To solve the extended fuzzy description logic with qualifying number restriction （EFALCQ） reasoning problems, EFALCQ is discretely simulated by description logic with qualifying number restriction （ALCQ）, and ALCQ...To solve the extended fuzzy description logic with qualifying number restriction （EFALCQ） reasoning problems, EFALCQ is discretely simulated by description logic with qualifying number restriction （ALCQ）, and ALCQ reasoning results are reused to prove the complexity of EFALCQ reasoning problems. The ALCQ simulation method for the consistency of EFALCQ is proposed. This method reduces EFALCQ satisfiability into EFALCQ consistency, and uses EFALCQ satisfiability to discretely simulate EFALCQ satdomain. It is proved that the reasoning complexity for EFALCQ satisfiability, consistency and sat-domain is PSPACE-complete.展开更多
The problem of fault reasoning has aroused great concern in scientific and engineering fields.However,fault investigation and reasoning of complex system is not a simple reasoning decision-making problem.It has become...The problem of fault reasoning has aroused great concern in scientific and engineering fields.However,fault investigation and reasoning of complex system is not a simple reasoning decision-making problem.It has become a typical multi-constraint and multi-objective reticulate optimization decision-making problem under many influencing factors and constraints.So far,little research has been carried out in this field.This paper transforms the fault reasoning problem of complex system into a paths-searching problem starting from known symptoms to fault causes.Three optimization objectives are considered simultaneously: maximum probability of average fault,maximum average importance,and minimum average complexity of test.Under the constraints of both known symptoms and the causal relationship among different components,a multi-objective optimization mathematical model is set up,taking minimizing cost of fault reasoning as the target function.Since the problem is non-deterministic polynomial-hard(NP-hard),a modified multi-objective ant colony algorithm is proposed,in which a reachability matrix is set up to constrain the feasible search nodes of the ants and a new pseudo-random-proportional rule and a pheromone adjustment mechinism are constructed to balance conflicts between the optimization objectives.At last,a Pareto optimal set is acquired.Evaluation functions based on validity and tendency of reasoning paths are defined to optimize noninferior set,through which the final fault causes can be identified according to decision-making demands,thus realize fault reasoning of the multi-constraint and multi-objective complex system.Reasoning results demonstrate that the improved multi-objective ant colony optimization(IMACO) can realize reasoning and locating fault positions precisely by solving the multi-objective fault diagnosis model,which provides a new method to solve the problem of multi-constraint and multi-objective fault diagnosis and reasoning of complex system.展开更多
The composition of the modern aerospace system becomes more and more complex.The performance degradation of any device in the system may cause it difficult for the whole system to keep normal working states.Therefore,...The composition of the modern aerospace system becomes more and more complex.The performance degradation of any device in the system may cause it difficult for the whole system to keep normal working states.Therefore,it is essential to evaluate the performance of complex aerospace systems.In this paper,the performance evaluation of complex aerospace systems is regarded as a Multi-Attribute Decision Analysis(MADA)problem.Based on the structure and working principle of the system,a new Evidential Reasoning(ER)based approach with uncertain parameters is proposed to construct a nonlinear optimization model to evaluate the system performance.In the model,the interval form is used to express the uncertainty,such as error in testing data and inaccuracy in expert knowledge.In order to analyze the subsystems that have a great impact on the performance of the system,the sensitivity analysis of the evaluation result is carried out,and the corresponding maintenance strategy is proposed.For a type of Inertial Measurement Unit(IMU)used in a rocket,the proposed method is employed to evaluate its performance.Then,the parameter sensitivity of the evaluation result is analyzed,and the main factors affecting the performance of IMU are obtained.Finally,the comparative study shows the effectiveness of the proposed method.展开更多
The evidential reasoning(ER)rule framework has been widely applied in multi-attribute decision analysis and system assessment to manage uncertainty.However,traditional ER implementations rely on two critical limitatio...The evidential reasoning(ER)rule framework has been widely applied in multi-attribute decision analysis and system assessment to manage uncertainty.However,traditional ER implementations rely on two critical limitations:1)unrealistic assumptions of complete evidence independence,and 2)a lack of mechanisms to differentiate causal relationships from spurious correlations.Existing similarity-based approaches often misinterpret interdependent evidence,leading to unreliable decision outcomes.To address these gaps,this study proposes a causality-enhanced ER rule(CER-e)framework with three key methodological innovations:1)a multidimensional causal representation of evidence to capture dependency structures;2)probabilistic quantification of causal strength using transfer entropy,a model-free information-theoretic measure;3)systematic integration of causal parameters into the ER inference process while maintaining evidential objectivity.The PC algorithm is employed during causal discovery to eliminate spurious correlations,ensuring robust causal inference.Case studies in two types of domains—telecommunications network security assessment and structural risk evaluation—validate CER-e’s effectiveness in real-world scenarios.Under simulated incomplete information conditions,the framework demonstrates superior algorithmic robustness compared to traditional ER.Comparative analyses show that CER-e significantly improves both the interpretability of causal relationships and the reliability of assessment results,establishing a novel paradigm for integrating causal inference with evidential reasoning in complex system evaluation.展开更多
The property of NP_completeness of topologic spatial reasoning problem has been proved.According to the similarity of uncertainty with topologic spatial reasoning,the problem of directional spatial reasoning should be...The property of NP_completeness of topologic spatial reasoning problem has been proved.According to the similarity of uncertainty with topologic spatial reasoning,the problem of directional spatial reasoning should be also an NP_complete problem.The proof for the property of NP_completeness in directional spatial reasoning problem is based on two important transformations.After these transformations,a spatial configuration has been constructed based on directional constraints,and the property of NP_completeness in directional spatial reasoning has been proved with the help of the consistency of the constraints in the configuration.展开更多
Directly calculating the topolo gi cal and geometric complexity from the STEP (standard for the exchange of product model data, ISO 10303) file is a huge task. So, a case-based reasoning approac h is presented, which...Directly calculating the topolo gi cal and geometric complexity from the STEP (standard for the exchange of product model data, ISO 10303) file is a huge task. So, a case-based reasoning approac h is presented, which is based on the similarity between the new component and t he old one, to calculate the topological and geometric complexity of new compone nts. In order to index, retrieve in historical component database, a new way of component representation is brought forth. And then an algorithm is given to ext ract topological graph from its STEP files. A mathematical model, which describe s how to compare the similarity, is discussed. Finally, an example is given to s how the result.展开更多
The(3+1)-dimensional Boiti-Leon-Manna-Pempinelli(BLMP)equation serves as a crucial nonlinear evolution equation in mathematical physics,capable of characterizing complex nonlinear dynamic phenomena in three-dimensiona...The(3+1)-dimensional Boiti-Leon-Manna-Pempinelli(BLMP)equation serves as a crucial nonlinear evolution equation in mathematical physics,capable of characterizing complex nonlinear dynamic phenomena in three-dimensional space and one-dimensional time.With broad applications spanning fluid dynamics,shallow water waves,plasma physics,and condensed matter physics,the investigation of its solutions holds significant importance.Traditional analytical methods face limitations due to their dependence on bilinear forms.To overcome this constraint,this letter proposes a novel multi-modal neurosymbolic reasoning intelligent algorithm(MMNRIA)that achieves 100%accurate solutions for nonlinear partial differential equations without requiring bilinear transformations.By synergistically integrating neural networks with symbolic computation,this approach establishes a new paradigm for universal analytical solutions of nonlinear partial differential equations.As a practical demonstration,we successfully derive several exact analytical solutions for the(3+1)-dimensional BLMP equation using MMNRIA.These solutions provide a powerful theoretical framework for studying intricate wave phenomena governed by nonlinearity and dispersion effects in three-dimensional physical space.展开更多
In the complex orchard environment,precise picking point localization is crucial for the automation of fruit picking robots.However,existing methods are prone to positioning errors when dealing with complex scenarios ...In the complex orchard environment,precise picking point localization is crucial for the automation of fruit picking robots.However,existing methods are prone to positioning errors when dealing with complex scenarios such as short peduncles,partial occlusion,or complete misidentification,which can affect the actual work efficiency of the fruit picking robot.This study proposes an enhanced picking point localization method based on semantic reasoning for complex picking scenarios in vineyard.It innovatively designs three modules:the semantic reasoning module(SRM),the ROI threshold adjustment strategy(RTAS),and the picking point location optimization module(PPOM).The SRM is applied to handle the scenarios of grape peduncles being obstructed by obstacles,partial misidentification of peduncles,and complete misidentification of peduncles.The RTAS addresses the issue of low and short peduncles during the picking process.Finally,the PPOM optimizes the final position of the picking point,allowing the robotic arm to perform the picking operation with greater flexibility.Experimental results show that SegFormer achieves an mIoU(mean Intersection over Union)of 84.54%,with B_IoU and P_IoU reaching 73.90%and 75.63%,respectively.Additionally,the success rate of the improved fruit picking point localization algorithm reached 94.96%,surpassing the baseline algorithm by 8.12%.The algorithm's average processing time is 0.5428±0.0063 s,meeting the practical requirements for real-time picking.展开更多
The present study proposed the idea of segment manipulation of complex cognition (SMCC), and technically made it possible the quantitative treatment and systematical manipula-tion on the premise diversity. The segment...The present study proposed the idea of segment manipulation of complex cognition (SMCC), and technically made it possible the quantitative treatment and systematical manipula-tion on the premise diversity. The segment manipulation of complex cognition divides the previ-ous inductive strengths judgment task into three distinct steps, attempting to particularly distin-guish the psychological processes and their rules. The results in Experiment 1 showed that compared with the traditional method, the quantitative treatment and systematical manipulation of SMCC on the diversity did not change the task’s nature, and remain rational and a good measurement of inductive strength judgment. The results in Experiment 2 showed that the par-ticipants’ response rules in the triple-step task were expected from our proposal, and that in Step 2 the “feeling of surprise” (FOS), which seems implausible but predicted from the diversity premises, was measured, and its component might be the critical part that produced the diversity effect. The “feeling of surprise” may reflect the impact of emotion on cognition, representing a strong revision to premise probability principle of pure rational hypothesis proposed by Lo et al., and its roles in the diversity effect are worthy of further research. In this regards were discussed the mistakes that the premise probability principle makes when it takes posterity probability as prior probability.展开更多
The reasoning chain generated by the large language models(LLMs)during the reasoning process is often susceptible to illusions that lead to incorrect reasoning steps.Such misleading intermediate reasoning steps may tr...The reasoning chain generated by the large language models(LLMs)during the reasoning process is often susceptible to illusions that lead to incorrect reasoning steps.Such misleading intermediate reasoning steps may trigger a series of errors.This phenomenon can be alleviated by using validation methods to obtain feedback and adjust the reasoning process,similar to the human reflective process.In this paper,we propose a collaborative reasoning framework for mathematical reasoning called CRMR,where a generator is responsible for generating structured intermediate reasoning and a verifier provides detailed feedback on each step of the reason-ing.In particular,we formulate a rigorous form of structured intermediate reasoning called step-by-step rationale(SSR).We evaluated the CRMR framework not only on mathematical word problems but also conducted experiments using open-source and closed-source models with different parameter sizes independently.The results show that our method fully exploits the inference capabilities of the models and achieves significant results on the dataset compared to a single model.展开更多
Visual Question Answering(VQA)is a complex task that requires a deep understanding of both visual content and natural language questions.The challenge lies in enabling models to recognize and interpret visual elements...Visual Question Answering(VQA)is a complex task that requires a deep understanding of both visual content and natural language questions.The challenge lies in enabling models to recognize and interpret visual elements and to reason through questions in a multi-step,compositional manner.We propose a novel Transformer-based model that introduces specialized tokenization techniques to effectively capture intricate relationships between visual and textual features.The model employs an enhanced self-attention mechanism,enabling it to attend to multiple modalities simultaneously,while a co-attention unit dynamically guides focus to the most relevant image regions and question components.Additionally,a multi-step reasoning module supports iterative inference,allowing the model to excel at complex reasoning tasks.Extensive experiments on benchmark datasets demonstrate the model’s superior performance,with accuracies of 98.6%on CLEVR,63.78%on GQA,and 68.67%on VQA v2.0.Ablation studies confirm the critical contribution of key components,such as the reasoning module and co-attention mechanism,to the model’s effectiveness.Qualitative analysis of the learned attention distributions further illustrates the model’s dynamic reasoning process,adapting to task complexity.Overall,our study advances the adaptation of Transformer architectures for VQA,enhancing both reasoning capabilities and model interpretability in visual reasoning tasks.展开更多
针对大语言模型(large language models,LLMs),虽然现有方法在复杂多步推理任务中(如思维链(chain of thought)通过引导模型生成推理步骤来增强推理能力,但常出现生成的中间步骤错误和信息遗漏问题,一旦某环节出错,往往导致最终解答失...针对大语言模型(large language models,LLMs),虽然现有方法在复杂多步推理任务中(如思维链(chain of thought)通过引导模型生成推理步骤来增强推理能力,但常出现生成的中间步骤错误和信息遗漏问题,一旦某环节出错,往往导致最终解答失败。为此,提出了一种全新的推理方法——递进一致性推理(progressive consistent reasoning,PCR)。PCR通过构建一个动态已知量库(一个在推理过程中不断更新的结构化信息列表),从原始问题中提取显式关键信息建立初始已知量库,并将问题分解为多个子问题;在每一次解答子问题后,通过从子问题答案中提取新信息对已知量库进行动态更新,然后基于最新的已知量库对“原始问题”重新思考后进行一次完整的解答尝试,生成阶段候选解。最后,采用聚合策略整合各阶段候选解,输出更加稳健、准确的最终答案。与其他方法相比,PCR方法在GSM8K、CSQA等多项复杂推理基准上相比传统思维链和自一致性方法(self-consistency reasoning)提高了11.9%、5.73%和3.45%、0.95%。结果表明,PCR方法能够有效降低中间步骤错误和信息遗漏对结果的影响,增强推理的稳定性和准确性。展开更多
This research addresses the performance challenges of ontology-based context-aware and activity recognition techniques in complex environments and abnormal activities,and proposes an optimized ontology framework to im...This research addresses the performance challenges of ontology-based context-aware and activity recognition techniques in complex environments and abnormal activities,and proposes an optimized ontology framework to improve recognition accuracy and computational efficiency.The method in this paper adopts the event sequence segmentation technique,combines location awareness with time interval reasoning,and improves human activity recognition through ontology reasoning.Compared with the existing methods,the framework performs better when dealing with uncertain data and complex scenes,and the experimental results show that its recognition accuracy is improved by 15.6%and processing time is reduced by 22.4%.In addition,it is found that with the increase of context complexity,the traditional ontology inferencemodel has limitations in abnormal behavior recognition,especially in the case of high data redundancy,which tends to lead to a decrease in recognition accuracy.This study effectively mitigates this problem by optimizing the ontology matching algorithm and combining parallel computing and deep learning techniques to enhance the activity recognition capability in complex environments.展开更多
基金The National Natural Science Foundation of China(No60403016)the Weaponry Equipment Foundation of PLA Equip-ment Ministry (No51406020105JB8103)
文摘To solve the extended fuzzy description logic with qualifying number restriction （EFALCQ） reasoning problems, EFALCQ is discretely simulated by description logic with qualifying number restriction （ALCQ）, and ALCQ reasoning results are reused to prove the complexity of EFALCQ reasoning problems. The ALCQ simulation method for the consistency of EFALCQ is proposed. This method reduces EFALCQ satisfiability into EFALCQ consistency, and uses EFALCQ satisfiability to discretely simulate EFALCQ satdomain. It is proved that the reasoning complexity for EFALCQ satisfiability, consistency and sat-domain is PSPACE-complete.
基金supported by Sub-project of Key National Science and Technology Special Project of China(Grant No.2011ZX05056)
文摘The problem of fault reasoning has aroused great concern in scientific and engineering fields.However,fault investigation and reasoning of complex system is not a simple reasoning decision-making problem.It has become a typical multi-constraint and multi-objective reticulate optimization decision-making problem under many influencing factors and constraints.So far,little research has been carried out in this field.This paper transforms the fault reasoning problem of complex system into a paths-searching problem starting from known symptoms to fault causes.Three optimization objectives are considered simultaneously: maximum probability of average fault,maximum average importance,and minimum average complexity of test.Under the constraints of both known symptoms and the causal relationship among different components,a multi-objective optimization mathematical model is set up,taking minimizing cost of fault reasoning as the target function.Since the problem is non-deterministic polynomial-hard(NP-hard),a modified multi-objective ant colony algorithm is proposed,in which a reachability matrix is set up to constrain the feasible search nodes of the ants and a new pseudo-random-proportional rule and a pheromone adjustment mechinism are constructed to balance conflicts between the optimization objectives.At last,a Pareto optimal set is acquired.Evaluation functions based on validity and tendency of reasoning paths are defined to optimize noninferior set,through which the final fault causes can be identified according to decision-making demands,thus realize fault reasoning of the multi-constraint and multi-objective complex system.Reasoning results demonstrate that the improved multi-objective ant colony optimization(IMACO) can realize reasoning and locating fault positions precisely by solving the multi-objective fault diagnosis model,which provides a new method to solve the problem of multi-constraint and multi-objective fault diagnosis and reasoning of complex system.
基金supported by the National Natural Science Foundation of China(Nos.61773388,61751304,61833016,and 61702142)the Shaanxi Outstanding Youth Science Foundation(No.2020JC-34)the Key Research and Development Plan of Hainan(No.ZDYF2019007)。
文摘The composition of the modern aerospace system becomes more and more complex.The performance degradation of any device in the system may cause it difficult for the whole system to keep normal working states.Therefore,it is essential to evaluate the performance of complex aerospace systems.In this paper,the performance evaluation of complex aerospace systems is regarded as a Multi-Attribute Decision Analysis(MADA)problem.Based on the structure and working principle of the system,a new Evidential Reasoning(ER)based approach with uncertain parameters is proposed to construct a nonlinear optimization model to evaluate the system performance.In the model,the interval form is used to express the uncertainty,such as error in testing data and inaccuracy in expert knowledge.In order to analyze the subsystems that have a great impact on the performance of the system,the sensitivity analysis of the evaluation result is carried out,and the corresponding maintenance strategy is proposed.For a type of Inertial Measurement Unit(IMU)used in a rocket,the proposed method is employed to evaluate its performance.Then,the parameter sensitivity of the evaluation result is analyzed,and the main factors affecting the performance of IMU are obtained.Finally,the comparative study shows the effectiveness of the proposed method.
基金supported by the Natural Science Foundation of China(Nos.U22A2099,62273113,62203461,62203365)the Innovation Project of Guangxi Graduate Education under Grant YCBZ2023130by the Guangxi Higher Education Undergraduate Teaching Reform Project Key Project,grant number 2022JGZ130.
文摘The evidential reasoning(ER)rule framework has been widely applied in multi-attribute decision analysis and system assessment to manage uncertainty.However,traditional ER implementations rely on two critical limitations:1)unrealistic assumptions of complete evidence independence,and 2)a lack of mechanisms to differentiate causal relationships from spurious correlations.Existing similarity-based approaches often misinterpret interdependent evidence,leading to unreliable decision outcomes.To address these gaps,this study proposes a causality-enhanced ER rule(CER-e)framework with three key methodological innovations:1)a multidimensional causal representation of evidence to capture dependency structures;2)probabilistic quantification of causal strength using transfer entropy,a model-free information-theoretic measure;3)systematic integration of causal parameters into the ER inference process while maintaining evidential objectivity.The PC algorithm is employed during causal discovery to eliminate spurious correlations,ensuring robust causal inference.Case studies in two types of domains—telecommunications network security assessment and structural risk evaluation—validate CER-e’s effectiveness in real-world scenarios.Under simulated incomplete information conditions,the framework demonstrates superior algorithmic robustness compared to traditional ER.Comparative analyses show that CER-e significantly improves both the interpretability of causal relationships and the reliability of assessment results,establishing a novel paradigm for integrating causal inference with evidential reasoning in complex system evaluation.
文摘The property of NP_completeness of topologic spatial reasoning problem has been proved.According to the similarity of uncertainty with topologic spatial reasoning,the problem of directional spatial reasoning should be also an NP_complete problem.The proof for the property of NP_completeness in directional spatial reasoning problem is based on two important transformations.After these transformations,a spatial configuration has been constructed based on directional constraints,and the property of NP_completeness in directional spatial reasoning has been proved with the help of the consistency of the constraints in the configuration.
文摘Directly calculating the topolo gi cal and geometric complexity from the STEP (standard for the exchange of product model data, ISO 10303) file is a huge task. So, a case-based reasoning approac h is presented, which is based on the similarity between the new component and t he old one, to calculate the topological and geometric complexity of new compone nts. In order to index, retrieve in historical component database, a new way of component representation is brought forth. And then an algorithm is given to ext ract topological graph from its STEP files. A mathematical model, which describe s how to compare the similarity, is discussed. Finally, an example is given to s how the result.
基金supported by the National Natural Science Foundation of China(Grant No.62303289)Tianyuan Fund for Mathematics of the National Natural Science Foundation of China(Grant No.12426105)+3 种基金the Scientific and Technological Innovation Programs(STIP)of Higher Education Institutions in Shanxi(Grant No.2024L022)Fundamental Research Program of Shanxi Province(Grant Nos.202403021222001 and 202203021222003)the“Wen Ying Young Scholars”Talent Project of Shanxi University(Grant Nos.138541088,138541090,and 138541127)Funded by Open Foundation of Hubei Key Laboratory of Applied Mathematics(Hubei University)(Grant No.HBAM202401).
文摘The(3+1)-dimensional Boiti-Leon-Manna-Pempinelli(BLMP)equation serves as a crucial nonlinear evolution equation in mathematical physics,capable of characterizing complex nonlinear dynamic phenomena in three-dimensional space and one-dimensional time.With broad applications spanning fluid dynamics,shallow water waves,plasma physics,and condensed matter physics,the investigation of its solutions holds significant importance.Traditional analytical methods face limitations due to their dependence on bilinear forms.To overcome this constraint,this letter proposes a novel multi-modal neurosymbolic reasoning intelligent algorithm(MMNRIA)that achieves 100%accurate solutions for nonlinear partial differential equations without requiring bilinear transformations.By synergistically integrating neural networks with symbolic computation,this approach establishes a new paradigm for universal analytical solutions of nonlinear partial differential equations.As a practical demonstration,we successfully derive several exact analytical solutions for the(3+1)-dimensional BLMP equation using MMNRIA.These solutions provide a powerful theoretical framework for studying intricate wave phenomena governed by nonlinearity and dispersion effects in three-dimensional physical space.
基金supported by the National Natural Science Foundation of China,China(32171909,52205254,32301704)the Guangdong Basic and Applied Basic Research Foundation,China(2023A1515011255,2020B1515120050,2024A1515010199)+1 种基金the Key Research Projects of Ordinary Universities in Guangdong Province,China(2024ZDZX1042,2024ZDZX3057)the Guangdong key R&D projects,China(202080404030001).
文摘In the complex orchard environment,precise picking point localization is crucial for the automation of fruit picking robots.However,existing methods are prone to positioning errors when dealing with complex scenarios such as short peduncles,partial occlusion,or complete misidentification,which can affect the actual work efficiency of the fruit picking robot.This study proposes an enhanced picking point localization method based on semantic reasoning for complex picking scenarios in vineyard.It innovatively designs three modules:the semantic reasoning module(SRM),the ROI threshold adjustment strategy(RTAS),and the picking point location optimization module(PPOM).The SRM is applied to handle the scenarios of grape peduncles being obstructed by obstacles,partial misidentification of peduncles,and complete misidentification of peduncles.The RTAS addresses the issue of low and short peduncles during the picking process.Finally,the PPOM optimizes the final position of the picking point,allowing the robotic arm to perform the picking operation with greater flexibility.Experimental results show that SegFormer achieves an mIoU(mean Intersection over Union)of 84.54%,with B_IoU and P_IoU reaching 73.90%and 75.63%,respectively.Additionally,the success rate of the improved fruit picking point localization algorithm reached 94.96%,surpassing the baseline algorithm by 8.12%.The algorithm's average processing time is 0.5428±0.0063 s,meeting the practical requirements for real-time picking.
基金This work was supported by the National Natural Science Foundation of China(Grant No.30370488)National Key Subject of Basic Psychology,SWU(Grant No.XGZ04006).
文摘The present study proposed the idea of segment manipulation of complex cognition (SMCC), and technically made it possible the quantitative treatment and systematical manipula-tion on the premise diversity. The segment manipulation of complex cognition divides the previ-ous inductive strengths judgment task into three distinct steps, attempting to particularly distin-guish the psychological processes and their rules. The results in Experiment 1 showed that compared with the traditional method, the quantitative treatment and systematical manipulation of SMCC on the diversity did not change the task’s nature, and remain rational and a good measurement of inductive strength judgment. The results in Experiment 2 showed that the par-ticipants’ response rules in the triple-step task were expected from our proposal, and that in Step 2 the “feeling of surprise” (FOS), which seems implausible but predicted from the diversity premises, was measured, and its component might be the critical part that produced the diversity effect. The “feeling of surprise” may reflect the impact of emotion on cognition, representing a strong revision to premise probability principle of pure rational hypothesis proposed by Lo et al., and its roles in the diversity effect are worthy of further research. In this regards were discussed the mistakes that the premise probability principle makes when it takes posterity probability as prior probability.
基金supported by the National Natural Science Foundation of China(No.62176052)Natural Science Foundation of Shanghai,China(No.21ZR1401700)AI-Enhanced Research Program of Shanghai Municipal Education Commission(No.SMEC-AI-DHUZ-05).
文摘The reasoning chain generated by the large language models(LLMs)during the reasoning process is often susceptible to illusions that lead to incorrect reasoning steps.Such misleading intermediate reasoning steps may trigger a series of errors.This phenomenon can be alleviated by using validation methods to obtain feedback and adjust the reasoning process,similar to the human reflective process.In this paper,we propose a collaborative reasoning framework for mathematical reasoning called CRMR,where a generator is responsible for generating structured intermediate reasoning and a verifier provides detailed feedback on each step of the reason-ing.In particular,we formulate a rigorous form of structured intermediate reasoning called step-by-step rationale(SSR).We evaluated the CRMR framework not only on mathematical word problems but also conducted experiments using open-source and closed-source models with different parameter sizes independently.The results show that our method fully exploits the inference capabilities of the models and achieves significant results on the dataset compared to a single model.
文摘Visual Question Answering(VQA)is a complex task that requires a deep understanding of both visual content and natural language questions.The challenge lies in enabling models to recognize and interpret visual elements and to reason through questions in a multi-step,compositional manner.We propose a novel Transformer-based model that introduces specialized tokenization techniques to effectively capture intricate relationships between visual and textual features.The model employs an enhanced self-attention mechanism,enabling it to attend to multiple modalities simultaneously,while a co-attention unit dynamically guides focus to the most relevant image regions and question components.Additionally,a multi-step reasoning module supports iterative inference,allowing the model to excel at complex reasoning tasks.Extensive experiments on benchmark datasets demonstrate the model’s superior performance,with accuracies of 98.6%on CLEVR,63.78%on GQA,and 68.67%on VQA v2.0.Ablation studies confirm the critical contribution of key components,such as the reasoning module and co-attention mechanism,to the model’s effectiveness.Qualitative analysis of the learned attention distributions further illustrates the model’s dynamic reasoning process,adapting to task complexity.Overall,our study advances the adaptation of Transformer architectures for VQA,enhancing both reasoning capabilities and model interpretability in visual reasoning tasks.
文摘针对大语言模型(large language models,LLMs),虽然现有方法在复杂多步推理任务中(如思维链(chain of thought)通过引导模型生成推理步骤来增强推理能力,但常出现生成的中间步骤错误和信息遗漏问题,一旦某环节出错,往往导致最终解答失败。为此,提出了一种全新的推理方法——递进一致性推理(progressive consistent reasoning,PCR)。PCR通过构建一个动态已知量库(一个在推理过程中不断更新的结构化信息列表),从原始问题中提取显式关键信息建立初始已知量库,并将问题分解为多个子问题;在每一次解答子问题后,通过从子问题答案中提取新信息对已知量库进行动态更新,然后基于最新的已知量库对“原始问题”重新思考后进行一次完整的解答尝试,生成阶段候选解。最后,采用聚合策略整合各阶段候选解,输出更加稳健、准确的最终答案。与其他方法相比,PCR方法在GSM8K、CSQA等多项复杂推理基准上相比传统思维链和自一致性方法(self-consistency reasoning)提高了11.9%、5.73%和3.45%、0.95%。结果表明,PCR方法能够有效降低中间步骤错误和信息遗漏对结果的影响,增强推理的稳定性和准确性。
基金supported by the BK21 FOUR program of the National Research Foundation of Korea funded by the Ministry of Education(NRF5199991014091)Seok-Won Lee’s work was supported by Institute of Information&Communications Technology Planning&Evaluation(IITP)under the Artificial Intelligence Convergence Innovation Human Resources Development(IITP-2024-RS-2023-00255968)grant funded by the Korea government(MSIT).
文摘This research addresses the performance challenges of ontology-based context-aware and activity recognition techniques in complex environments and abnormal activities,and proposes an optimized ontology framework to improve recognition accuracy and computational efficiency.The method in this paper adopts the event sequence segmentation technique,combines location awareness with time interval reasoning,and improves human activity recognition through ontology reasoning.Compared with the existing methods,the framework performs better when dealing with uncertain data and complex scenes,and the experimental results show that its recognition accuracy is improved by 15.6%and processing time is reduced by 22.4%.In addition,it is found that with the increase of context complexity,the traditional ontology inferencemodel has limitations in abnormal behavior recognition,especially in the case of high data redundancy,which tends to lead to a decrease in recognition accuracy.This study effectively mitigates this problem by optimizing the ontology matching algorithm and combining parallel computing and deep learning techniques to enhance the activity recognition capability in complex environments.
