-based modeling (ABM). This model constitutes the design protocol of ABM called Overview, Design, and Details (ODD). This model constitutes that every category of agents is offering information as a service. Hence, a federation of services through protocol for the communication between sensors and software components is desired. Such integration of services supports monitoring and tracking of vehicles on the route. The model simulations provide useful results for the integration of services based on smart objects.
{"title":"Context-Aware Agent-based Model for Smart Long Distance Transport System","authors":"Muhammad Raees, Afzal Ahmed","doi":"arxiv-2409.02434","DOIUrl":"https://doi.org/arxiv-2409.02434","url":null,"abstract":"Long-distance transport plays a vital role in the economic growth of\u0000countries. However, there is a lack of systems being developed for monitoring\u0000and support of long-route vehicles (LRV). Sustainable and context-aware\u0000transport systems with modern technologies are needed. We model for\u0000long-distance vehicle transportation monitoring and support systems in a\u0000multi-agent environment. Our model incorporates the distance vehicle transport\u0000mechanism through agent-based modeling (ABM). This model constitutes the design\u0000protocol of ABM called Overview, Design, and Details (ODD). This model\u0000constitutes that every category of agents is offering information as a service.\u0000Hence, a federation of services through protocol for the communication between\u0000sensors and software components is desired. Such integration of services\u0000supports monitoring and tracking of vehicles on the route. The model\u0000simulations provide useful results for the integration of services based on\u0000smart objects.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"42 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190509","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Segev Shlomov, Ben wiesel, Aviad Sela, Ido Levy, Liane Galanti, Roy Abitbol
General web-based agents are increasingly essential for interacting with complex web environments, yet their performance in real-world web applications remains poor, yielding extremely low accuracy even with state-of-the-art frontier models. We observe that these agents can be decomposed into two primary components: Planning and Grounding. Yet, most existing research treats these agents as black boxes, focusing on end-to-end evaluations which hinder meaningful improvements. We sharpen the distinction between the planning and grounding components and conduct a novel analysis by refining experiments on the Mind2Web dataset. Our work proposes a new benchmark for each of the components separately, identifying the bottlenecks and pain points that limit agent performance. Contrary to prevalent assumptions, our findings suggest that grounding is not a significant bottleneck and can be effectively addressed with current techniques. Instead, the primary challenge lies in the planning component, which is the main source of performance degradation. Through this analysis, we offer new insights and demonstrate practical suggestions for improving the capabilities of web agents, paving the way for more reliable agents.
{"title":"From Grounding to Planning: Benchmarking Bottlenecks in Web Agents","authors":"Segev Shlomov, Ben wiesel, Aviad Sela, Ido Levy, Liane Galanti, Roy Abitbol","doi":"arxiv-2409.01927","DOIUrl":"https://doi.org/arxiv-2409.01927","url":null,"abstract":"General web-based agents are increasingly essential for interacting with\u0000complex web environments, yet their performance in real-world web applications\u0000remains poor, yielding extremely low accuracy even with state-of-the-art\u0000frontier models. We observe that these agents can be decomposed into two\u0000primary components: Planning and Grounding. Yet, most existing research treats\u0000these agents as black boxes, focusing on end-to-end evaluations which hinder\u0000meaningful improvements. We sharpen the distinction between the planning and\u0000grounding components and conduct a novel analysis by refining experiments on\u0000the Mind2Web dataset. Our work proposes a new benchmark for each of the\u0000components separately, identifying the bottlenecks and pain points that limit\u0000agent performance. Contrary to prevalent assumptions, our findings suggest that\u0000grounding is not a significant bottleneck and can be effectively addressed with\u0000current techniques. Instead, the primary challenge lies in the planning\u0000component, which is the main source of performance degradation. Through this\u0000analysis, we offer new insights and demonstrate practical suggestions for\u0000improving the capabilities of web agents, paving the way for more reliable\u0000agents.","PeriodicalId":501315,"journal":{"name":"arXiv - CS - Multiagent Systems","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142190511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：