Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599702
Yang Xu, Sui Ai-na, Tang Zhan-kun
In this study, through researching and analyzing the technology of Topical Crawler, we improve the critical algorithm, and present the Topic-Relevance judgment algorithm based on Multi-Level Vector Space Model and the Topical Search Strategy based on the content evaluation, which combine with the simple link structure analysis and link tags data analysis. Through the analysis of the experimental data, the algorithm and the strategy proposed in this research have higher accuracy and efficiency, which can improve the performance of the Traditional Topical Crawler greatly.
{"title":"Topical Crawler based on multi-level vector space model and optimized hyperlink chosen strategy","authors":"Yang Xu, Sui Ai-na, Tang Zhan-kun","doi":"10.1109/COGINF.2010.5599702","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599702","url":null,"abstract":"In this study, through researching and analyzing the technology of Topical Crawler, we improve the critical algorithm, and present the Topic-Relevance judgment algorithm based on Multi-Level Vector Space Model and the Topical Search Strategy based on the content evaluation, which combine with the simple link structure analysis and link tags data analysis. Through the analysis of the experimental data, the algorithm and the strategy proposed in this research have higher accuracy and efficiency, which can improve the performance of the Traditional Topical Crawler greatly.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123126759","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599705
A. Faghfouri, W. Kinsner, David Swatek
Lightning strikes are physical phenomena which have adverse effects on power delivery systems and lines. Predicting the ge ographical and te mporal distr ibution of the lightning str ike de nsities thr ough modeling can help power s ystem d esigners to imp rove th e protection of the existing and new power systems and transmission lines. In order to comp are th e res ults of the model simulations to the p hysical d ata, ch aracterization is required. The distribution of the lightning str ike de nsities ge nerates patterns w hich ar e highly nonline ar, nonstationar y, and stochastic. Traditional single-s cale analyzes and measures are insufficient to characterize such structures. Multiscale analysis of these patterns indicates their self-affinity over multiple scales, which is an indication of multifractality of these structures. Con sequently, mu ltifractal an alysis methods such as the Rényi fractal dimension spectrum are appropriate can didates for ch aracterization of these density map s. Th is p aper in vestigates con vergence in distribution of the lightning strike maps of Manitoba from 1998 to 2004. In addition, the sensitivity of the multifractal characterization to convergence in distribution is studied. The res ults in dicate a con vergence in statistical distribution for the lightning strike de nsity distributions and strong s ensitivity of th e Rén yi s pectra to th e d ata variations. This approach could be classified as cognitive engineering of p ower-delivery systems because it in eludes the sensing of th e environment and extraction of features over many scales and extended duration of time.
{"title":"A new approach to the study of lightning-strike density distributions","authors":"A. Faghfouri, W. Kinsner, David Swatek","doi":"10.1109/COGINF.2010.5599705","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599705","url":null,"abstract":"Lightning strikes are physical phenomena which have adverse effects on power delivery systems and lines. Predicting the ge ographical and te mporal distr ibution of the lightning str ike de nsities thr ough modeling can help power s ystem d esigners to imp rove th e protection of the existing and new power systems and transmission lines. In order to comp are th e res ults of the model simulations to the p hysical d ata, ch aracterization is required. The distribution of the lightning str ike de nsities ge nerates patterns w hich ar e highly nonline ar, nonstationar y, and stochastic. Traditional single-s cale analyzes and measures are insufficient to characterize such structures. Multiscale analysis of these patterns indicates their self-affinity over multiple scales, which is an indication of multifractality of these structures. Con sequently, mu ltifractal an alysis methods such as the Rényi fractal dimension spectrum are appropriate can didates for ch aracterization of these density map s. Th is p aper in vestigates con vergence in distribution of the lightning strike maps of Manitoba from 1998 to 2004. In addition, the sensitivity of the multifractal characterization to convergence in distribution is studied. The res ults in dicate a con vergence in statistical distribution for the lightning strike de nsity distributions and strong s ensitivity of th e Rén yi s pectra to th e d ata variations. This approach could be classified as cognitive engineering of p ower-delivery systems because it in eludes the sensing of th e environment and extraction of features over many scales and extended duration of time.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"635 ","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120939498","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599731
F. Pelayo, J. Valverde, Maria L. Pelayo, F. Cuartero
Formal methods in Computer Science are a particular kind of mathematically-based techniques for the specification, development and verification of software and hardware systems. Among them, Petri Nets have been widely used because of their high expressive power and accurate description capabilities. On the other hand, Dynamical Systems theory has not been used for these purposes so far. In this paper we present the first use of Discrete Dynamical Systems under the aim to better analyze Concurrent Computing Systems. In order to do that, Petri Nets, PNs, are properly encoded into DDSs, so defining the corresponding Phase Space with its metric structure and the evolution operator of the system.
{"title":"Discrete Dynamical Systems for encoding Concurrent Computing Systems","authors":"F. Pelayo, J. Valverde, Maria L. Pelayo, F. Cuartero","doi":"10.1109/COGINF.2010.5599731","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599731","url":null,"abstract":"Formal methods in Computer Science are a particular kind of mathematically-based techniques for the specification, development and verification of software and hardware systems. Among them, Petri Nets have been widely used because of their high expressive power and accurate description capabilities. On the other hand, Dynamical Systems theory has not been used for these purposes so far. In this paper we present the first use of Discrete Dynamical Systems under the aim to better analyze Concurrent Computing Systems. In order to do that, Petri Nets, PNs, are properly encoded into DDSs, so defining the corresponding Phase Space with its metric structure and the evolution operator of the system.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129976628","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599834
S. Fujita, K. Sugawara, C. Moulin, J. Barthès
The symbiotic application is designed to achieve the function of supporting a user based on the knowledge of social and perceptual interaction. The architecture of the symbiotic application was described at the ICCI'08. In this paper, we described the design of symbiotic base mechanism for the cognition functions and the decision functions in a symbiotic application. The symbiotic base mechanism converts the signals from the real space and the digital space to the awareness. The symbiotic base mechanism do the action on actuators based on the operation by the cognition functions and the decision functions. The cognition functions and the decision functions are achieved by the intelligent multi-agent system such as an OMAS and a ADIPS/DASH. On the other hand, the symbiotic base mechanism is constructed by a multi-agent system which is called by the Symbiotic Multi-Agent: SYMA. The example scenario is shown by the combination of multi-paradigm/multi-agent framework.
{"title":"The design of awareness and operation module for the symbiotic applications","authors":"S. Fujita, K. Sugawara, C. Moulin, J. Barthès","doi":"10.1109/COGINF.2010.5599834","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599834","url":null,"abstract":"The symbiotic application is designed to achieve the function of supporting a user based on the knowledge of social and perceptual interaction. The architecture of the symbiotic application was described at the ICCI'08. In this paper, we described the design of symbiotic base mechanism for the cognition functions and the decision functions in a symbiotic application. The symbiotic base mechanism converts the signals from the real space and the digital space to the awareness. The symbiotic base mechanism do the action on actuators based on the operation by the cognition functions and the decision functions. The cognition functions and the decision functions are achieved by the intelligent multi-agent system such as an OMAS and a ADIPS/DASH. On the other hand, the symbiotic base mechanism is constructed by a multi-agent system which is called by the Symbiotic Multi-Agent: SYMA. The example scenario is shown by the combination of multi-paradigm/multi-agent framework.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132924040","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599729
Zied Kechaou, M. B. Ammar, A. Alimi
With the growth of the Web2.0, e-commerce has become very popular in use, many websites offer the opportunity to make sales online and give the opportunity to get own an online review about objects, persons, and products. New opportunities and challenges arise as people can now actively use information technologies to seek and understand other people's opinions (sentiments) when to making their choices. In this paper, we propose a new two-step approach for an automatic sentiment processing, consisting a first step dealing with extract the subjective portions of reviews and a second one which determines the review overall sentiment by identifying whether a review appreciates (positive) or desappreciates (negative) its purpose. This approach consists of a four levels processing chain operating in a parallel way (i.e syntactic, semantic, linguistic and sentiment analysis).
{"title":"A new linguistic approach to sentiment automatic processing","authors":"Zied Kechaou, M. B. Ammar, A. Alimi","doi":"10.1109/COGINF.2010.5599729","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599729","url":null,"abstract":"With the growth of the Web2.0, e-commerce has become very popular in use, many websites offer the opportunity to make sales online and give the opportunity to get own an online review about objects, persons, and products. New opportunities and challenges arise as people can now actively use information technologies to seek and understand other people's opinions (sentiments) when to making their choices. In this paper, we propose a new two-step approach for an automatic sentiment processing, consisting a first step dealing with extract the subjective portions of reviews and a second one which determines the review overall sentiment by identifying whether a review appreciates (positive) or desappreciates (negative) its purpose. This approach consists of a four levels processing chain operating in a parallel way (i.e syntactic, semantic, linguistic and sentiment analysis).","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"62 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133320297","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599815
Zhou Lingli, Lai Jianghuang
With the growing applications of biometrics, there are many concerns about the security and privacy of the biometrics data. This paper proposes a security algorithm of face recognition, which bases on the local binary pattern (LBP) and random projection. Feature extraction is first performed on face images by LBP operator. Random projection is done using the random matrix which is generated by user specific key. All the above successive procedures produce the cancelable and non-invertible feature data, which is stored in template databases and can achieve the protection of biometrics data. In extensive experiments with publicly available face datasets ORL, YALE and FERET, higher recognition accuracy is reached, which demonstrates that the proposed approach is able to not only protect the biometrics data, but also is robust to various complex conditions, such as illumination, changes in the pose and expression, etc.
{"title":"Security algorithm of face recognition based on local binary pattern and random projection","authors":"Zhou Lingli, Lai Jianghuang","doi":"10.1109/COGINF.2010.5599815","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599815","url":null,"abstract":"With the growing applications of biometrics, there are many concerns about the security and privacy of the biometrics data. This paper proposes a security algorithm of face recognition, which bases on the local binary pattern (LBP) and random projection. Feature extraction is first performed on face images by LBP operator. Random projection is done using the random matrix which is generated by user specific key. All the above successive procedures produce the cancelable and non-invertible feature data, which is stored in template databases and can achieve the protection of biometrics data. In extensive experiments with publicly available face datasets ORL, YALE and FERET, higher recognition accuracy is reached, which demonstrates that the proposed approach is able to not only protect the biometrics data, but also is robust to various complex conditions, such as illumination, changes in the pose and expression, etc.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114082913","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599837
Akira Sakatoku, Akifumi Kawato, Toshiaki Osada, G. Kitagata, Tetsuo Kinoshita
In this paper, we propose 3D space which is based on concept of real and digital spaces symbiosis. Recently, collaborative work environments which are using Augmented Reality have been expected to realize effective collaborative works. However, these environments have some limitations. One of limitations is restricted availability of devices, and another limitation is low usability. Because of these limitations, it is difficult to utilize these environments, for general users. To break these limitations, we propose the construction scheme of 3D symbiotic space. Collaborative work environments based on 3D symbiotic space enable users to do collaborative works intuitively as if virtual spaces united with the real space. First, we present fundamental technologies for 3D symbiotic space. Next, we show the design and implementation of the agent-aided collaborative work environments based on 3D symbiotic space. Finally, we confirm feasibility of 3D symbiotic space with experimental results.
{"title":"3D Symbiotic space for agent-aided collaborative work","authors":"Akira Sakatoku, Akifumi Kawato, Toshiaki Osada, G. Kitagata, Tetsuo Kinoshita","doi":"10.1109/COGINF.2010.5599837","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599837","url":null,"abstract":"In this paper, we propose 3D space which is based on concept of real and digital spaces symbiosis. Recently, collaborative work environments which are using Augmented Reality have been expected to realize effective collaborative works. However, these environments have some limitations. One of limitations is restricted availability of devices, and another limitation is low usability. Because of these limitations, it is difficult to utilize these environments, for general users. To break these limitations, we propose the construction scheme of 3D symbiotic space. Collaborative work environments based on 3D symbiotic space enable users to do collaborative works intuitively as if virtual spaces united with the real space. First, we present fundamental technologies for 3D symbiotic space. Next, we show the design and implementation of the agent-aided collaborative work environments based on 3D symbiotic space. Finally, we confirm feasibility of 3D symbiotic space with experimental results.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114852514","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599734
Yan Ma, Robert E. Mercer
This paper mainly addresses the commutativity issue discovered in iterated belief revisions under the framework of AGM. Namely, given an iterated formula K ∗ x ∗ y, under what condition K ∗ x ∗ y = K ∗ (x ⁁ y) = K ∗ y ∗ x can hold needs to be reconsidered. As we shall demonstrate, AGM 7th and 8th postulates together falsely imply the commutativity existing in Rott's selection puzzle. On the other hand, Nayak et al.'s counteracting problem which demands commutativity is rejected by AGM postulates. The worse thing is the two realistic puzzles cannot be correctly simulated by AGM postulates under classic logics. Towards these issues, we suggest the inaccurate definition of consequence operation Cn needs to be modified in order to accommodate part of Rott's puzzle. Furthermore, we illustrated neither of the two puzzles can be simulated under the AGM framework in the content level. Therefore, we present our approach as follows: first, we built a belief revision framework which satisfies the first six AGM postulates in the content level. Second, by setting appropriate OCF, our framework is able to accommodate the two puzzles in epistemic state level. Third, we prove the commutativity can be held in the epistemic state level if and only if (K ∗ x) ∩ (K ∗ y) ≠ φ and (K ∗ x) ∗ (K ∗ y) ├/┴ and thus eliminate the intrusion by the selection puzzle. Last, we demonstrate the hold of commutativity is only a special case of the selection criterion we suggest, that is, the most explainable interpretation(s) that satisfy the most recent evidence should be chosen to be the final revision result. In this sense, we can tell in the counteracting problems, namely, when only (K ∗ x) ∩ (K ∗ y) ≠ φ is satisfied, why the revision result had better be represented by K ∗ (x ∗ y) is because the interpretation(s) suggested in our selection criterion is only guaranteed to be in K ∗ (x ∗ y) rather than its iterated form. Hence, we unify both of puzzles with the modified AGM framework.
{"title":"A solution to commutativity problems found in iterated belief revision","authors":"Yan Ma, Robert E. Mercer","doi":"10.1109/COGINF.2010.5599734","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599734","url":null,"abstract":"This paper mainly addresses the commutativity issue discovered in iterated belief revisions under the framework of AGM. Namely, given an iterated formula K ∗ x ∗ y, under what condition K ∗ x ∗ y = K ∗ (x ⁁ y) = K ∗ y ∗ x can hold needs to be reconsidered. As we shall demonstrate, AGM 7th and 8th postulates together falsely imply the commutativity existing in Rott's selection puzzle. On the other hand, Nayak et al.'s counteracting problem which demands commutativity is rejected by AGM postulates. The worse thing is the two realistic puzzles cannot be correctly simulated by AGM postulates under classic logics. Towards these issues, we suggest the inaccurate definition of consequence operation Cn needs to be modified in order to accommodate part of Rott's puzzle. Furthermore, we illustrated neither of the two puzzles can be simulated under the AGM framework in the content level. Therefore, we present our approach as follows: first, we built a belief revision framework which satisfies the first six AGM postulates in the content level. Second, by setting appropriate OCF, our framework is able to accommodate the two puzzles in epistemic state level. Third, we prove the commutativity can be held in the epistemic state level if and only if (K ∗ x) ∩ (K ∗ y) ≠ φ and (K ∗ x) ∗ (K ∗ y) ├/┴ and thus eliminate the intrusion by the selection puzzle. Last, we demonstrate the hold of commutativity is only a special case of the selection criterion we suggest, that is, the most explainable interpretation(s) that satisfy the most recent evidence should be chosen to be the final revision result. In this sense, we can tell in the counteracting problems, namely, when only (K ∗ x) ∩ (K ∗ y) ≠ φ is satisfied, why the revision result had better be represented by K ∗ (x ∗ y) is because the interpretation(s) suggested in our selection criterion is only guaranteed to be in K ∗ (x ∗ y) rather than its iterated form. Hence, we unify both of puzzles with the modified AGM framework.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"122 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116026250","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599827
H. D. Garis, Chen Xiaoxi, Yang Ye, Chen Shuo, B. Goertzel, Ruiting Lian
This paper introduces a Partially Connected Neural Evolutionary (Parcone) model and its application to object and gesture recognition as part of the "China Brain Project", which aims to build China's first artificial brain, consisting of 10,000s of evolved interconnected neural net modules, each with its own little task, similar to Minsky's agents.
{"title":"Object/gesture recognition software in the “China Brain Project”","authors":"H. D. Garis, Chen Xiaoxi, Yang Ye, Chen Shuo, B. Goertzel, Ruiting Lian","doi":"10.1109/COGINF.2010.5599827","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599827","url":null,"abstract":"This paper introduces a Partially Connected Neural Evolutionary (Parcone) model and its application to object and gesture recognition as part of the \"China Brain Project\", which aims to build China's first artificial brain, consisting of 10,000s of evolved interconnected neural net modules, each with its own little task, similar to Minsky's agents.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114677547","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}
Pub Date : 2010-07-07DOI: 10.1109/COGINF.2010.5599743
Rongbin Li, Eddie C. L. Chan, G. Baciu
Singularity problem in human face feature extraction is very challenging that has gained a lot of attentions during the last decade. A pseudo-inverse linear discriminant analysis (LDA) plays a important role to solve the singularity problem of the scatter matrices. In this paper, we make use of Linear Discriminant Analysis via QR decomposition (LDA/QR) and Direct Linear Discriminant Analysis (DLDA) to solve the singularity problem in face feature recognition. We also show that an equivalent relationship between DLDA and LDA/QR. They can be regarded as a special case of pseudo-inverse LDA. Similar to LDA/QR algorithm, DLDA could be a two-stage LDA method. Interestingly, we find that the first stage of DLDA can act as a dimensionality reduction algorithm. In our experiment, we compare DLDA and LDA/QR algorithms in terms of classification accuracy, computational complexity in ORL and Yale face datasets. We have also conducted experiments to compare their first stages on these datasets. Our results indicate that the empirical and theoretic proofs of equivalence between DLDA and LDA/QR algorithms coincidentally converge and verify their same capabilities in the dimension reduction.
{"title":"DLDA and LDA/QR equivalence framework for human face recognition","authors":"Rongbin Li, Eddie C. L. Chan, G. Baciu","doi":"10.1109/COGINF.2010.5599743","DOIUrl":"https://doi.org/10.1109/COGINF.2010.5599743","url":null,"abstract":"Singularity problem in human face feature extraction is very challenging that has gained a lot of attentions during the last decade. A pseudo-inverse linear discriminant analysis (LDA) plays a important role to solve the singularity problem of the scatter matrices. In this paper, we make use of Linear Discriminant Analysis via QR decomposition (LDA/QR) and Direct Linear Discriminant Analysis (DLDA) to solve the singularity problem in face feature recognition. We also show that an equivalent relationship between DLDA and LDA/QR. They can be regarded as a special case of pseudo-inverse LDA. Similar to LDA/QR algorithm, DLDA could be a two-stage LDA method. Interestingly, we find that the first stage of DLDA can act as a dimensionality reduction algorithm. In our experiment, we compare DLDA and LDA/QR algorithms in terms of classification accuracy, computational complexity in ORL and Yale face datasets. We have also conducted experiments to compare their first stages on these datasets. Our results indicate that the empirical and theoretic proofs of equivalence between DLDA and LDA/QR algorithms coincidentally converge and verify their same capabilities in the dimension reduction.","PeriodicalId":365083,"journal":{"name":"9th IEEE International Conference on Cognitive Informatics (ICCI'10)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133164321","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：