Upload
trinhcong
View
217
Download
4
Embed Size (px)
Citation preview
R eferences
1. Akaike, H. (1973) . Information theory and an exte nsion of the maximum likelihood principle. In : Th e 2nd International Symposium on Information Th eory.Akad emiai Kiado, Budapest . pp. 267-281.
2. Alford, A. and C. J . Harris (1998) . Using B-splines to represent optimal actuator demands for systems featuring position /rate limited actuators . In : Proc.of the 5th IFA C Workshop, AIARTC'9S. Mexico. pp. 271-276.
3. An, P. E . and C . J . Harris (1995) . An intelligent driver warning syste m for vehicle collision avoidance. IEEE Tran sactions on Sy st ems, Man and Cybern etics26 (2), 254-261.
4. An , P. E ., M. Brown and C. J . Harris (1995) . A global gradient noise covarian ceexpress ion for stationar y real ga ussian input s. IEEE Tran saction s on NeuralNetworks 6(6), 1549-1 551.
5. An, P . E., M . Brown and C. J . Harris (1997) . On the convergence rate of t henormalised leas t mean square adaption . IEEE Transactions on Neural Networks8 (5) , 1211-1215.
6. An , P . E. , M. Brown, C. J . Harris , A. J . Lawrence and C. G. Moore (1993) .Comparat ive aspects of asso ciative memory networks for mod elling. In : Proc.2nd European Control Conference, pp. 454-459.
7. Aronszajin, N. (1950) . T heory of reproducing kernels . Tran sactions of American Mathem atics Society 68 , 337-404.
8. Astrom , K. J . and B. W ittenmark (1989) . Adaptive Con trol. Addison Wesley,MS.
9. Atkinson, A. C. and A. N. Donev (1992) . Optimum Experimental Designs.Clarendon Press , Oxford.
10. Avedyan, E . D., M. Brown and C. J . Harris (1995). A deterministic and stat istical an alysis of the modifi ed NLMS rules. B elizer Jou rnals 30, 14-43.
11. Bar-Shalom , Y. (1981) . On the t rack-to-t rack correlation problem . IEEE Tran sactions on Automatic Control 26 (2), 571-572.
12. Bar-Shalom, Y. and L. Campo (1986) . The effect of the common process noiseon the two-sensor fused -track covariance. IEEE Tran sactions on Aerospace andElectronic Systems 22 (6), 803- 805.
13. Bar-Shalom , Y. and T . E . Fortmann (1988) . Tracking and Data A ssociation.Academic P ress, New York .
14. Bar-Shalom , Y. and X. R. Li (1995). Multitarget-Multisensor Tracking: Principles and Techniques. YBS , St orrs, Conn.
15. Barto, A. G., R. S. Sutton and C. H. Anderson (1983). Neuronlike adapt iveelements that can solve difficult learn ing control prob lems. IEEE Tran sactionson System s, Man and Cybern etics 13(5) , 834- 846.
16. Belsley, D. (1991). Condit ionin g Diagonost ics: Collin earity and W eak Data inRegression. W iley, Chichester .
30S References
17. Berger , C. S. (1994) . Linear splines with ada pt ive mesh sizes for modellin gnonl inear dyn amical systems . Proc. lEE Control Th eory and Applications141(5) , 77- 2S4.
IS. Billings, S. A. and Q. H. Tao (1991). Mode l validity tests for nonlinear signa lprocessing applications . Intern ational Journ al of Control 54(1) , 157-194.
19. Billings , S. A. and Q. M. Zhu (1994) . Nonlinear mod el validation usin g correla t ion tests. In tern ational Journal of Control 60 , 1107-1120.
20. Billings , S. A. and Q. M. Zhu (1995) . Model valida t ion tests for mu lti vari ablenonlinear models including neural networks. International Journal of Control62 , 749-766.
21. Billings, S. A. and W. S. F. Voon (19S6). Correlation based mod el validity testsfor nonlinear models. In terna tional Journ al of Control 44(1) , 235-244.
22. Billings , S. A. and W . S. F . Voon (19S7). Piecewise linear identification ofnonlinear sys te ms. Int ernational Journal of Control 46 , 215-235.
23. Bishop , C. M. (1995). Neural Networks for Pattern Recognit ion. Oxford University Press , Oxford.
24. Bossley, K. M . (1997) . Neurofuzzy mod elling approaches in syste m identification. PhD thesis. Dept of ECS, University of Southampton.
25. Bossley, K. M., M. Brown and C . J . Harris (1995) . Neurofuzzy model construction for the modelling of nonlinear processes . In : Proc. of the 3rd Europ eanControl Conference. Vol. 3. Rome, Italy. pp. 243S-1443.
26. Bossley, K. M. , M. Brown and C. J . Harris (1999) . Neurofuz zy identification ofan auto nomous underwater vehicle . International Journal of Sy st ems Science30(9) ,901-913.
27. Box , G. E. P . and G. M. Jenkins (1976). Tim e Series Analysis , Forecast ing andControl. Hold en-Day, London .
2S. Breim an , L. (1991 ). The I1 method for est imating mul tivari ate fun ctions fromnoisy dat a . Technometri cs 33(2), 125-1 53.
29. Bridgett , N. A. and C . J . Harris (1994) . Neurofuzzy iden ti fica t ion and controlof a gas turbine jet engine. In : Proc. In terna tional Symp. on Signa l Processing,Robotics And Neural Networks. Lille , France. pp . 462-467.
30. Bridget t , N. A. , J . Brandt and C. J . Harris (1995) . A neural network for use inthe diagonosis and treatment of br east cancer. In : Proc. of the 4t h Int ernationalConference on ANN. Cambridge , UK. pp . 44S-453.
31. Bro omhead , D. S. and D. Lowe (19SS). Multivari able fun ction al interpolationand adapt ive networks. Complex Systems 2, 321- 355.
32. Brown, M. and C . J . Harris (1994). Neurofu zzy Adaptive Modelling and Cont rol.Prentice Hall, Hemel Hempstead .
33. Brown, M . and T . Ullri ch (1997). Comparison of nod e inser tion algorit hms forDelaunay networks. In: IMA CS 2nd Mathmod. Vol. 2. Vienna , Austria, pp . 7757S0.
34. Brown, M., S. R. Gunn, C. N. Ng and C . J . Harris (1997) . Neurofuzzy modelling: a transparent approach. In : Dealing with Complexity: A N eural NetworkApproach (K. Warwick, Ed .) . pp. 110-124, Springer-Verlag, Beilin.
35. Buj a, A., T . Hastie and R. Tibshiran i (19S9). Linear smoot hers and additivemodels. Annals of St atistics 17(2) , 453-535.
36. Buntine, W . L. and A. S. Weigend (1991). Bayesian back-p rop agation . ComplexSystems 5, 603- 643.
37. Casdagli, M. C. (19S9) . Nonlinear pr ediction of chaot ic ti me series. Physica D20, 335- 356.
3S. Casdagli, M. C., S. Eubank, J. D. Farmer and J . Gibso n (1991). State spacereconstruction in the pr esence of noise. Physica D 51 , 52-9S.
References 309
39. Chan, C. W ., W . C. Chan , K. C. Cheung and C. J. Harr is (2000) . Support vecto r neurofuzzy networks for mod elling nonlinear dyn am ic systems . Submittedto International Journal of Systems Science.
40. Chan, W . C., C. W . Chan, K. C. Cheung and C. J . Har ris (2001). On the modelling of nonlinear dynami c systems using suppo rt vector machines. En gin eeringand Ap plications of AI 14(2), 105-114.
41. Chang, K. C., R . K. Saha and Y. Bar-Sh alom (1997) . On opt imal track-to-t rackfusion . IEEE Transactions on Aerospace and Electronic Syst em s 33(4) , 12711276.
42. Chang , K. C., Z. T ian and R . K. Sah a (1998) . Performance evalua t ion oftrack fusion with information filter . In : Proc. of Int erna tional Conference onMultisource-Multisensor Information Fusion. USA. pp. 648-655.
43. Chen , F . C. and C. C. Liu (1994). Adaptively cont rolling nonlinear cont inuoustime systems usin g multilayer neural networks. IEEE Transactions on Autom atic Control 39(6), 1306-1310.
44. Chen, F . C . and H. K. Khalil (1995) . Adaptive control of a class of nonlineardiscret e-time sys tems using neural networks . IEEE Transactions on Autom aticControl 40(5) , 791-801.
45. Chen, S. and S. A. Billings (1992). Neur al netwo rks for nonlinear dynamicsys te m mod ellin g and identifi cation . In ternational Journal of Control 56, 319346.
46. Chen, S. , C. F . Cowan and P. M. Grant (1991). Or thogon al least squares learning algorithm for radial basis funct ion networks. IEEE Transactions on NeuralNetworks 2, 302-309.
47. Chen, S., S. A. Bill ings and W . Luo (1989). Orthogon al least squa res methodsand their applications to nonl inear sys tem identification . Intern ational Journalof Control 50, 1873-1896.
48. Chen , S., Y. Wu and B. L. Luk (1999). Combined genetic algorithm optimizat ion and regularized orthogonal least squa res learning for radial basis functionnetworks. IEEE Transaction s on Neural Networks 10, 1239-1243.
49. Ch erkassky, V. S. and F . M. Mulier (1999) . Learning from Data: Concepts,Th eory, and Methods. Wiley.
50. Chui, C. K. and G . Ch en (1987). Kalman Filt ering: with Real-Time and Applications. Springer-Verl ag, Berlin.
51. Cox, E. (1994). Th e Fuzzy Sy stem s Handbook. AP Professional.52. Daubechies,1. (1992) . Ten lectures on wavelets. In : CBMS- NSF Regional Con
ference Series in Applied Math ematics. Vol. 61. SIAM, Princeton, R1.53. Dempster , A. P., N. M. Laird and D. B. Rubin (1977) . Maximum likelihood
from incomplete data via the EM algorit hm (with discussions) . Journal of theRoyal Statistical Society : B 39, 1- 39.
54. Dixon, L. C. W . and R . Price (1986) . Numerical experience with the truncatedNewton method. Techni cal rep ort . Numerical Op timisation Cent re, Univ. Hertfordshire, Hatfield , UK.
55. Dodd, T . J . (2000) . Prior knowledge for time series modelling. PhD thesis .Dept . of ECS, Univers ity of Southampton.
56. Dowdy, S. M. and S. Wearden (1991). Statistics for Research. Wil ey, New York .57. Doyle, R. S. (1996) . Neurofuzzy multi-sensor data fusion for helicopter obstacle
avoidance . PhD thesis . Dept of ECS, University of Sou thampton.58. Doyle , R. S. and C. J. Harris (1996) . Multi-sensor data fusion for helicopter
guidance using neurofuzzy estimat ion algorit hms . Th e Royal A eronautical Society Journal 98, 241-251.
59. Dri ankov, D., H. Hellendoorn and M. Reinfrank (1993). An Introduction toFuzzy Control. Springe r-Verlag, Berlin.
310 References
60. Durrant-Whyte, H. F . (1988) . Integration, Coordination and Control of MultiSensor Robot Systems. Kluwer Academic, New York .
61. Ellacott , S. W . (1994) . Aspects of the numerical analysis of neural networks .Acta Numerica 18, 145-202.
62. Ergenion, T., M. Pontil and T . Poggio (1999) . A unified framework for regularisation networks and support vector machines. Technical report . MIT AI LabMemo (BCL paper no. 171).
63. Farin, G. (1994) . Curves and Surfaces fo r Computer-aided Geom etric Design:a Practical Guide. Academic Press, Boston.
64. Femminella, O. P., M. J . Starink, M. Brown, 1. Sinclair, C . J. Harris and P. A. S.Reed (1999) . Data preprocessing/model initialisation in neurofuzzy modellingof structure property relationships in AI-Zn-Mg-Cu alloys . ISIJ 39(10) , 10271037.
65. Femminella, O. P. (2000) . Neurofuzzy and SUPANOVA modelling of theprocessing-property relationships of aerospace AI-alloys. PhD thesis. Facultyof Engineering, University of Southampton.
66. Femminella, O. P, M. J. Starink, S. R. Gunn, C. J . Harris and P. A. S. Reed(2000) . Neurofuzzy and SUPANOVA modelling of structural property relationships in AI-Zn-Mg-Cu alloys. In : Proc. of 7th International ConferenceAluminium Alloys, Material Science Forum. Vol. 331-337. Charlottesville, Va.pp. 1255-1260.
67. Feng, M. and C. J. Harris (1998). Adaptive neurofuzzy control for a class ofstate-dependent nonlinear processes. International Jou rnal of Systems Science29(7), 759-771.
68. Feng , M. and C. J. Harris (2001) . Pie cewise Iyapunov stability conditions offuzzy systems. IEEE Trans actions on Systems, Man and Cybernetics 31:PartB(l) , 259-262.
69. Feraday, S. (2001). Intelligent approaches to modelling and intepreting discbrake squeal data. PhD thesis. Dept of ECS, University of Southampton.
70. Fletcher, R . (1987) . Practical Methods for Optimisation. 2nd edn., Wiley,Chichester.
71. French, M. (1998) . Adaptive control of functionally un certain systems. PhDthesis, University of Southampton.
72. French, M. and E. Rogers (1997). Approximate models for adaptive feedbacklinearisation. Internation al Journal of Control 68(6), 1305-1323.
73. Friedman , J . H. (1991) . Multivariate adaptive regression splines. The Annalsof Statistics 19(1),1-141.
74. Gan , Q. and C. J . Harris (2001) . A hybrid learning scheme combining EM andMASMOD algorithms for nonlinear system modelling. IEEE Transa ctions onNeural Networks 12(1) , 43-53.
75. Gan, Q. and C. J . Harris (2001) . Comparison of two measurement fusion methods for Kalman-filter-based multisensor data fusion. IEEE Transactions onAerospace and Electronic Sy st em s 37(1) , 273-280.
76. Gan, Q. and C. J . Harris (1999). Fuzzy locallinearisation and local basis function exp ansion in nonlinear system modelling. IEEE Transactions on Systems ,Man and Cybern etics 29: Part B(4) , 559-565.
77. Gan, Q . and C. J. Harris (1999) . Linearisation and state estimation of unknowndiscrete-time nonlinear dynamic systems using recurrent neurofuzzy networks.IEEE Tran sactions on Systems, Man and Cybern etics 29: Part B(6) , 802-817.
78. Gan , Q. and C. J. Harris (1999) . Multisensor data fusion using Kalman filtersbased on linearised process models. In : Proc. of International Conference onData Fusi on, EuroFusion 99. UK. pp . 105-112.
Referen ces 311
79. Gan, Q . and C. J . Harris (1999) . Neurofuzzy state estimato rs using a modifi edASMOD and Kalman filter a lgorit hm. In : Proc. of the Intern ational Conf erenceon Compu tational In telligen ce for Modelling Con trol and Autom ation. Vol. 1.Vienna , Austria. pp . 214-219.
80. Gao, J . B., C. J . Harris and S. R. Gunn (2001). On a class of suppo rt vect orkernels based on frames in fun ction Hilbert spaces. Neural Computation 13, 120.
81. Gao, J . B ., S . R. Gunn and C. J . Harris (2001). Mean field t heo ry for theSVM regression . ISIS Technical Rep ort , Dep artment of El ectron ics and Computer Scien ce, Uni versity of Southampton. Submit te d to IEEE Transactionson Neural Networks.
82. Geman, S., E . Bien enstock and R. Doursat (1992) . Neural networks and thebias /variance dil emma . Neural Computation 4 , 1- 58.
83. Gers henfeld , N., B. Schoner and E. Met ois (1999). Clus te r-weighte d mod elin gfor time-seri es analys is. Nature 39 7, 329- 332.
84. Gill, P. E. , W . Murray and M. H. Wr ight (1993). Pra ctical Optimisat ion. Academi c Press , San Diego.
85. Girosi , F ., M. J ones and T . Poggio (1995). Regularisation theory for neuralnetworks. Neural Computation 7, 219-269.
86. Golub , G. H., M. Heath and G . Wahba (1979). Gener aliz ed cross-validat ion asa method for cho osing goo d ridge parameter. Technometri cs 21(2), 215-223.
87. Goodwin , G . C. and K. S. Sin (1984). Adaptive Filt ering Predict ion and Control.Prentice Hall, Englecliff, NJ .
88. Grime, S. and H. Durrant-Whyt e (1994). Data fus ion in decen tra lised senso rnetworks. Control Eng in eering Pra ctice 2(5) , 849- 863.
89. Gull, S. F . (1989) . Developments in maximum ent ro py data analysis . In : Maximum Entropy and Bayesian Methods (J . Skilling, Ed.) . pp . 53-71, KluwerAcademic.
90. Gunn , S. R . and J . S. Kandola (2001). Structural mod elling with sp arse kernels .Machin e Learning, Accepted .
91. Gunn, S. R. and M. Brown (1999) . SUPANOV - a sparse transparent modellingapproach . In: Proc. of IEEE International Workshop on N eural Networks forSignal processing . Madison, Wi s., USA .
92. Gunn, S. R ., M. Brown and K. Bossley (1997) . Network performance assessme ntfor neurofuzzy data mod elling. In : Intelligent Data A nalysis , Lecture Notes inCompute r Scien ce, pp . 313- 323, Springer , Berlin.
93. Hall , M. and C. J . Harris (1998). Neurofuzzy syste ms for command and cont rol.In : Proc. of the 0 2 R eseach and Technology Symposium. Monterey, Naval PGS.pp . 461-471.
94. Hardier , G . (1997) . Recurrent neural networks for ship mod elling and cont rol.In : Proc. of the 7th Sh ip Control Sy stems Symposium. Vol. 1. pp . 39-61.
95. Harris, C . J ., A . Bailey and T . J . Dodd (1998). Multi-sensor data fusion inde fence and aerospace . Th e Royal Aeronautical Jou rn al 102 (1015), 229-244.
96. Harris , C. J . and P. E. An (1998) . An intelligent driver warning system . In :Knowledge Ba sed Intelligent Techniques in Industry. pp . 1- 52, CRC Press,Boca Raton .
97. Harris, C . J . and Q. Gan (2001) . State est imat ion and multisen sor data fusionusing data-based neurofuzzy local lineari sation process models. InformationFusion 2(1), 17-29.
98. Harris , C . J . and X. Hon g (2000). Neurofuzzy network mod el const ructionusing Bezier-Bernst ein polynomial fun ction s. Proc. lEE Control Th eory andApplications 14 7, 337-343 .
312 References
99. Harris , C. J . and X. Hong (2001). A par allel learning and const ructio nalgor it hm. In : Proc. of SPIE- A pplications and Scien ce of Com putational Intelligen ce II. Orl ando, USA.
100. Harri s , C. J. , C. G. Moore and M. Brown (1993) . Intelligent Con trol: SomeA spects of Fuzzy Logic and Neural Networks. World Scient ific Press, Singapore.
101. Harris , C. J ., X . Hong and M. Feng (1999) . Op timal piecewise locally linearmod elling. In : Proc. of SPIE - A pplicatio ns and Science of Comp utationalIntelligen ce II. Vol. 3722 . Orl ando, USA. pp. 486- 493.
102. Harris , C . J ., X. Hong and P. A. Wilson (1999) . An intelligent guidance andcontrol system for ship obstacle avoidance. Proc. IMechE Part I, Journal ofSy st ems and Control Eng in eering 213, 311-320.
103. Hastie, T . J. and R . J . Tibshirani (1996) . Generalised Additiv e Models. Chapman and Hall , London .
104. Haykin, S. (1999) . Neural Networks: A Comprehensive Foundation. PrenticeHal l.
105. Hong, X. and C . J Harris (2000) . Generalised neurofuzzy network mod ellingalgorit hms usin g Bezier-Bernstein polynomi al functions and addit ive decomposition . IEEE Transactions on Neural Networks 11(1) , 889- 902.
106. Hong, X. and C. J Harris (2001) . A mix ture of experts network struc t ure construction algorit hm for modellin g and control. Applied In telligence, In Press .
107. Hong, X. and C. J Harris (2001) . Nonlinear mod el st ruc t ur e detecti on usingopt imum expe rimental design and orthogo nal least squa res . IEEE Trans actio ns on Neural Networks 12(2), 425-439.
108. Hong, X. and C . J Harris (2001). Neurofuzzy design and mod el const ruc t ionof nonlinear dyn amical processes from data . Proc. lEE Control Th eory andApplications, to appear.
109. Hong, X. and C. J Harris (2001) . Neurofuzzy networks knowledge extract ionand exte nded Gram-Schmidt algorit hm for subspace decomposition. TechnicalRep ort, ISIS Group, Dept of Electronics and Compute r Science, University ofSouthampton .
110. Hong, X. and C. J Harris (2001). Vari able select ion algorithm for the constru ction of MIMO operat ing point dependent neurofuzzy network . IEEE Transactions on Fuzzy Sys tems 9 , 88-101.
111. Hong, X., C. J . Harri s and P. A. Wilson (1999). Neurofuzzy state identificationusing prefilt ering. Proc. lEE Control Th eory and Applications 146, 234-240.
112. Horn, R. A. and Johnson J . R (1985) . Matrix Analysis. Cambridge UniversityPress, Cambridge.
113. Huber , P. (1985). P roj ection pursuit. Annals of Statistics 13 ,1465-1 481.114. Hunt , K. J ., R . Haas and M. Brown (1995) . On the fun ction al equivalence of
fuzzy infer ence systems and spline-based networks. Int ern ational Journal ofNeural Sys tem s 6 (2),171-184.
115. Ingber , L. (1992). Geneti c algorithms and very fast simulate d rean ealling: acomparison. Mathem atical and Computer Modelling 11,87-100.
116. Isidori, A. (1989) . Nonlinear Control Sy st ems. Springer-Verlag , Berlin.117. Ivakhnenko, A. G . (1971) . Polyn omial theory of complex syste ms. IEEE
Transactions on Systems , Man and Cyberne tics 1 (4) , 364- 378.118. J acobs, R. A., M. 1. Jordan, S. J . Nowlan and G . E . Hint (1991) . Ada pt ive
mix tures of local expe rts. Neural Computatio n 3 , 79- 87.119. Jagannathan, S. and F . L. Lewis (1996) . Multilayer discrete-time neural net
controller with gua ranteed performance. IEEE Transactions on Neural Networks 7(1) , 107-130.
References 313
120. J agannathan , S. and F . L. Lewis (1997 ). Feedback lineari sation in discret et ime usin g neural network s. In : Proc. of the 12th IEEE Internation al S ym p.on Intelligent Control. Istanbul, Turkey. pp. 181-186.
121. J an g, J . S. R. (1993). ANFIS: Adap tiv e-network -based fuzzy inference systems . IEEE Transactions on Systems, Man and Cybernetics 23 , 665-685.
122. J an g, J . S. R., C.T. Sun and E. Mizutani (1997). N euro-Fuzzy and Soft Com puting: A Computati onal Approach to Learning and Ma chine Intelligence.Uppe r Saddle River , Prentice Hall , NJ.
123. Jin , L., P . N. Nikiforuk and M. M. Gupta (1995) . Approximation of discretetime state-space tra jectories using dynamic recurrent neural networks. IEEETransact ions on Automatic Control 40(7) , 1266-1270.
124. Joh an sen , T . A. and B. A. Foss (1993). Construc t ing NARMAX models usingARMAX models . International Journal of Control 58, 1125-1153.
125. Jordan , M. and L. Xu (1995) . Convergence results for the EM approach tomixtures of expe rts' architec t ures. N eural N etworks 8(9) , 1409-1431.
126. Jordan, M. and R. A. Jacobs (1994). Hierar chical mixtures of experts and theEM algorit hm. N eural Computation 6 , 181-214.
127. Kan ellakopoulos , I. , P. V. Kokotovic and A. S. Morse (1991) . Systemati c design of adapt ive controllers for feedback linearisable syste ms. IEEE Transactions on Automatic Control 36(11) , 1241-1253.
128. Karlsson , M. , A. Malmberg , T . J ensen and L. Axelsson (1999) . Tr ack fusionalgorithms in decentralised tracking systems with feedback in a flight air craftapplication . In : Pro c. of FUSION'99. USA . pp. 733-740.
129. Kavli , T . (1993). ASMOD - an algorithm for adapt ive spline modelling ofobs ervation data. International Journal of Control 58(4) , 947-967.
130. Kim, Y . H., F. L. Lewis and C. T. Abdall ah (1997) . A dynamic recurrentneural-network-based ad aptive observer for a class of nonlinear systems. Automatica 33(8), 1539-1543.
131. Krstic, M. and P. V. Kokotovic (1996). Adaptive nonlinear output-feedbackschemes with Marino-Tomei controller. IEEE Transactions on AutomaticControl 41(2), 274-280.
132. Lane, S. H. (1992). Theory and development of higher order CMAC neuralnetworks. IEEE Control System s Maga zin e 12, 23-30.
133. Leontaritis, I. J . and S. A. Billings (1985) . Input-output parametric models fornonlinear systems - part 1: deterministic nonlinear systems; part 2: st ochasticnonlinear systems. International Journal of Control 41(1) , 303-344.
134. Leontaritis , I. J. and S. A. Billings (1987) . Model select ion and validationmethods for nonlinear systems . International Journal of Control 45(1) , 311341.
135. Lin , C . T. (1996) . N eural Fuzzy Syst ems: A N euro -Fuzzy Synergism to Intelligent Syst ems. Prentice Hall , Upper Saddle River , NJ.
136. Liu , C . C. and F . C. Chen (1993). Adaptive cont rol of nonlinear continuoustime systems using neural networks - general relative degree and MIMO cases.International Journal of Control 58 , 317-355.
137. Ljung, L. (1987) . System Identification: Theory fo r the User, Prentice Hall.138. Lo, S. H. (1989 ). Delaunay triangulation of nonconvex plan domains . Inter
national Journal of Numerical Methods in Engineering 28 , 2695-2707.139. Mackay, D. J . (1997) . Gaussian processes: a replacement for
supervised neural networks. In : Lecture Notes at NIPS 1991.http:/ /www.ra .phy.cam.ac.uk/rnackay,
140. MacKay, D. J. C . (1991) . Bayesian methods for adapt ive models. PhD thesis .California Institute of Technology, USA.
314 References
141. Manyika, J . and H. Durran t-Whyte (1994). Data Fusion and Sens or Man agem ent: A Decentral ized Informa tion-Theoretic Approach. Ellis Horwood, NewYork.
142. Mar enbach , P. and M. Brown (1997). Evolutionary versus inductive const ruction of neurofuzzy sys te ms . In : Genetic Algorithms in Engi neering Sy st em s:Innovations and Applications (GA LESIA '97) . Glasgow, UK.
143. Marino, R. and P. Tomei (1992). Global adaptive observers for nonlinear syst ems via filtered transformations. IEEE Transactions on Automatic Control37(8) , 1239-1245.
144. Marino, R . and P. Tomei (1993) . Global adaptive output-feedback controlof nonlinear systems, part I: linear parameterisation. IEEE Transactions onAutomatic Control 38(1) , 17-32.
145. Marino, R . and P. Tom ei (1993) . Global adaptive out put-feedback control ofnonlinear systems , part II: nonlinear parameterisation. IEEE Transa ctions onAutomatic Control 38(1) , 33-48.
146. Marino, R. and P. Tom ei (1995) . Adaptive observers with arbit rary exponential rate of convergence for nonlinear systems. IEEE Transactions on Automatic Control 40(7) , 1300-1304.
147. Michie, D. and R. A. Chambers (1968) . Boxes: an expe riment in adaptivecontrol. In : Mach in e Intelligence 2 (E . Dale and D. Michie, Eds.) . pp . 137152. Oliver and Boyd, Edinburgh.
148. Miller , W . T ., R .S. Sutton and P. J. Werbos (Eds.) (1990) . Neural Networksfor Control. MIT Press , Cambridge, MA.
149. Moody, J . (1994) . Prediction risk and architecture select ion for neural networks. In : From Statist ics to Neural Networks: Th eory and Pattern Recognition and Applications (V . Cherkassky, Ed .). Springer-Verlag, Berlin.
150. Moore, C. G. and C. J. Harris (1994) . Aspects of fuzzy control and est imat ion .In: Advan ces in Intelligent Control (C . J . Harris, Ed .). pp . 201-242. Taylor &Francis , London.
151. Murray-Smith, R . (1994) . A local model network approach to nonlinear modelling . PhD thesis. Dept. of Computer Science , Univers ity of Strathclyde.
152. Murray-Smith, R. and T. A. Johan sen (1997) . Mult iple Model Approaches toModelling and Control. Taylor & Francis, London.
153. Nea l, R. M. (1996) . Bayesian Learning for Neural Networks. Lecture Notes inStatistics 118, Springer , New York.
154. Nelles , 0 ., S. Sinsel and R . Isermann (1996) . Local basis function networksfor identification of a turbocharg er. In : lEE UKACC Control'96. Exeter.
155. Netter, J ., M. H. Kutner , C. J . Nathtshein and W . Wasserman (1985) . AppliedLinear Statistical Models. Irwin , London.
156. Oinohundro, S. M. (1989) . The Delaunay triangulation and function learning.Technical report . Berkeley Tech. Report,90-001.
157. Orr, M. J. L. (1995) . Regularisation in the selection of radial basis functioncenters. Neural Computation 7(3), 954-975.
158. Papoulis, A. (1965) . Probability, Random Variables, and Stochastic Processes .McGraw-Hill, New York .
159. Parzen , E . (1967). Times Series Papers. Holden Day, San Francisco.160. Pearlmutter, B. A. (1995). Gr adi ent calculations for dynami c recurrent neural
networks: a survey. IEEE Transa ctions on Neural Networks 6( 5) , 1212-1228.161. Pedrycz, W . (1993) . Fuzzy Control and Fuzzy Sy st em s. 2nd edn. , Research
Studies Press, Wil ey, Taunton.162. Pineda, F . J . (1987). Generalisation of back-propagation to recurrent neural
networks . Physical Review Lett ers 59(19), 2229-2232.
References 315
163. Poggio, T ., V . Torre and C. Koch (1985) . Computational vision and regularisation theory. Nat ure 317(26), 314-319.
164. Priestley, M. B. (1988) . Nonlin ear and Nonstationary Time Series Analysis.Academic Press , London.
165. Rao, B. and H. F . Durrant-Whyte (1993). A decentralised Bayesian algorithmfor identification of tracked targets. IEEE Tran sactions on Sy st ems, Man andCybern etics 23(6) , 1683-1698.
166. Rao, B., H. F . Durrant-Whyte and A. Sheen (1991 ). A fully decentralisedmulti-sensor system for tracking and surveillance. International Journal ofRobotics R esearch 12(1), 20-45.
167. Roecker , J . A. and C. D. McGillem (1988) . Comparison of two-senso r trackingmethods based on state vector fusion and measurement fusion . IEEE Transact ions on Aerospace and Electronic Systems 24(4) , 447-449.
168. Saha, R . K. (1996) . Track-to-track fusion with diss imilar sensors . IEEE Transactions on Aerospace and Electronic Systems 32(3) , 1021-1029.
169. Saha, R. K. and K. C . Chang (1998) . An efficient algor it hm for mu lt isensor track fusion. IEEE Tmnsactions on Aerospace and Electronic Systems34(1), 200-210.
170. Saitoh , S. (1988). Theory of Reproducing K ern els and its and Applications.Longman, Harlow, Essex.
171. Sakamoto , Y., M. Ishgur i and G . Kitagawa (1986) . A kaik e Inform ation Criterion Statistics. D. Reidel, Tokyo :
172. Sanner, R. and J . Slotine (1992) . Gaussian networks for direct adaptive control. IEEE Transactions on Neural Networks 3(6) , 837-863.
173. Sastry, S. and A. Isidori (1989) . Adaptive control of linearisab le systems. IEEETransactions on Automatic Control 34, 1123- 1131.
174. Scholkopf, B., P. Barlett and A. J. Smola (1999) . Shrinking the data model:a new support vector regression algorithm. In : Advan ces in Neural Inf orm ation Processing Systems (M. S. Kearns and S.A. Solla, Eds.). MIT P ress ,Cambridge, Mass.
175. Sinha, N. K. and M. M. Gupta (Eds.) (2000) . Soft Computing & IntelligentSystems: Theory & and Applicat ions. Academic Press.
176. Sjoberg , J ., Q . Zhang and L. Ljung (1995) . Nonlinear black-box modelling insystem ident ificat ion: a unified overview. Automatica 31(12), 1691- 1724.
177. Sjoberg, J ., Q . Zhang and L. Ljung (1995) . Nonlinear black-box modelling insystem identification: mathematical foundations . Automatica 31 (12) , 17251752.
178. Shewchuk, J . R. (1994) . An introduct ion to conjugate gradient method without the agonizing pain. Tech . Report CMU-CS-94, Carnegie Mellon Unive rs ity.
179. Slotine , J . and W . Li (1991) . Applied Nonlinear Control. Prent ice Hall , Englewoo d Cliffs, NJ.
180. Smola, A. J . (1998). Learn ing with kerne ls. PhD thesis. Informat ik der Technischen Universitat Berlin .
181. Soderstrom, T . and P. Stoica (1989). Sys tem Identifi cation. Prentice Hall .182. Sontag, E . (1997) . Recurrent neural networks: some systems-theoretic aspects.
In: Dealing with Complexity: a Neural Network Approach (M . Karny, K. Warwick and V. Kurkova , Eds.) . pp. 1- 12. Springer-Verlag, London.
183. Stone, C. J ., M. H. Hansen , C. Kooperberg and Y. K. Truong (1997) . 1994Wald Memorial Lecture: Polynomial sp lines and their tensor products in extended linear modeling . Th e Annals of Statistics 25 (4), 1371-1470.
184. Sun, C. (1994) . Ru le based structure identification in an adapt ive networkbased inference sys te m. IEEE Tran sactions on Fuzzy Systems 2(1), 64-73.
316 References
185. Takagi , T . and M. Sugeno (1985) . Fuzzy identification of sys tems and itsapplica t ions to mod elling and control. IEEE Transactions on System s, Manand Cybern etics 15 , 116-1 32.
186. Takens, F . (1981). Det ecting strange at t rac to rs in turbulence. In : DynamicalSy st em s and Turbu lence (D. A. Rand and L. S. Young, Eds.) . Lectu re Notesin Mathematics 898, pp . 336-381, Springer-Verlag, Berlin.
187. Tolle, H. and E. Ersu (1992) . Neuroconirol: Learning Control Sy st ems Inspired by Neuronal Archit ectures and Human Problem Solving. Lecture Not esin Control and Information Sciences, 172 , Springer-Verlag , Berlin.
188. Tong, R. M. (1977). A control enginee ring review of fuzzy systems. Aut om atica13 , 559-569.
189. Tr an, T . and C. J . Harris (1999). Maritime avoidance navigation, totally integrated systems (MANTIS) . In : Proc, IEEE Fusion 99. Vol. 2. pp . 394- 401.
190. Vapnik, V . (1995) . Th e Nature of S tatis tical Learning Th eory. Springer-Verlag,New York .
191. Vapnik, V. (1998) . Statistical Learn ing Th eory : Adaptive Learning Sy st em sfor Signal Processing , Communicat ion and Control. Wiley, Chicheste r.
192. Verbruggen, H. B. and R. Babuska (1999). Fuzzy Logic Advances in and Applications. Series in Robotics and Intelligent Systems , Vo1.23, World Scientific,Singapore.
193. Vidyasagar, M. (1993). Nonlin ear Syst ems Analysi s. P rentice Hall , SaddleRiver, NJ .
194. Wahba , G . (1990) . Spline Models for Observat ional Data. Society for Industrialand Applied Mathematics, Princeton, RI.
195. Wang, H., G . P. Liu , C . J. Harris and M. Brown (1995). Advan ced AdaptiveControl. Pergamon P ress, Oxford .
196. Wang, H., M. Brown and C. J . Harris (1996) . Mod ellin g and control of nonlinear, operating point dep endent syste ms via associative memory networks.Journal of Dynamics and Control 6 , 199-218.
197. Wang, L. X. (1997) . A Cours e in Fuzzy Systems and Control. Prentice Hall ,Englewood Cliffs, NJ .
198. Wang, L.X. (1994) . Adaptive Fuzzy Sy st em s and Control: Design and Analysis.Prentice Hall , Englewood Cliffs, NJ.
199. Webb, A. (1993). An approach to nonlinear principal components an alysisusing radial basis functions. Defence Research Agency Malvern Memorandum.
200. Webb, A. (1999) . Statistical Pattern Recognition. Arnold , London.201. Werbos, P. J . (1990) . Backpropagation through time: What it does and how
to do it . Proc. of the IEEE 78 (10), 1550-1560.202. Werntges , H. W . (1993). Partitions of unity improve neural functional approxi
mation. In : Proc. IEEE Intern ational Confere nce Neural Networks. Vol. 2. SanFrancisco, USA . 914-918.
203. Williams, R. J . and D. Zipser (1989) . A learning algorithm for continuallyrunning fully recurrent neural networks. Neural Computation 1(2) , 270- 280.
204. Williams, R. J . and D. Zipser (1995) . Gradient-based learn ing algorithms forrecurrent networks and t heir computational comp lexity. In: Back-Propagation:Th eory, Architectures and Applications (Y. Chauvin and D. E . Rumelhart ,Eds.) , Erlbaum , Hillsdale, N.J .
205. Wi lliams, R. J . and J . Peng (1990) . An efficient gradient- based algor ithmfor on-line training of recur rent network trajectories . Neural Computation2 (4) , 490- 501.
206. Wolke nhauer , O. (1998) . Pos sibility Th eory with Application to Data Analysis.Research Studies Press, Wiley.
References 317
207. Wu, Z.-Q. and C. J. Harris (1997) . A neurofuzzy network structure for modelling and state estimation of unknown nonlinear systems. International Journal of Systems Science 28, 335-345.
208. Xu, L. and M. I. Jordan (1996) . On convergence properties of the EM algorithm for Gaussian mixtures. Neural Computation 8, 129-151.
209. Yeslidirek, A. and F. L. Lewis (1995) . Feedback linearisation using neuralnetworks. Automatica 31 , 1659-1664.
210. Zhang, Q. (1993) . Regressor selection and wavelet network construction. Technical Report, IRISA, Compus unversitaire de Beaulieu, 35042 Rennes.
211. Zheng, G . L. and S. A. Billings (2001). Effects of overparameterisation in nonlinear system identification and neural networks. Submitted to InternationalJournal of System Science.
212. Zimmermann, H. J . (1996) . Fuzzy Set Theory and its and Applications. KluwerAcademic, Boston.
Index
adaptivefuzzy systems, 72learning rates, 70modelling history, 21networks based fuzzy inferencesyst ems (ANFIS), 155spline modelling algorithm(ASMOD), 107, 226
addit iveneurofuzzy modellingalgorit hms, 106splines, 40
Akaike informati on crite rion(AIC) , 144algebraic sum operator, 85analysis of var iance (ANOVA),18, 106a posteriori prob abili ty, 32A-optimality
design criterion, 145neurofuzzy mod el construc tion(NeuDec) algorit hm, 143
approximat ion err or, 9autocorre lat ion matrix, 56a -cut , 77
backpropagation algorithm, 214Barycentric coordinates, 204basis functi ons, 14
expa nsion, global, localand radi al, 30
batch learning laws, 58Bayesian regularisati on , 129
Bayes theorem, 32, 129bias-vari ance dilemma, 33bivariate
Bernst ein polynomial , 215de Cast eljau algorit hm, 215
B-splines, 30, 41memb ership function, 79
Bezier- Bernsteinpolynomial models, 18, 41mod elling network , 209model construct ionalgorit hm, 219t riangle patch, 216
canonical state spacerepresent ations, 19centre of grav ity defuzzification , 90compact support sets, 14, 77condit ion numb er , 15, 36, 57corre lat ion test s, 45cross valid ation , 44, 105curse of dim ensionality, 17, 73, 156cyclic mod elling, 7
dataacquisit ion, 2preprocessing, 3
de Caste ljau algorithm, 212decentralised dat a fusion , 256
architecture , 262, 265defuzzification , 90
mean of maxima, 90
320 Index
cent re of gravity, 90defuzzifier , 75Delaunay
input space par ti ti oning, 204partitioning modelling, 201simplex, 203trian gulation, 18, 202
dir ect state est imation method , 169divide and conquer principle, 18
Ein st ein 's principle of simplicity, 17empirical
risk, 5minimisation principle (ERM), 10
loss function , 33err or
bar , 132sur face, 55
est imat ion error, 10expectat ion-maximisation (EM)algorit hm, 226, 237, 237exte nded
addit ive neurofuzzy mod el,119, 120
barycentric coordinates, 217
feature space , 284feedback linearisation , 227
neurofuzzy models, 239par ametric feedback form , 241
feedforward Gram-8chimidt OL8pro cedure, 191final prediction error , 43, 144first ord er infinite splines, 299forward const rained regression(FeR) , 177functional equivalence of outputsaugmente d fusion (OAF) andopt imal weighting measurementfusion (OWMF) , 260fuzzification , 89fuzzifier , 75fuzzy
boxtrees, 161dat a clustering algorit hm, 155implicati on, 85intersection , 83memb ership funct ion , 75modelling, 71operators, 83relati on sur face, 87set , 75support , 77systems, 74un ion , 84variable, 78
Gaussianmembership functions, 82radial basis function (RBF) ,41, 285
generallearning laws, 58recurrent neur al network, 243
generalisati on , 1, 13, 72generalised linear models, 6global basis functi ons, 30gradient
descent algorithms, 59, 237noise, 68
Gr am-8chimidt OL8 algorithm, 193grey box models, 4group method of data handling(GMDH) ,104growing and pruning modelconstruction, 107Gustafson-Kessel algorit hm, 155
Hessian matrix, 59, 123hierarchical
multisensor dat a fusion , 266neurofuzzy model, 125
hinging hyp erpl anes, 30hybrid
hierarchical multi sensor dat afusion architecture, 20
learning scheme, 236, 239te nsor / addit ive splines, 41
hypothesis testing, 6, 43
ill-posed problems, 5, 143"incomplete-data" log likelihood,238
independ ent component analysis(ICA), 17ind ex vector , 146indirect state est imationmethods, 169inference engine, 75inferencing, 88informat ion
filter, 249measures, 43
inst ant aneous learning laws, 54, 61intelligent or adaptive modelling , 7input-output models, 26introduction to modelling, 1inverse pro cedure of the deCasteljau algorithm, 209iterative
neurofuzzy model const ruction, 104(cyclic) , 7
Kalman filter , 20, 169, 248, 255Karush-Kuhn-Tucker (KKT)algorit hm, 288kernel functions , 6, 30, 281, 284k-d trees, 17knot insert ion/ removal, 109knowledge base , 74
lat t ice based associative networks,15learning t heory, 9
laws, 53, 58least mean squares , 61
likelihood function , 32, 129linear smoother , 36
Index 321
linguistic vagueness , 71local
basi s functi ons, 30basis funct ion expa nsion(LBF E) , 230linear model t ree (LOLIMOT)algorithm, 159model,136neurofuzzy modelling, 153regulari sed neur ofuzzy
loss functions, 11 , 284
Mamdani model, 155maximum
a posteriori (MAP) est imate ,32
likelihood (ML) est imate,33, 54, 236
likelihood (ML) identifi cation, 129max-NLMS , 64mean
of maxima (MOM), 90squared error , 27
measurement fusionmethods, 255, 258minimal capt ure zone, 66, 69minimum description lengt h, 44mixture of experts
algorithm, 18, 227modelling, 173, 204, 236
modelapproximation erro r, 9bias , 34construction algorithm, 143cross validat ion, 44, 105identificat ion , 3parameter est imatio n, 31parsimony, 34quality, 33regressor order, 14selection criteria sensit ivity, 44selection methods, 42tran sparency, 3, 73st ructural regular isat ion , 34
322 Index
valid ity tests, 46varia nce , 34
modi fied ada pt ive spline modelling(MA8MOD) algorit hm, 244mul t i-dim ensional kernels, 286mult ilayer perceptron (MLP) , 55,41mult isensor dat a fusion , 20, 255
neurofuzzyfeedback linea risation, 239Kalm an filtering, 173local function basis expa nsionmodels, 230local linearis ation (NFLL) , 225modelling, 71models, 91network, 94condit ioning, 15, 36, 57
regulari sed model, 129state esti mators , 245systems, 72
Newtonalgorit hm, 122- Raphson algorithm, 59
nonli nearaffine syste ms, 240autoregress ive models(NARX), 28auto regress ive movingaverage models (NARMAX) , 28finit e impulse responsemodels, 28output error (NOE)model, 28regression approximat ion, 288
normal fuzzy set, 77normalised
condition numbers, 68least mean squ ares (NLM8)weight convergence , 63
learning , 62, 166, 170
operati ng point neuro fuzzymodel, 164optimal weight ing measurementfusion (OWMF) , 259ort hogonal least squares (OL8)algorithm, 143, 177over-determined learning, 13Occam 's razor , 15output
augmented fusion (OAF), 259feedb ack linearisation , 19, 241measurement fusion , 20
par ametric-pure-feedback form , 241parsimonious
neurofuzzy modelling, 103par allel modelling algorit hm, 183
partition of unity, 78polynomial
kernels, 41, 286Bezier-Bern stein , 41, 215B-splines, 30, 41, 286
principal component analysis(PCA), 17, 144priors for neuro fuzzy model, 133problem
computational, 282conceptual , 282
quad- trees models, 17
radi albasis functi ons (RBF), 30const ruction, 30
recurrent neur al network , 242recursive least squares est imat ion,67, 166regulari sat ion , 5, 36
Bayesian , 129function , 283networks, 39techniques, 106, 143
regul arised neurofuzzy model, 129reg ular iser coefficient , 35regression mat rix , 27reproducing kernels, 39
Hilb ert spaces (RKHS) , 6, 39, 281rid ge const ruct ion, 30Robbins- Munro stochasticapproximation algorit hm, 70rul e
confidence , 86completeness , 78
sgn-NLMS, 65sigmoidal neural networks, 30simplexes, 18, 203
Delaunay, 203singular value decomposition(SVD) , 143slack varia bles, 287splines , 41, 286
te nsor/ addit ive, 41st agewise const ruction, 110state
dir ect estimation method , 169feedback linearisation , 240indirect state estimationmethods, 169-sp ace models, 26-space representations ofneurofuzzy mod els, 168vector fusion methods, 255, 263vector assimil ati on fusion(SVAF ),263vector fusion , 20
stat ist icallearn ing theory, 10significance metrics, 105
steepes t descent, 60st rict feedback syste m, 241st ructural
risk minimisation , 11,226, 230, 245, 282
regularisat ion , 5
Index 323
support vectors, 288analysis of var iance(SUPANOVA) algorit hm,20, 297, 298
machine, 6, 282models, 281, 286, 289neural networks, 39neurofuzzy network
S-norm, 85
Takagi-Sugeno (T- S) mod elsfuzzy ty pe, 18local neurofuzzy mod el,95, 155, 181, 225
target tracking, 20tensor
multiplication/tensorsubmodel split t ing, 108product , 30, 286product splines , 40
track-t o-track fusion , 264t ransparency, 3, 73T sukam oto model, 155T-norm, 83
under-d etermined learning, 13unimodal fuzzy sets, 76univariate Bezier-Bernst einpolynomials, 209
valid ation /verificati on , 3Vapnik-Cherr onenki s (VC) , 11very fast simulated reannealing(VFSR) algorit hm, 206
wavelet s, 30weighting
fun cti on , 204identificati on , 122
Wiener and Hammerst ein mod els,28