Large language models (LLMs) have been utilized in solving diverse reasoning tasks, encompassing common sense, arithmetic and deduction tasks. However, with difficulties of reversing thinking patterns and irrelevant premises, how to determine the authenticity of the cause in abductive logical reasoning remains underexplored. Inspired by hypothesis and verification method and identification of irrelevant information in human thinking process, we propose a new framework for LLMs abductive logical reasoning called CauseJudger (CJ), which identifies the authenticity of possible cause by transforming thinking from reverse to forward and removing irrelevant information. In addition, we construct an abductive logical reasoning dataset for decision task called CauseLogics, which contains 200,000 tasks of varying reasoning lengths. Our experiments show the efficiency of CJ with overall experiments and ablation experiments as well as case studies on our dataset and reconstructed public dataset. Notably, CJ's implementation is efficient, requiring only two calls to LLM. Its impact is profound: when using gpt-3.5, CJ achieves a maximum correctness improvement of 41% compared to Zero-Shot-CoT. Moreover, with gpt-4, CJ attains an accuracy exceeding 90% across all datasets.
{"title":"CauseJudger: Identifying the Cause with LLMs for Abductive Logical Reasoning","authors":"Jinwei He, Feng Lu","doi":"arxiv-2409.05559","DOIUrl":"https://doi.org/arxiv-2409.05559","url":null,"abstract":"Large language models (LLMs) have been utilized in solving diverse reasoning\u0000tasks, encompassing common sense, arithmetic and deduction tasks. However, with\u0000difficulties of reversing thinking patterns and irrelevant premises, how to\u0000determine the authenticity of the cause in abductive logical reasoning remains\u0000underexplored. Inspired by hypothesis and verification method and\u0000identification of irrelevant information in human thinking process, we propose\u0000a new framework for LLMs abductive logical reasoning called CauseJudger (CJ),\u0000which identifies the authenticity of possible cause by transforming thinking\u0000from reverse to forward and removing irrelevant information. In addition, we\u0000construct an abductive logical reasoning dataset for decision task called\u0000CauseLogics, which contains 200,000 tasks of varying reasoning lengths. Our\u0000experiments show the efficiency of CJ with overall experiments and ablation\u0000experiments as well as case studies on our dataset and reconstructed public\u0000dataset. Notably, CJ's implementation is efficient, requiring only two calls to\u0000LLM. Its impact is profound: when using gpt-3.5, CJ achieves a maximum\u0000correctness improvement of 41% compared to Zero-Shot-CoT. Moreover, with gpt-4,\u0000CJ attains an accuracy exceeding 90% across all datasets.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"62 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142224948","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}
In pursuit of a more sustainable and cost-efficient last mile, parcel lockers have gained a firm foothold in the parcel delivery landscape. To fully exploit their potential and simultaneously ensure customer satisfaction, successful management of the locker's limited capacity is crucial. This is challenging as future delivery requests and pickup times are stochastic from the provider's perspective. In response, we propose to dynamically control whether the locker is presented as an available delivery option to each incoming customer with the goal of maximizing the number of served requests weighted by their priority. Additionally, we take different compartment sizes into account, which entails a second type of decision as parcels scheduled for delivery must be allocated. We formalize the problem as an infinite-horizon sequential decision problem and find that exact methods are intractable due to the curses of dimensionality. In light of this, we develop a solution framework that orchestrates multiple algorithmic techniques rooted in Sequential Decision Analytics and Reinforcement Learning, namely cost function approximation and an offline trained parametric value function approximation together with a truncated online rollout. Our innovative approach to combine these techniques enables us to address the strong interrelations between the two decision types. As a general methodological contribution, we enhance the training of our value function approximation with a modified version of experience replay that enforces structure in the value function. Our computational study shows that our method outperforms a myopic benchmark by 13.7% and an industry-inspired policy by 12.6%.
{"title":"Dynamic Demand Management for Parcel Lockers","authors":"Daniela Sailer, Robert Klein, Claudius Steinhardt","doi":"arxiv-2409.05061","DOIUrl":"https://doi.org/arxiv-2409.05061","url":null,"abstract":"In pursuit of a more sustainable and cost-efficient last mile, parcel lockers\u0000have gained a firm foothold in the parcel delivery landscape. To fully exploit\u0000their potential and simultaneously ensure customer satisfaction, successful\u0000management of the locker's limited capacity is crucial. This is challenging as\u0000future delivery requests and pickup times are stochastic from the provider's\u0000perspective. In response, we propose to dynamically control whether the locker\u0000is presented as an available delivery option to each incoming customer with the\u0000goal of maximizing the number of served requests weighted by their priority.\u0000Additionally, we take different compartment sizes into account, which entails a\u0000second type of decision as parcels scheduled for delivery must be allocated. We\u0000formalize the problem as an infinite-horizon sequential decision problem and\u0000find that exact methods are intractable due to the curses of dimensionality. In\u0000light of this, we develop a solution framework that orchestrates multiple\u0000algorithmic techniques rooted in Sequential Decision Analytics and\u0000Reinforcement Learning, namely cost function approximation and an offline\u0000trained parametric value function approximation together with a truncated\u0000online rollout. Our innovative approach to combine these techniques enables us\u0000to address the strong interrelations between the two decision types. As a\u0000general methodological contribution, we enhance the training of our value\u0000function approximation with a modified version of experience replay that\u0000enforces structure in the value function. Our computational study shows that\u0000our method outperforms a myopic benchmark by 13.7% and an industry-inspired\u0000policy by 12.6%.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"36 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142224949","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}
Ruiting Dai, Yuqiao Tan, Lisi Mo, Tao He, Ke Qin, Shuang Liang
Recently, prompt learning has garnered considerable attention for its success in various Vision-Language (VL) tasks. However, existing prompt-based models are primarily focused on studying prompt generation and prompt strategies with complete modality settings, which does not accurately reflect real-world scenarios where partial modality information may be missing. In this paper, we present the first comprehensive investigation into prompt learning behavior when modalities are incomplete, revealing the high sensitivity of prompt-based models to missing modalities. To this end, we propose a novel Multi-step Adaptive Prompt Learning (MuAP) framework, aiming to generate multimodal prompts and perform multi-step prompt tuning, which adaptively learns knowledge by iteratively aligning modalities. Specifically, we generate multimodal prompts for each modality and devise prompt strategies to integrate them into the Transformer model. Subsequently, we sequentially perform prompt tuning from single-stage and alignment-stage, allowing each modality-prompt to be autonomously and adaptively learned, thereby mitigating the imbalance issue caused by only textual prompts that are learnable in previous works. Extensive experiments demonstrate the effectiveness of our MuAP and this model achieves significant improvements compared to the state-of-the-art on all benchmark datasets
{"title":"MuAP: Multi-step Adaptive Prompt Learning for Vision-Language Model with Missing Modality","authors":"Ruiting Dai, Yuqiao Tan, Lisi Mo, Tao He, Ke Qin, Shuang Liang","doi":"arxiv-2409.04693","DOIUrl":"https://doi.org/arxiv-2409.04693","url":null,"abstract":"Recently, prompt learning has garnered considerable attention for its success\u0000in various Vision-Language (VL) tasks. However, existing prompt-based models\u0000are primarily focused on studying prompt generation and prompt strategies with\u0000complete modality settings, which does not accurately reflect real-world\u0000scenarios where partial modality information may be missing. In this paper, we\u0000present the first comprehensive investigation into prompt learning behavior\u0000when modalities are incomplete, revealing the high sensitivity of prompt-based\u0000models to missing modalities. To this end, we propose a novel Multi-step\u0000Adaptive Prompt Learning (MuAP) framework, aiming to generate multimodal\u0000prompts and perform multi-step prompt tuning, which adaptively learns knowledge\u0000by iteratively aligning modalities. Specifically, we generate multimodal\u0000prompts for each modality and devise prompt strategies to integrate them into\u0000the Transformer model. Subsequently, we sequentially perform prompt tuning from\u0000single-stage and alignment-stage, allowing each modality-prompt to be\u0000autonomously and adaptively learned, thereby mitigating the imbalance issue\u0000caused by only textual prompts that are learnable in previous works. Extensive\u0000experiments demonstrate the effectiveness of our MuAP and this model achieves\u0000significant improvements compared to the state-of-the-art on all benchmark\u0000datasets","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"37 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142193945","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}
The increasing complexity of robots and autonomous agents that interact with people highlights the critical need for approaches that systematically test them before deployment. This review paper presents a general framework for solving this problem, describes the insights that we have gained from working on each component of the framework, and shows how integrating these components leads to the discovery of a diverse range of realistic and challenging scenarios that reveal previously unknown failures in deployed robotic systems interacting with people.
{"title":"Algorithmic Scenario Generation as Quality Diversity Optimization","authors":"Stefanos Nikolaidis","doi":"arxiv-2409.04711","DOIUrl":"https://doi.org/arxiv-2409.04711","url":null,"abstract":"The increasing complexity of robots and autonomous agents that interact with\u0000people highlights the critical need for approaches that systematically test\u0000them before deployment. This review paper presents a general framework for\u0000solving this problem, describes the insights that we have gained from working\u0000on each component of the framework, and shows how integrating these components\u0000leads to the discovery of a diverse range of realistic and challenging\u0000scenarios that reveal previously unknown failures in deployed robotic systems\u0000interacting with people.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"39 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142194070","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}
This research presents a computational framework for describing and recognizing episodes and for logical reasoning. This framework, named cognitive-logs, consists of a set of relational and graph databases. Cognitive-logs record knowledge, particularly in episodes that consist of "actions" represented by verbs in natural languages and "participants" who perform the actions. These objects are connected by arrows (morphisms) that link each action to its participant and link cause to effect. Operations based on category theory enable comparisons between episodes and deductive inferences, including abstractions of stories. One of the goals of this study is to develop a database-driven artificial intelligence. This artificial intelligence thinks like a human but possesses the accuracy and rigour of a machine. The vast capacities of databases (up to petabyte scales in current technologies) enable the artificial intelligence to store a greater volume of knowledge than neural-network based artificial intelligences. Cognitive-logs serve as a model of human cognition and designed with references to cognitive linguistics. Cognitive-logs also have the potential to model various human mind activities.
{"title":"Action is the primary key: a categorical framework for episode description and logical reasoning","authors":"Yoshiki Fukada","doi":"arxiv-2409.04793","DOIUrl":"https://doi.org/arxiv-2409.04793","url":null,"abstract":"This research presents a computational framework for describing and\u0000recognizing episodes and for logical reasoning. This framework, named\u0000cognitive-logs, consists of a set of relational and graph databases.\u0000Cognitive-logs record knowledge, particularly in episodes that consist of\u0000\"actions\" represented by verbs in natural languages and \"participants\" who\u0000perform the actions. These objects are connected by arrows (morphisms) that\u0000link each action to its participant and link cause to effect. Operations based\u0000on category theory enable comparisons between episodes and deductive\u0000inferences, including abstractions of stories. One of the goals of this study\u0000is to develop a database-driven artificial intelligence. This artificial\u0000intelligence thinks like a human but possesses the accuracy and rigour of a\u0000machine. The vast capacities of databases (up to petabyte scales in current\u0000technologies) enable the artificial intelligence to store a greater volume of\u0000knowledge than neural-network based artificial intelligences. Cognitive-logs\u0000serve as a model of human cognition and designed with references to cognitive\u0000linguistics. Cognitive-logs also have the potential to model various human mind\u0000activities.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"30 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142194069","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}
As LLMs become increasingly proficient at producing human-like responses, there has been a rise of academic and industrial pursuits dedicated to flagging a given piece of text as "human" or "AI". Most of these pursuits involve modern NLP detectors like T5-Sentinel and RoBERTa-Sentinel, without paying too much attention to issues of interpretability and explainability of these models. In our study, we provide a comprehensive analysis that shows that traditional ML models (Naive-Bayes,MLP, Random Forests, XGBoost) perform as well as modern NLP detectors, in human vs AI text detection. We achieve this by implementing a robust testing procedure on diverse datasets, including curated corpora and real-world samples. Subsequently, by employing the explainable AI technique LIME, we uncover parts of the input that contribute most to the prediction of each model, providing insights into the detection process. Our study contributes to the growing need for developing production-level LLM detection tools, which can leverage a wide range of traditional as well as modern NLP detectors we propose. Finally, the LIME techniques we demonstrate also have the potential to equip these detection tools with interpretability analysis features, making them more reliable and trustworthy in various domains like education, healthcare, and media.
{"title":"HULLMI: Human vs LLM identification with explainability","authors":"Prathamesh Dinesh Joshi, Sahil Pocker, Raj Abhijit Dandekar, Rajat Dandekar, Sreedath Panat","doi":"arxiv-2409.04808","DOIUrl":"https://doi.org/arxiv-2409.04808","url":null,"abstract":"As LLMs become increasingly proficient at producing human-like responses,\u0000there has been a rise of academic and industrial pursuits dedicated to flagging\u0000a given piece of text as \"human\" or \"AI\". Most of these pursuits involve modern\u0000NLP detectors like T5-Sentinel and RoBERTa-Sentinel, without paying too much\u0000attention to issues of interpretability and explainability of these models. In\u0000our study, we provide a comprehensive analysis that shows that traditional ML\u0000models (Naive-Bayes,MLP, Random Forests, XGBoost) perform as well as modern NLP\u0000detectors, in human vs AI text detection. We achieve this by implementing a\u0000robust testing procedure on diverse datasets, including curated corpora and\u0000real-world samples. Subsequently, by employing the explainable AI technique\u0000LIME, we uncover parts of the input that contribute most to the prediction of\u0000each model, providing insights into the detection process. Our study\u0000contributes to the growing need for developing production-level LLM detection\u0000tools, which can leverage a wide range of traditional as well as modern NLP\u0000detectors we propose. Finally, the LIME techniques we demonstrate also have the\u0000potential to equip these detection tools with interpretability analysis\u0000features, making them more reliable and trustworthy in various domains like\u0000education, healthcare, and media.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142194072","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}
Daniel J. Tan, Qianyi Xu, Kay Choong See, Dilruk Perera, Mengling Feng
Multi-organ diseases present significant challenges due to their simultaneous impact on multiple organ systems, necessitating complex and adaptive treatment strategies. Despite recent advancements in AI-powered healthcare decision support systems, existing solutions are limited to individual organ systems. They often ignore the intricate dependencies between organ system and thereby fails to provide holistic treatment recommendations that are useful in practice. We propose a novel hierarchical multi-agent reinforcement learning (HMARL) framework to address these challenges. This framework uses dedicated agents for each organ system, and model dynamic through explicit inter-agent communication channels, enabling coordinated treatment strategies across organs. Furthermore, we introduce a dual-layer state representation technique to contextualize patient conditions at various hierarchical levels, enhancing the treatment accuracy and relevance. Through extensive qualitative and quantitative evaluations in managing sepsis (a complex multi-organ disease), our approach demonstrates its ability to learn effective treatment policies that significantly improve patient survival rates. This framework marks a substantial advancement in clinical decision support systems, pioneering a comprehensive approach for multi-organ treatment recommendations.
{"title":"Advancing Multi-Organ Disease Care: A Hierarchical Multi-Agent Reinforcement Learning Framework","authors":"Daniel J. Tan, Qianyi Xu, Kay Choong See, Dilruk Perera, Mengling Feng","doi":"arxiv-2409.04224","DOIUrl":"https://doi.org/arxiv-2409.04224","url":null,"abstract":"Multi-organ diseases present significant challenges due to their simultaneous\u0000impact on multiple organ systems, necessitating complex and adaptive treatment\u0000strategies. Despite recent advancements in AI-powered healthcare decision\u0000support systems, existing solutions are limited to individual organ systems.\u0000They often ignore the intricate dependencies between organ system and thereby\u0000fails to provide holistic treatment recommendations that are useful in\u0000practice. We propose a novel hierarchical multi-agent reinforcement learning\u0000(HMARL) framework to address these challenges. This framework uses dedicated\u0000agents for each organ system, and model dynamic through explicit inter-agent\u0000communication channels, enabling coordinated treatment strategies across\u0000organs. Furthermore, we introduce a dual-layer state representation technique\u0000to contextualize patient conditions at various hierarchical levels, enhancing\u0000the treatment accuracy and relevance. Through extensive qualitative and\u0000quantitative evaluations in managing sepsis (a complex multi-organ disease),\u0000our approach demonstrates its ability to learn effective treatment policies\u0000that significantly improve patient survival rates. This framework marks a\u0000substantial advancement in clinical decision support systems, pioneering a\u0000comprehensive approach for multi-organ treatment recommendations.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"41 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142193858","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}
Tan D. Tran, Canh V. Pham, Dung T. K. Ha, Phuong N. H. Pham
This work proposes an efficient parallel algorithm for non-monotone submodular maximization under a knapsack constraint problem over the ground set of size $n$. Our algorithm improves the best approximation factor of the existing parallel one from $8+epsilon$ to $7+epsilon$ with $O(log n)$ adaptive complexity. The key idea of our approach is to create a new alternate threshold algorithmic framework. This strategy alternately constructs two disjoint candidate solutions within a constant number of sequence rounds. Then, the algorithm boosts solution quality without sacrificing the adaptive complexity. Extensive experimental studies on three applications, Revenue Maximization, Image Summarization, and Maximum Weighted Cut, show that our algorithm not only significantly increases solution quality but also requires comparative adaptivity to state-of-the-art algorithms.
{"title":"Improved Parallel Algorithm for Non-Monotone Submodular Maximization under Knapsack Constraint","authors":"Tan D. Tran, Canh V. Pham, Dung T. K. Ha, Phuong N. H. Pham","doi":"arxiv-2409.04415","DOIUrl":"https://doi.org/arxiv-2409.04415","url":null,"abstract":"This work proposes an efficient parallel algorithm for non-monotone\u0000submodular maximization under a knapsack constraint problem over the ground set\u0000of size $n$. Our algorithm improves the best approximation factor of the\u0000existing parallel one from $8+epsilon$ to $7+epsilon$ with $O(log n)$\u0000adaptive complexity. The key idea of our approach is to create a new alternate threshold\u0000algorithmic framework. This strategy alternately constructs two disjoint\u0000candidate solutions within a constant number of sequence rounds. Then, the\u0000algorithm boosts solution quality without sacrificing the adaptive complexity.\u0000Extensive experimental studies on three applications, Revenue Maximization,\u0000Image Summarization, and Maximum Weighted Cut, show that our algorithm not only\u0000significantly increases solution quality but also requires comparative\u0000adaptivity to state-of-the-art algorithms.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"22 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142193857","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}
Directed graphical models such as Bayesian nets are often used to implement intelligent tutoring systems able to interact in real-time with learners in a purely automatic way. When coping with such models, keeping a bound on the number of parameters might be important for multiple reasons. First, as these models are typically based on expert knowledge, a huge number of parameters to elicit might discourage practitioners from adopting them. Moreover, the number of model parameters affects the complexity of the inferences, while a fast computation of the queries is needed for real-time feedback. We advocate logical gates with uncertainty for a compact parametrization of the conditional probability tables in the underlying Bayesian net used by tutoring systems. We discuss the semantics of the model parameters to elicit and the assumptions required to apply such approach in this domain. We also derive a dedicated inference scheme to speed up computations.
{"title":"Intelligent tutoring systems by Bayesian networks with noisy gates","authors":"Alessandro Antonucci, Francesca Mangili, Claudio Bonesana, Giorgia Adorni","doi":"arxiv-2409.04102","DOIUrl":"https://doi.org/arxiv-2409.04102","url":null,"abstract":"Directed graphical models such as Bayesian nets are often used to implement\u0000intelligent tutoring systems able to interact in real-time with learners in a\u0000purely automatic way. When coping with such models, keeping a bound on the\u0000number of parameters might be important for multiple reasons. First, as these\u0000models are typically based on expert knowledge, a huge number of parameters to\u0000elicit might discourage practitioners from adopting them. Moreover, the number\u0000of model parameters affects the complexity of the inferences, while a fast\u0000computation of the queries is needed for real-time feedback. We advocate\u0000logical gates with uncertainty for a compact parametrization of the conditional\u0000probability tables in the underlying Bayesian net used by tutoring systems. We\u0000discuss the semantics of the model parameters to elicit and the assumptions\u0000required to apply such approach in this domain. We also derive a dedicated\u0000inference scheme to speed up computations.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"405 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142193859","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}
Mehwish Alam, Genet Asefa Gesese, Pierre-Henri Paris
Knowledge graphs (KGs) have recently been used for many tools and applications, making them rich resources in structured format. However, in the real world, KGs grow due to the additions of new knowledge in the form of entities and relations, making these KGs dynamic. This chapter formally defines several types of dynamic KGs and summarizes how these KGs can be represented. Additionally, many neurosymbolic methods have been proposed for learning representations over static KGs for several tasks such as KG completion and entity alignment. This chapter further focuses on neurosymbolic methods for dynamic KGs with or without temporal information. More specifically, it provides an insight into neurosymbolic methods for dynamic (temporal or non-temporal) KG completion and entity alignment tasks. It further discusses the challenges of current approaches and provides some future directions.
知识图谱（KGs）近来已被用于许多工具和应用中，成为结构化格式的丰富资源。然而，在现实世界中，由于实体和关系形式的新知识的加入，知识图谱会不断增长，从而使这些知识图谱成为动态图谱。本章正式定义了几种类型的动态 KG，并总结了这些 KG 的表示方法。此外，许多神经符号方法已被提出，用于学习静态 KG 的表示方法，以完成 KG 补充和实体对齐等任务。本章将进一步关注用于有时间信息或无时间信息动态 KG 的神经符号方法。更具体地说，本章深入探讨了用于动态（时态或非时态）KG补全和实体配准任务的神经符号方法。它进一步讨论了当前方法所面临的挑战，并提供了一些未来发展方向。
{"title":"Neurosymbolic Methods for Dynamic Knowledge Graphs","authors":"Mehwish Alam, Genet Asefa Gesese, Pierre-Henri Paris","doi":"arxiv-2409.04572","DOIUrl":"https://doi.org/arxiv-2409.04572","url":null,"abstract":"Knowledge graphs (KGs) have recently been used for many tools and\u0000applications, making them rich resources in structured format. However, in the\u0000real world, KGs grow due to the additions of new knowledge in the form of\u0000entities and relations, making these KGs dynamic. This chapter formally defines\u0000several types of dynamic KGs and summarizes how these KGs can be represented.\u0000Additionally, many neurosymbolic methods have been proposed for learning\u0000representations over static KGs for several tasks such as KG completion and\u0000entity alignment. This chapter further focuses on neurosymbolic methods for\u0000dynamic KGs with or without temporal information. More specifically, it\u0000provides an insight into neurosymbolic methods for dynamic (temporal or\u0000non-temporal) KG completion and entity alignment tasks. It further discusses\u0000the challenges of current approaches and provides some future directions.","PeriodicalId":501479,"journal":{"name":"arXiv - CS - Artificial Intelligence","volume":"37 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142193863","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：