一种基于时序关系网络的逻辑推理方法

doi:10.11896/jsjkx.201000171

计算机科学 ›› 2021, Vol. 48 ›› Issue (5): 239-246.doi: 10.11896/jsjkx.201000171

一种基于时序关系网络的逻辑推理方法

张姝楠^1,2, 曹峰^1,2, 郭倩^1,2, 钱宇华^1,2,3

1 山西大学大数据科学与产业研究院太原030006
2 山西大学计算机与信息技术学院太原030006
3 山西大学计算智能与中文信息处理教育部重点实验室太原030006

收稿日期:2020-10-28 修回日期:2021-03-25 出版日期:2021-05-15 发布日期:2021-05-09
通讯作者: 钱宇华(jinchengqyh@126.com)
基金资助:
国家自然科学基金项目(61672332,61802238,61603228,62006146,F060308);山西省拔尖创新人才支持计划;山西省重点研发计划(国际科技合作)项目(201903D421003);山西省三晋学者;山西省回国留学人员科研项目(2017023,2018172,HGKY2019001);山西省青年基金项目(201901D211171,201901D211169)

Logical Reasoning Method Based on Temporal Relation Network

ZHANG Shu-nan^1,2, CAO Feng^1,2, GUO Qian^1,2, QIAN Yu-hua^1,2,3

1 Institute of Big Data Science and Industry,Shanxi University,Taiyuan 030006,China
2 School of Computer and Information Technology,Shanxi University,Taiyuan 030006,China
3 Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education,Shanxi University,Taiyuan 030006,China

Received:2020-10-28 Revised:2021-03-25 Online:2021-05-15 Published:2021-05-09
About author:ZHANG Shu-nan,born in 1996,postgraduate.Her main research interests include deep learning and visual logic learning.(15735105265@163.com)
QIAN Yu-hua,born in 1976,Ph.D,professor,is a member of China Computer Federation.His main research interests include pattern recognition,feature selection,rough set theory,granular computing,and artifificial intelligence.
Supported by:
National Natural Science Foundation of China(61672332,61802238,61603228,62006146,F060308),Program for the Outstanding Innovative Teams of Higher Learning Institutions of Shanxi,Key R&D Program(International Science and Technology Cooperation Project) of Shanxi Province,China(201903D421003),Program for the San Jin Young Scholars of Shanxi,Overseas Returnee Research Program of Shanxi Province(2017023,2018172,HGKY2019001) and Shanxi Province Science Foundation for Youths(201901D211171,201901D211169).

摘要/Abstract

摘要： 逻辑推理是人类智能的核心,是人工智能领域一个富有挑战性的研究课题。人类的IQ测试问题是衡量人类智商水平高低和逻辑推理能力的常用手段之一,如何让计算机学习拥有类似人类的逻辑推理能力是一个非常重要的研究内容,其目的是使计算机从给定的图像中直接学习逻辑推理模式,而无需事先为计算机设计先验推理模式。基于此目的,提出了一种新的数据集Fashion-IQ,该数据集中的每个样本包含7张输入图片和1个标签,这7张图片分别为3张包含一种或多种逻辑的问题输入图片和4张选项输入图片,目的是利用机器学习3张问题输入图片中包含的逻辑来预测下一张图片,从而选择正确的选项。为了解决这个问题,提出了一种时序关系模型。针对每个选项,该模型首先使用卷积神经网络提取前3张输入图片和选项图片的空间特征;接着采用关系网络将这4个空间特征两两组合;然后采用LSTM提取前3张问题输入图片和该选项的时序特征,将时序特征与组合好的空间特征相结合得到时序-空间融合特征;最后对前3张输入图片与每个选项得到的时序-空间融合特征进行进一步推理,采用softmax函数进行打分,得分最高的选项就是正确答案。实验结果证明,该模型在此数据集上实现了比较高的推理准确度。

关键词: IQ测试, 逻辑推理, 时序关系网络, 时序-空间融合特征, 推理模式

Abstract: Logical reasoning is the core of human intelligence and a challenging research topic in the field of artificial intelligence.Human IQ test is one of the common methods to measure the level of human IQ and logical reasoning ability.How to let the computer learn to have the logical reasoning ability similar to human is a very important research content,the purpose is to make the computer from a given image directly learn the logical reasoning mode without having to design a priori reasoning mode for the computer in advance.For this purpose,a new data set Fashion-IQ is proposed.Each sample of the data set contains seven input pictures and a label.The seven pictures are three question input pictures that contain one or more logics,and four option input pictures.The purpose is to let the machine learn to predict the next picture based on the logic contained in the three question input pictures,so as to select the correct option.In order to solve this problem,the temporal relationship model is proposed.For each option,the model first uses a convolutional neural network to extract the spatial features of the first three input pictures and option pictures,and then uses a relation network to combine these four spatial features in pairs.Then,it uses LSTM to extract the first three question input pictures combining the time series feature with the time series feature of this option,the time series feature and the combined space feature are combined to obtain the time series-space fusion feature.Finally,the first three input pictures and the temporal-spatial fusion features obtained from each option are further reasoned,and the softmax function is used for scoring.The option with the highest score is the correct answer.Experiments prove that the model has achieved a relatively high inference accuracy on this data set.

Key words: Inference pattern, IQ test, Logical reasoning, Temporal relation network, Temporal-spatial fusion features

中图分类号:

TP181

张姝楠, 曹峰, 郭倩, 钱宇华. 一种基于时序关系网络的逻辑推理方法[J]. 计算机科学, 2021, 48(5): 239-246. https://doi.org/10.11896/jsjkx.201000171

ZHANG Shu-nan, CAO Feng, GUO Qian, QIAN Yu-hua. Logical Reasoning Method Based on Temporal Relation Network[J]. Computer Science, 2021, 48(5): 239-246. https://doi.org/10.11896/jsjkx.201000171

参考文献

[1]COLOM R,KARAMA S,JUNG R E,et al.Human intelligence and brain networks[J].Dialogues Clin Neuro,2010,12(4):489-501.
[2]ZHOU Z H.Abductive learning:towards bridging machinelearning and logical reasoning[J].Science China Information Sciences,2019,62(7):191-193.
[3]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2016:770-778.
[4]KAIMING H K,GEORGIA G,PIOTR D,et al.Mask R-CNN[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,2020,42(2):386-397.
[5]STERN W.The Psychological Methods of Testing Intelligence[J].Psychological Clinic,1915,9(2):56-59.
[6]TAIGMAN Y,YANG M,RANZATO M,et al.Deepface:Clo-sing the gap to human-level performance in face verification[C]//IEEE Conference on Computer Vision & Pattern Recognition.2014:1701-1708.
[7]XIONG W,DROPPO J,HUANG X,et al.The Microsoft 2016 Conversational Speech Recognition System[C]//2017 IEEE International Conference on Acoustics,Speech and Signal Proces-sing (ICASSP).2017:5255-259.
[8]HOSHEN D,WERMAN M.Iq of neural networks[J].arXiv:1710.01692,2017.
[9]SANTORO A,RAPOSO D,BARRETT D G,et al.A simple neural network module for relational reasoning[C]//31st Conference on Neural Information Processing Systems.2017:4974-4983.
[10]RAVEN J C.Raven's progressive matrices[M].Western Psychological Services,1938.
[11]SNOW R E,KYLLONEN P C,MARSHALEK B.The topography of ability and learning correlations[J].Advances in the Psychology of Human Intelligence,1984,2(S 47):103.
[12]KUNDA M,MCGREGGOR K,GOEL A.Addressing the Ra-ven's Progressive Matrices Test of “General” Intelligence[C]//2009 AAAI Fall Symposium Series.2009.
[13]WANG H,TIAN F,GAO B,et al.Solving verbal comprehension questionsin iq test by knowledge-powered word embedding[J].arXiv:1505.07909,2015.
[14]HOSHEN Y,PELEG S.Visual Learning of Arithmetic Operations[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2016:3733-3739.
[15]JOHNSON J,HARIHARAN B,LAURENS V D M,et al.Inferring and Executing Programs for Visual Reasoning[C]//2017 IEEE International Conference on Computer Vision.2017:2989-2998.
[16]XIAO H,RASUL K,VOLLGRAF R.Fashion-MNIST:a Novel Image Dataset for Benchmarking Machine Learning Algorithms[J].arXiv:1708.07747,2017.
[17]BARRETT D G T,HILL F,SANTORO A,et al.Measuring abstract reasoning in neural networks[C]//InternationalConfe-rence on Machine Learning.PMLR,2018:511-520.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

一种基于时序关系网络的逻辑推理方法

Logical Reasoning Method Based on Temporal Relation Network

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 10

Metrics

本文评价

推荐阅读 0

[1]	肖治鸿, 韩晔彤, 邹永攀. 基于多源数据和逻辑推理的行为识别技术研究 Study on Activity Recognition Based on Multi-source Data and Logical Reasoning 计算机科学, 2022, 49(6A): 397-406. https://doi.org/10.11896/jsjkx.210300270
[2]	吴立波, 黄玉芳. 基于DNA链置换的逻辑推理问题研究 Logical Reasoning Based on DNA Strand Displacement 计算机科学, 2022, 49(1): 259-263. https://doi.org/10.11896/jsjkx.210200131
[3]	琚安康,郭渊博,朱泰铭,王通. 网络安全事件关联分析技术与工具研究 Survey on Network Security Event Correlation Analysis Methods and Tools 计算机科学, 2017, 44(2): 38-45. https://doi.org/10.11896/j.issn.1002-137X.2017.02.004
[4]	王坚,史朝辉,郭新鹏,李伟平. Mamdani模糊推理算法的直觉化扩展 Intuitionistic Extension of Mamdani Fuzzy Reasoning Arithmetic 计算机科学, 2016, 43(Z6): 44-45. https://doi.org/10.11896/j.issn.1002-137X.2016.6A.009
[5]	贾志淳,邢星. 基于贝叶斯与多故障推理的Web服务诊断 Diagnosis of Web Service Based on Bayes and Multi-faults Reasoning 计算机科学, 2014, 41(6): 225-230. https://doi.org/10.11896/j.issn.1002-137X.2014.06.044
[6]	徐俊,肖刚,张元鸣,高飞,方赵林. 基于逻辑推理的构件行为片段提取与重组研究 Research on Component Behavior Fragment Extraction and Composition Based on Logical Reasoning 计算机科学, 2012, 39(5): 120-123.
[7]	胡小风邢永康. 融合概率和逻辑的推理模型研究计算机科学, 2006, 33(B12): 239-241.
[8]	张小红何华灿李伟华. 形式系统UL的弱完备性计算机科学, 2003, 30(12): 103-107.
[9]	周青. 关于程序验证方法的讨论计算机科学, 1995, 22(3): 58-60.
[10]	高全泉陆汝钤. 微机上实现的逻辑推理语言 Tuili 1.1 计算机科学, 1992, 19(5): 19-25.