Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning)
El análisis de estructuras biológicamente relevantes trae consigo diferentes problemas representativos, debido a los cambios que pueden surgir en un paciente a la hora de realizar procedimientos médicos necesarios para la determinación de algún tipo de anormalidad corporal, como puede ser los cambio...
- Autores:
-
Velásquez Minoli, Juan Pablo
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2022
- Institución:
- Universidad Tecnológica de Pereira
- Repositorio:
- Repositorio Institucional UTP
- Idioma:
- OAI Identifier:
- oai:repositorio.utp.edu.co:11059/14267
- Acceso en línea:
- https://hdl.handle.net/11059/14267
https://repositorio.utp.edu.co/home
- Palabra clave:
- 620 - Ingeniería y operaciones afines::629 - Otras ramas de la ingeniería
Inference (logic)
Shape correspondence
Brain structure
Unsupervised learning
Variational inference
Multiview learning
- Rights
- openAccess
- License
- Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0)
| id |
UTP2_647fe269492adbc6133b3e2dcfe281a2 |
|---|---|
| oai_identifier_str |
oai:repositorio.utp.edu.co:11059/14267 |
| network_acronym_str |
UTP2 |
| network_name_str |
Repositorio Institucional UTP |
| repository_id_str |
|
| dc.title.spa.fl_str_mv |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| title |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| spellingShingle |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) 620 - Ingeniería y operaciones afines::629 - Otras ramas de la ingeniería Inference (logic) Shape correspondence Brain structure Unsupervised learning Variational inference Multiview learning |
| title_short |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| title_full |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| title_fullStr |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| title_full_unstemmed |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| title_sort |
Análisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning) |
| dc.creator.fl_str_mv |
Velásquez Minoli, Juan Pablo |
| dc.contributor.advisor.none.fl_str_mv |
García Arias, Hernan Felipe |
| dc.contributor.author.none.fl_str_mv |
Velásquez Minoli, Juan Pablo |
| dc.subject.ddc.none.fl_str_mv |
620 - Ingeniería y operaciones afines::629 - Otras ramas de la ingeniería |
| topic |
620 - Ingeniería y operaciones afines::629 - Otras ramas de la ingeniería Inference (logic) Shape correspondence Brain structure Unsupervised learning Variational inference Multiview learning |
| dc.subject.other.none.fl_str_mv |
Inference (logic) Shape correspondence Brain structure |
| dc.subject.proposal.eng.fl_str_mv |
Unsupervised learning Variational inference Multiview learning |
| description |
El análisis de estructuras biológicamente relevantes trae consigo diferentes problemas representativos, debido a los cambios que pueden surgir en un paciente a la hora de realizar procedimientos médicos necesarios para la determinación de algún tipo de anormalidad corporal, como puede ser los cambios en la respiración que llevan al mismo tiempo un aumento o disminución en la frecuencia cardíaca, también, cambios en el humor del paciente pueden generar ansiedad o algún otro tipo de anomalía, resultando así en múltiples estados en estructuras corporales complejas. Encontrar correspondencias entre diferentes formas complejas proveniente de mallas en 3D suele ser una tarea engorrosa, ya que no siempre es posible encontrar medidas de similaridad entre elementos en imágenes de resonancia magnética u otro estudio médico. Dado que encontrar correspondencias en estructuras biológicas aleatoriamente cambiantes de acuerdo a las circunstancias del sujeto, en este proyecto se propone una metodología para describir estructuras cerebrales basada en variational multiview unsupervised learning, el cuál al ser un aprendizaje no supervisado permite la búsqueda de variabilidad similar entre elementos no rígidos a pesar de que ocurran algunos cambios en ellos, también se incluye el uso de múltiples vistas como una estrategia para mejorar la forma en la que se observa y trata la información proveniente de imágenes de resonancia magnética facilitando así la extracción de características representativas y finalmente se aplica inferencia variacional, la cual permite el manejo de modelos Bayesianos intratables computacionalmente usualmente requeridos en la creación de variables latentes, al inferir funciones de densidad de probabilidad simples sobre estos, aumentando la velocidad de convergencia del mismo y su grado de acierto. Los resultados muestran como los modelos propuestos logran capturar en gran medida las no linealidades presentes en formas 3D no rígidas, incluso cuando estas presentan oclusión o par cialidades, demostrando así ser técnicas viables para el manejo de correspondencia de formas. |
| publishDate |
2022 |
| dc.date.accessioned.none.fl_str_mv |
2022-09-15T16:13:57Z |
| dc.date.available.none.fl_str_mv |
2022-09-15T16:13:57Z |
| dc.date.issued.none.fl_str_mv |
2022 |
| dc.type.none.fl_str_mv |
Trabajo de grado - Pregrado |
| dc.type.version.none.fl_str_mv |
info:eu-repo/semantics/acceptedVersion |
| dc.type.coar.none.fl_str_mv |
http://purl.org/coar/resource_type/c_7a1f |
| dc.type.content.none.fl_str_mv |
Text |
| dc.type.driver.none.fl_str_mv |
info:eu-repo/semantics/bachelorThesis |
| format |
http://purl.org/coar/resource_type/c_7a1f |
| status_str |
acceptedVersion |
| dc.identifier.uri.none.fl_str_mv |
https://hdl.handle.net/11059/14267 |
| dc.identifier.instname.none.fl_str_mv |
Universidad Tecnológica de Pereira |
| dc.identifier.reponame.none.fl_str_mv |
Repositorio Institucional Universidad Tecnológica de Pereira |
| dc.identifier.repourl.none.fl_str_mv |
https://repositorio.utp.edu.co/home |
| url |
https://hdl.handle.net/11059/14267 https://repositorio.utp.edu.co/home |
| identifier_str_mv |
Universidad Tecnológica de Pereira Repositorio Institucional Universidad Tecnológica de Pereira |
| dc.relation.references.none.fl_str_mv |
A. Damianou, C. H. Ek, M. K. Titsias, and N. D. Lawrence, “Manifold relevance determination,” in Proceedings of the International Conference in Machine Learning, J. Langford and J. Pineau, Eds., vol. 29. San Francisco, CA: Morgan Kauffman, 00 2012. [Online]. Available: http://inverseprobability.com/publications/damianou-manifold12.html M. Ovsjanikov, M. Ben-Chen, J. Solomon, A. Butscher, and L. Guibas., “Fun ctional maps: A flexible representation of maps between shapes.” ACM Trans. Graph., 2012 V.-M. R. L.-R. J. M.-S. J. L.-M. F. N. V. B. . Legaz-Aparicio, A.-G., “Efficient variational approach to multimodal registration of anatomical and functional intra-patient tumorous brain data,” Int. J. Neural Syst., 2017. P.-M. Olson, L.D., “Localization of epileptic foci using multimodality neuro imaging,” 2013. M. Talo, O. Yildirim, U. B. Baloglu, G. Aydin, and U. R. Acharya, “Con volutional neural networks for multi-class brain disease detection using mri images,” Computerized Medical Imaging and Graphics, vol. 78, dec 2019. H. Hardy, L. Brynildson, and B. Bronson, “Computer rendering of stereo tactic atlas data with whole brain mapping with computed tomography and magnetic resonance imaging,” Computers in Stereotactic Neurosurgery, pp. 109–133, 1992 Y. Katayama, H. Oshima, T. Kano, K. Kobayashi, C. Fukaya, and T. Yama moto, “Direct effect of subthalamic nucleus stimulation on levodopa-induced peak-dose dyskinesia in patients with parkinson’s disease,” Stereotactic and Functional Neurosurgery, vol. 84, pp. 176–179, aug 2006. G. Schaltenbrand and P. Bailey, “Introduction to stereotaxis with an atlas of the human brain,” Georg Thieme, 1959. M. Yoshida, “Three-dimensional maps by interpolation from the schalten brand and bailey atlas,” Computers in Stereotactic Neurosurgery, pp. 143– 152, 1992. A. H. A. Shon, K. Grochow and R. Rao, “Learning shared latent structure for image synthesis and robotic imitation,” Advances in Neural Information Processing Systems, vol. 18, no. 1233, 2006 R. J. T. P. R.-G. Ek, Carl Henrik and N. Lawrence, “Ambiguity modeling in latent spaces,” Machine Learning and Multimodal Interaction, 2008. R. T. Angela Serra, Paola Galdi, Artificial Intelligence in the Age of Neural Networks and Brain Computing. ACADEMIC PRESS, 2019, ch. Chapter 13 - Multiview Learning in Biomedical Applications. G. Sfikas and C. Nikou, “Bayesian multiview manifold learning applied to hippocampus shape and clinical score data,” in Medical Computer Vision and Bayesian and Graphical Models for Biomedical Imaging. Springer Interna tional Publishing, 2017, pp. 160–171. C. X. Chang Xu, Dacheng Tao, “A survey on multi-view learning,” ArXiv, apr 2013. H. K. S. M. Cheng Zhang, Judith Butepage, “Advances in variational infe rence,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, pp. 2008–2026, dec 2018. T. B. M. Michael Salter-Townshend, “Variational bayesian inference for the latent position cluster model for network data,” Computational Statistics & Data Analysis, vol. 57, pp. 661–671, jan 2013 M. G. b. P. L. a. M. L. a. GuoJun Liu a, Yang Liu a, “Variational inference with gaussian mixture model and householder flow,” Neural Networks, vol. 109, pp. 43–55, jan 2019. D. M.Pavlovic, B. R.L.Guillaume, E. K.Towlson, N. M.Y.Kuek, S. Afyouni, ´ P. E.Vértes, B. Yeo, E. T.Bullmore, and T. E.Nichols, “Multi-subject stochas tic blockmodels for adaptive analysis of individual differences in human brain network cluster structure,” NeuroImage, 2020. C. R. Madan, “Advances in studying brain morphology: The benefits of open access data,” Frontier in Human Neuroscience, 2017. T. Shiohama, J. Levman, N. Baumer, and E. Takahashi, “Structural magnetic resonance imaging-based brain morphology study in infants and toddlers with down syndrome: The effect of comorbidities,” Pediatric Neurobiology, pp. 67–73, 2019. J. Corps and I. Rekik, “Morphological brain age prediction using multi-view brain networks derived from cortical morphology in healthy and disordered participants,” Scientific Reports, 2019. P. M. C. F. B. Alberto Llera, Thomas Wolfers, “Inter-individual differences in human brain structure and morphology link to variation in demographics and behavior,” eLife, 2019. V. Chouvatut and E. Boonchieng, “Brain tumor’s approximate corresponden ce and area with interior holes filled,” in 2017 14th International Joint Con ference on Computer Science and Software Engineering (JCSSE), 2017, pp. 1–5. J. V. Buren and R. Borke, “Variations and connections of the human thala mus,” in Springer Berlin, 1 E. Sirnes, L. Oltedal, H. Bartsch, G. E. Eide, I. B. Elgen, and S. M. Aukland, “Brain morphology in school-aged children with prenatal opioid exposure: A structural mri study,” Early Human Development, pp. 33–39, 2017. O. van Kaick Hao Zhang Ghassan Hamarneh Daniel Cohen-Or, “A survey on shape correspondence,” Computer Graphics Forum, jul 2011. Z. Lahner, M. Vestner, A. Boyarski, O. Litany, R. Slossberg, T. Remez, E. Ro dolà, A. Bronstein, M. Bronstein, R. Kimmel, and D. Cremers, “Efficient deformable shape correspondence via kernel matching,” arXiv, Cornell Uni versity, sep 2017 J. Krüger, S. Schultz, and H. Handels, “Registration with probabilistic corres pondences – accurate and robust registration for pathological and inhomoge neous medical data,” Computer Vision and Image Understanding, 2019. M. Ovsjanikov, Handbook of Numerical Analysis. LIX, Ecole Polytechni que, CNRS, Palaiseau, France, 2018, ch. T. R. Song Wang, Brent Munsell, Statistical shape and Deformation Analysis. Elsevier Ltd., 2017, ch. 3. H. F. García, . A. Orozco, and M. A. Álvarez, “Nonlinear probabilistic latent variable models for groupwise correspondence analysis in brain structures,” in 2018 IEEE 28th International Workshop on Machine Learning for Signal Processing (MLSP), 2018, pp. 1–6. D. Boscaini, J. Masci, E. Rodolà, and M. M. Bronstein, “Learning shape co rrespondence with anisotropic convolutional neural networks,” arXiv, Cornell University, may 2016. T. B. Alan Brunton, Augusto Salazar and StefanieWuhrer, “Review of statis tical shape spaces for 3d data with comparative analysis for human faces,” Computer Vision and Image Understanding, vol. 128, pp. 1–6, 2014. U. S. M. Aubry and D. Cremers, “The wave kernel signature: A quantum mechanical approach to shape analysis,” In Proc. 4DMOD, 2011. M. M. Bronstein and I. Kokkinos, “Scale-invariant heat kernel signatures for non-rigid shape recognition,” In Proc. CVPR, 2010. R. M. Rustamov, “Laplace-beltrami eigenfunctions for deformation invariant shape representation,” In Proc. SGP, 2007. Y. Aflalo, A. Bronstein, and R. Kimmel, “On convex relaxation of graph iso morphism,” PNAS, 2015. M. M. B. A. M. Bronstein and R. Kimmel, “Generalized multidimensional scaling: a framework for isometryinvariant partial surface matching,” PNAS,, 2006. Q. Chen and V. Koltun., “Robust nonrigid registration by convex optimiza tion,” In Proc. ICCV, 2015. I. Kezurer, S. Z. Kovalsky, R. Basri, , and Y. Lipman, “Tight relaxation of quadratic matching,” In Computer Graphics Forum, 2015. K. A. Sidorov, S. Richmond, and D. Marshall, “Efficient groupwise non-rigid registration of textured surfaces,” in Proceedings of the 2011 IEEE Conferen ce on Computer Vision and Pattern Recognition, Washington, DC, pp. 2401– 2408, 2011. T. Iwata, T. Hirao, and N. Ueda, “Probabilistic latent variable models for unsupervised many-to-many object matching,” Information Processing and Management, pp. 682–697, 2016. J. Masci, D. Boscaini, M. M. Bronstein, and P. Vandergheynst, “Geodesic convolutional neural networks on riemannian manifolds,” In Proc. 3dRR,, pp. 682–697, 2015. D. C. E. V. L. Wei, Q. Huang and H. Li, “Dense human body correspondences using convolutional networks,” In Proc. CVPR, 2016. A. K. Z. Wu, S. Song, “3d shapenets: A deep representation for volumetric shapes,” In Proc. CVPR, pp. 682–697, 2015. L. Z. F. a. I. Z. H. X. J. W. R. . E. L. G. S. P. a. X. H. R. W.-R. Tuo Zhang Methodology; Conceptualization; Writing-Original draft preparation, Ying Huang Formal analysis; Investigation and T. L. C. Editing, “Identifying cross individual correspondences of 3-hinge gyri,” Medical Image Analysis, 2020. J. A. Claudia Blaiottaa, M. Jorge Cardosob, “Variational inference for me dical image segmentation,” Computer Vision and Image Understanding, vol. 151, pp. 14–28, oct 2016. Z. G. Tomoharu Iwata, David duvenaud, “Warped mixtures for nonparametric cluster shapes,” in UAI’13: Proceedings of the Twenty-Ninth Conference on Uncertainty in Artificial Intelligence, aug 2013, pp. 311–320. M. Cootes T. F., Taylor C.J., “Statistical models of appearance for computer vision,” 2004. X. X. S. S. Jing Zhao, Xijiong Xie, “Multi-view learning overview: Recent progress and new challenges,” Information Fusion, vol. 38, pp. 43–54, feb 2017. N. Lawrence, “Gaussian process latent variable models for visualisation of high dimensional data,” Advances in Neural Information Processing Systems, vol. 16, pp. 329–336, 2004 Q. L. Zhenglong Li and H. Lu, “A variational multi-view learning framework and its application to image segmentation,” IEEE International Conference on Multimedia and Expo, pp. 1516–1519, jul 2009. L. H. Changde Du, Changying Du and H. He, “Reconstructing perceived images from human brain activities with bayesian deep multiview learning,” IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 8, pp. 2310–2323, aug 2019. Y. Q. C. L. X. L. Bingqian Lin, Yuan Xie, “Jointly deep multi-view learning for clustering analysis,” ArXiv, nov 2018. Z. C. Liang Zhao and Z. J. Wang, “Unsupervised multiview nonnegative co rrelated feature learning for data clustering,” IEEE Signal Processing Letters, vol. 25, no. 1, pp. 60–64, nov 2017. R. R. Shashini De Silva, Jinsub Kim, “Unsupervised multiview learning with partial distribution information,” in 2017 IEEE 27th International Workshop on Machine Learning for Signal Processing (MLSP), 2017, pp. 1–6. P. H. Y. S. Xu Wang, Dezhong Peng, “Adversarial correlated autoencoder for unsupervised multi-view representation learning,” Knowledge-Based Sys tems, vol. 168, pp. 109–120, mar 2019. D. Z. Wentao Fan, Nizar Bouguila, “Unsupervised hybrid feature extraction selection for high-dimensional non-gaussian data clustering with variational inference,” IEEE Transactions on Knowledge and Data Engineering, vol. 25, no. 7, pp. 1670–1685, may 2013 X. C. P. A. W.-C. L. Tien Thanh Nguyen, Thi Thu Thuy Nguyen, “A novel combining classifier method based on variational inference,” Pattern Recog nition, vol. 49, pp. 198–212, jan 2016. N. B. Hieu Nguyen, Muhammad Azam, “Data clustering using variational learning of finite scaled dirichlet mixture models,” in 2019 IEEE 28th Inter national Symposium on Industrial Electronics (ISIE), 2019, pp. 1391–1 R. K. N. W. C. B. D. Z. L. M. M. B. A. L. Klaus Greff, Raphaël Lopez Kauf man, “Multi-object representation learning with iterative variational inferen ce,” ArXiv, may 2019 M. A. l. Hernán Darío Vargas Cardona, Álvaro Ángel Orozco, “Unsupervi sed learning applied in mer and ecg signals through gaussians mixtures with the expectation-maximization algorithm and variational bayesian inference,” 2013 35th Annual International Conference of the IEEE Engineering in Me dicine and Biology Society (EMBC), pp. 4326–4329, jul 2013 Q. Zheng, A. Sharf, A. Tagliasacchi, H. Z. Baoquan Chen, A. Sheffer, and D. C. Or., “Consensus skeleton for non-rigid space-time registration.” Com puter Graphcis Forum (Special Issue of Eurographics), vol. 29, no. 2, pp. 635–644, 2010. A. A. Salama, R. A. Alarabawy, W. El-shehaby, D. El-amrousy, M. S. Bagh dadi, and M. F. Rizkallah., “Consensus skeleton for non-rigid space-time re gistration.” The Egyptian Journal of Radiology and Nucle F. Kanavati, T. Tong, K. Misawa, K. M. Michitaka Fujiwara, D. Rueckert, , and B. Glocker., “Supervoxel classification forests for estimating pairwise image correspondences.” Pattern Recognition., vol. 63, pp. 561–569, 2017. M. Cabezas, A. Oliver, X. Lladó, J. Freixenet, and M. B. Cuadra., “A review of atlas-based segmentation for magnetic resonance brain images.” Computer Methods and Programs in Biomedicine., vol. 104, no. 3, pp. 158–177, 2011. D. Aiger, N. J. Mitra, and D. Cohen-Or., “4pointss congruent sets for ro bust pairwise surface registration.” ACM SIGGRAPH 2008 Papers., pp. 84–1, 2008. R. Panda, S. Agrawal, M. Sahoo, and R. Nayak, “novel evolutionary rigid body docking algorithm for medical image registration.” Swarm and Evolu tionary Computation., vol. 33, pp. 108–118, 2017. C. P. Weingarten, M. H. Sundman, P. Hickey, and N. kuei Chen., “Neuroima ging of parkinson’s disease: Expanding views.” Neuroscience and Biobeha vioral Reviews., vol. 59, pp. 16–52, 2015 L. Wang and C. Pan., “Nonrigid medical image registration with locally linear reconstruction.” Neurocomputing., vol. 145, pp. 303–315, 2014. A. S. Lundervold and A. Lundervold., “An overview of deep learning in me dical imaging focusing on mri.” Zeitschrift für Medizinische Physik., vol. 145, pp. 102–127, 2019. C.-Y. Wee, C. Liu, A. Lee, J. S. Poh, H. Ji, and A. Qiu., “Cortical graph neural network for ad and mci diagnosis and transfer learning across populations.” NeuroImage: Clinical., vol. 23, 2019. K. A. Sidorov, S. Richmond, and D. Marshall., “Efficient groupwise non-rigid registration of textured surfaces.” In Proceedings of the 2011 IEEE Conferen ce on Computer Vision and Pattern Recognition., pp. 2401–2408, 20 M. M. Bronstein and I. Kokkinos, “Scale-invariant heat kernel signatures for non-rigid shape recognition,” In Proc. CVPR, 2010. K. Cutajar, E. V. Bonilla, P. Michiardi, and M. Filippone, “Random feature expansions for deep gaussian processes,” Proceedings of the 34th Internatio nal Conference on Machine Learning, vol. 70, pp. 884–893, 2017. A. M. Bronstein, M. M. Bronstein, L. J. Guibas, and M. Ovsjanikov, “Shape google: Geometric words and expressions for invariant shape retrieval,” ACM Trans. Graph., vol. 30, pp. 1–20, 2011. A. Rahimi and B. Recht, “Random features for largescale kernel machines,” In Neural Infomration Processing Systems., 2007. A. Corduneanu and C. M. Bishop, “Variational bayesian model selection for mixture distributions,” Artificial Intelligence and Statistics, 2001. K. V. G., L. Y., and funkhouser T. A., “Blended intrinsic maps.” TOG 30, 2011. A. M. Bronstein, M. M. Bronstein, and R. Kimmel, “Calculus of nonrigid surfaces for geometry and texture manipulation,” IEEE Trans. Vis. Comput. Graph., 2007. S. Y. and A. Y., “Coarse-to-fine com-binatorial matching for dense isometric shape correspondence.” Computer Graphics Forum, 2011. M. Vestner, Z. Lähner, A. Boyarski, O. Litany, R. Slossberg, T. Remez, E. Ro dola, A. Bronstein, M. Bronstein, R. Kimmel, and D. Cremers, “Efficient de formable shape correspondence via kernel matching.” 017 International Con ference on 3D Vision (3DV), pp. 517–526, Y. Aflalo, A. Dubrovina, and R. Kimmel., “Spectral generalized multidimen sional scaling.” Int. J. Comput. Vision, pp. 380–392, 2016. E. Rodolà, S. R. Bulò, T. Windheuser, M. Vestner, and D. Cremers., “Dense non-rigid shape correspondence using random forests.” Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, CVPR ’14, pp. 4177–4184, 2014. L. Cosmo, E. Rodolà, M. M. Bronstein, A. Torsello, D. Cremers, and Y. Sahi llio glu, “Shrec’16: Partial matching of deformable shapes,” Eurographics Workshop on 3D Object Retrieval, 2016. E. Rodola, L. Cosmo, M. M. Bronstein, A. Torsello, and D. Cremers., “Partial functional correspondence.” Computer Graphics Forum, 2016. I. S. Gousias, A. D. Edwards, M. A. Rutherford, S. J. Counsell, J. V. Hajnal, D. Rueckert, and A. Hammers., “Magnetic resonance imaging of the newborn brain: Manual segmentation of labelled atlases in term-born and preterm in fants.” NeuroImage, pp. 1499–1509, 201 A. C. Damianou, M. K. Titsias, and N. D. Lawrence, “Variational inference for latent variables and uncertain inputs in gaussian processes,” Journal of Machine Learning Research, 2016. P. WD, “Kl-divergences of normal, gamma, dirichlet and wishart densities,” University College, London, 2001. A. G. d. G. Matthews, M. van der Wilk, T. Nickson, K. Fujii, A. Boukouvalas, P. León-Villagrá, Z. Ghahramani, and J. Hensman, “GPflow: A Gaussian process library using TensorFlow,” Journal of Machine Learning Research, vol. 18, no. 40, pp. 1–6, apr 2017. [Online]. Available: http://jmlr.org/papers/v18/16-537.html J. Hensman, F. Nicolo, and N. D. Lawrence, “Gaussian processes for big data.” Uncertainty in Artificial Intelligence, 2013. M. Johnson, D. K. Duvenaud, A. Wiltschko, R. P. Adams, , and S. R. Dat ta., “Composing graphical models with neural networks for structured repre sentations and fast inference.” Advances in Neural Information Processing Systems, vol. 29, pp. 2946–2954, 2016. A. Hebbal, L. Brevault, M. Balesdent, E.-G. Talbi, and N. Melab., “Bayesian optimization using deep gaussian processes.” hal-02924230ff, 2020. |
| dc.rights.license.none.fl_str_mv |
Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0) |
| dc.rights.uri.none.fl_str_mv |
https://creativecommons.org/licenses/by-nc-nd/4.0/ |
| dc.rights.coar.none.fl_str_mv |
http://purl.org/coar/access_right/c_abf2 |
| dc.rights.accessrights.none.fl_str_mv |
info:eu-repo/semantics/openAccess |
| rights_invalid_str_mv |
Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0) https://creativecommons.org/licenses/by-nc-nd/4.0/ http://purl.org/coar/access_right/c_abf2 |
| eu_rights_str_mv |
openAccess |
| dc.format.extent.none.fl_str_mv |
89 Páginas |
| dc.format.mimetype.none.fl_str_mv |
application/pdf |
| dc.publisher.none.fl_str_mv |
Universidad Tecnológica de Pereira |
| dc.publisher.program.none.fl_str_mv |
Maestría en Ingeniería Eléctrica |
| dc.publisher.faculty.none.fl_str_mv |
Facultad de Ingenierías |
| dc.publisher.place.none.fl_str_mv |
Pereira |
| publisher.none.fl_str_mv |
Universidad Tecnológica de Pereira |
| institution |
Universidad Tecnológica de Pereira |
| bitstream.url.fl_str_mv |
https://dspace7-utp.metabuscador.org/bitstreams/85695f8e-9e74-4841-99f4-1488cfc0d090/download https://dspace7-utp.metabuscador.org/bitstreams/90452e11-b733-430f-a151-94159f0d98bd/download https://dspace7-utp.metabuscador.org/bitstreams/17df1aa1-b68b-4fe3-ae08-70f3a16038c1/download https://dspace7-utp.metabuscador.org/bitstreams/e92cd28c-37bc-424a-bbb1-f2f8698e2afb/download |
| bitstream.checksum.fl_str_mv |
4f19f4b4a0844163025fcb8cadd1ef85 2f9959eaf5b71fae44bbf9ec84150c7a f3455e61e5da9b76c168ab1f0fef031a 155e157c60bf79c1df340474bf7f26b7 |
| bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 |
| repository.name.fl_str_mv |
Repositorio de la Universidad Tecnológica de Pereira |
| repository.mail.fl_str_mv |
bdigital@metabiblioteca.com |
| _version_ |
1814021948949135360 |
| spelling |
Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0)Manifiesto (Manifestamos) en este documento la voluntad de autorizar a la Biblioteca Jorge Roa Martínez de la Universidad Tecnológica de Pereira la publicación en el Repositorio institucional (http://biblioteca.utp.edu.co), la versión electrónica de la OBRA titulada: ________________________________________________________________________________________________ ________________________________________________________________________________________________ ________________________________________________________________________________________________ La Universidad Tecnológica de Pereira, entidad académica sin ánimo de lucro, queda por lo tanto facultada para ejercer plenamente la autorización anteriormente descrita en su actividad ordinaria de investigación, docencia y publicación. La autorización otorgada se ajusta a lo que establece la Ley 23 de 1982. Con todo, en mi (nuestra) condición de autor (es) me (nos) reservo (reservamos) los derechos morales de la OBRA antes citada con arreglo al artículo 30 dehttps://creativecommons.org/licenses/by-nc-nd/4.0/http://purl.org/coar/access_right/c_abf2info:eu-repo/semantics/openAccessGarcía Arias, Hernan FelipeVelásquez Minoli, Juan Pablo2022-09-15T16:13:57Z2022-09-15T16:13:57Z2022https://hdl.handle.net/11059/14267Universidad Tecnológica de PereiraRepositorio Institucional Universidad Tecnológica de Pereirahttps://repositorio.utp.edu.co/homeEl análisis de estructuras biológicamente relevantes trae consigo diferentes problemas representativos, debido a los cambios que pueden surgir en un paciente a la hora de realizar procedimientos médicos necesarios para la determinación de algún tipo de anormalidad corporal, como puede ser los cambios en la respiración que llevan al mismo tiempo un aumento o disminución en la frecuencia cardíaca, también, cambios en el humor del paciente pueden generar ansiedad o algún otro tipo de anomalía, resultando así en múltiples estados en estructuras corporales complejas. Encontrar correspondencias entre diferentes formas complejas proveniente de mallas en 3D suele ser una tarea engorrosa, ya que no siempre es posible encontrar medidas de similaridad entre elementos en imágenes de resonancia magnética u otro estudio médico. Dado que encontrar correspondencias en estructuras biológicas aleatoriamente cambiantes de acuerdo a las circunstancias del sujeto, en este proyecto se propone una metodología para describir estructuras cerebrales basada en variational multiview unsupervised learning, el cuál al ser un aprendizaje no supervisado permite la búsqueda de variabilidad similar entre elementos no rígidos a pesar de que ocurran algunos cambios en ellos, también se incluye el uso de múltiples vistas como una estrategia para mejorar la forma en la que se observa y trata la información proveniente de imágenes de resonancia magnética facilitando así la extracción de características representativas y finalmente se aplica inferencia variacional, la cual permite el manejo de modelos Bayesianos intratables computacionalmente usualmente requeridos en la creación de variables latentes, al inferir funciones de densidad de probabilidad simples sobre estos, aumentando la velocidad de convergencia del mismo y su grado de acierto. Los resultados muestran como los modelos propuestos logran capturar en gran medida las no linealidades presentes en formas 3D no rígidas, incluso cuando estas presentan oclusión o par cialidades, demostrando así ser técnicas viables para el manejo de correspondencia de formas.Analysis of biologically relevant structures brings different representative pro blems due to further changes in a patient when performing medical procedures necessary to determine some bodily abnormality. Such as changes in breathing that lead to an increase or decrease in the heart rate. Also, changes in the patient’s mood can generate anxiety or other abnormality. Thus problems produced multiple states in complex body structures. From a computer-vision perspective, finding corres pondence between different complex shapes from 3D meshes is often a cumbersome task since sometimes it is impossible to find similarity measures between elements in magnetic resonance imaging. For that reason, we proposed a methodology to find and describe brain structures based on variational multiview unsupervised learning. This proposed model includes unsupervised learning that permits finding similar variability automatically. Also, we use multiple views as a strategy to observe and teat the information from magnetic resonance images. Thus techniques facilitate the extraction of representative characteristics. Finally, the variational inference is applied, which allows the management of computationally intractable Bayesian models that require the creation of latent variables by inferring simple probability density functions on the system model. The use of variational inference increases scalability, convergence speed, and success. The results show how the proposed models capture the non-linearities present in non-rigid 3D shapes, even when they present occlusion or partialities. Thus results prove that the proposed techniques are suitable for shape correspondence analysis.Índice general 1. Introducción 12 1.1. Introducción . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.2. Planteamiento del problema . . . . . . . . . . . . . . . . . . . . 13 1.2.1. Perspectiva Clínica . . . . . . . . . . . . . . . . . . . . . 13 1.2.2. Correspondencia de Formas . . . . . . . . . . . . . . . . 14 1.2.3. Formulación del Problema . . . . . . . . . . . . . . . . . 16 1.3. Justificación . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.3.1. Pertinencia . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.3.2. Viabilidad . . . . . . . . . . . . . . . . . . . . . . . . . . 17 1.3.3. Impacto . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.4. Objetivos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.4.1. General . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.4.2. Específicos . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.5. Publicaciones Asociadas y Software . . . . . . . . . . . . . . . . 19 2. Antecedentes 20 2.1. Multiview Learning . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2. Variational inference en aprendizaje no supervisado . . . . . . . . 25 2.3. Correspondencia de formas . . . . . . . . . . . . . . . . . . . . . 31 2.3.1. Descripción general del problema de la correspondencia . 31 2.4. Imágenes de resonancia magnética . . . . . . . . . . . . . . . . . 35 3. Non-Linear Correspondence Analysis Using Variational Gaussian Mix ture Models 37 3.1. Materiales y Métodos . . . . . . . . . . . . . . . . . . . . . . . . 38 3.1.1. Scale-invariant Heat Kernel Signature (SI-HKS) . . . . . 38 3.1.2. Non-linear Gaussian Mixture Model para Corresponden cias Groupwise . . . . . . . . . . . . . . . . . . . . . . . 39 3.1.3. Variational Inference . . . . . . . . . . . . . . . . . . . . 40 3.1.4. Random Fourier Feature . . . . . . . . . . . . . . . . . . 42 3.1.5. Normalized Geodesic Error como medida de evaluación . 42 3.2. Resultados . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 3.2.1. Base de datos Tosca non-rigid world . . . . . . . . . . . . 43 6 ÍNDICE GENERAL ÍNDICE GENERAL 3.2.2. Base de datos SHREC’16 . . . . . . . . . . . . . . . . . 44 3.2.3. Base de datos de estructuras cerebrales . . . . . . . . . . 47 3.3. Conclusiones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4. Multiview Bayesian Variational Mixtures for Gaussian Process Latent Variable Models (M-VMGPLVM) 49 4.1. El Modelo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.2. Variational Inference . . . . . . . . . . . . . . . . . . . . . . . . 52 4.2.1. El evidence lower bound (ELBO) . . . . . . . . . . . . . 53 4.3. Resultados . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.3.1. Base de datos Toy . . . . . . . . . . . . . . . . . . . . . 57 4.3.2. Base de datos Tosca Non-rigid world . . . . . . . . . . . 58 4.3.3. Base de datos SHREC’16 . . . . . . . . . . . . . . . . . 61 4.4. Base de datos de estructuras cerebrales . . . . . . . . . . . . . . . 64 4.5. Conclusiones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 5. Conclusiones y Trabajos Futuros 68 5.1. Conclusiones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 5.2. Trabajos Futuros . . . . . . . . . . . . . . . . . . . . . . . . . . 69 A. Momentos Estadísticos de Distribuciones de Probabilidad 71 A.1. Primer y segundo momento de q(X | Z) sobre X . . . . . . . . . 71 A.2. Primer y segundo momento de q(Z) . . . . . . . . . . . . . . . . 72 A.3. Primer y segundo momento de q(X) . . . . . . . . . . . . . . . . 73 B. Derivaciones Matemáticas de Multiview Bayesian Variational Mixtu res for Gaussian Process Latent Variable Models (M-VMGPLVM) 74 B.1. El evidence lower bound (ELBO) . . . . . . . . . . . . . . . . . . 74 B.2. Segundo término en el ELBO . . . . . . . . . . . . . . . . . . . . 75 B.3. Tercer término en ELBO . . . . . . . . . . . . . . . . . . . . . . 78 B.4. Quinto término en ELBO . . . . . . . . . . . . . . . . . . . . . . 79MaestríaMagíster en Ingeniería Eléctrica89 Páginasapplication/pdfUniversidad Tecnológica de PereiraMaestría en Ingeniería EléctricaFacultad de IngenieríasPereira620 - Ingeniería y operaciones afines::629 - Otras ramas de la ingenieríaInference (logic)Shape correspondenceBrain structureUnsupervised learningVariational inferenceMultiview learningAnálisis de morfología estructural cerebral a partir de correspondencias de forma Usando (Variational multiview unsupervised learning)Trabajo de grado - Pregradoinfo:eu-repo/semantics/acceptedVersionhttp://purl.org/coar/resource_type/c_7a1fTextinfo:eu-repo/semantics/bachelorThesisA. Damianou, C. H. Ek, M. K. Titsias, and N. D. Lawrence, “Manifold relevance determination,” in Proceedings of the International Conference in Machine Learning, J. Langford and J. Pineau, Eds., vol. 29. San Francisco, CA: Morgan Kauffman, 00 2012. [Online]. Available: http://inverseprobability.com/publications/damianou-manifold12.htmlM. Ovsjanikov, M. Ben-Chen, J. Solomon, A. Butscher, and L. Guibas., “Fun ctional maps: A flexible representation of maps between shapes.” ACM Trans. Graph., 2012V.-M. R. L.-R. J. M.-S. J. L.-M. F. N. V. B. . Legaz-Aparicio, A.-G., “Efficient variational approach to multimodal registration of anatomical and functional intra-patient tumorous brain data,” Int. J. Neural Syst., 2017.P.-M. Olson, L.D., “Localization of epileptic foci using multimodality neuro imaging,” 2013.M. Talo, O. Yildirim, U. B. Baloglu, G. Aydin, and U. R. Acharya, “Con volutional neural networks for multi-class brain disease detection using mri images,” Computerized Medical Imaging and Graphics, vol. 78, dec 2019.H. Hardy, L. Brynildson, and B. Bronson, “Computer rendering of stereo tactic atlas data with whole brain mapping with computed tomography and magnetic resonance imaging,” Computers in Stereotactic Neurosurgery, pp. 109–133, 1992Y. Katayama, H. Oshima, T. Kano, K. Kobayashi, C. Fukaya, and T. Yama moto, “Direct effect of subthalamic nucleus stimulation on levodopa-induced peak-dose dyskinesia in patients with parkinson’s disease,” Stereotactic and Functional Neurosurgery, vol. 84, pp. 176–179, aug 2006.G. Schaltenbrand and P. Bailey, “Introduction to stereotaxis with an atlas of the human brain,” Georg Thieme, 1959.M. Yoshida, “Three-dimensional maps by interpolation from the schalten brand and bailey atlas,” Computers in Stereotactic Neurosurgery, pp. 143– 152, 1992.A. H. A. Shon, K. Grochow and R. Rao, “Learning shared latent structure for image synthesis and robotic imitation,” Advances in Neural Information Processing Systems, vol. 18, no. 1233, 2006R. J. T. P. R.-G. Ek, Carl Henrik and N. Lawrence, “Ambiguity modeling in latent spaces,” Machine Learning and Multimodal Interaction, 2008.R. T. Angela Serra, Paola Galdi, Artificial Intelligence in the Age of Neural Networks and Brain Computing. ACADEMIC PRESS, 2019, ch. Chapter 13 - Multiview Learning in Biomedical Applications.G. Sfikas and C. Nikou, “Bayesian multiview manifold learning applied to hippocampus shape and clinical score data,” in Medical Computer Vision and Bayesian and Graphical Models for Biomedical Imaging. Springer Interna tional Publishing, 2017, pp. 160–171.C. X. Chang Xu, Dacheng Tao, “A survey on multi-view learning,” ArXiv, apr 2013.H. K. S. M. Cheng Zhang, Judith Butepage, “Advances in variational infe rence,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, pp. 2008–2026, dec 2018.T. B. M. Michael Salter-Townshend, “Variational bayesian inference for the latent position cluster model for network data,” Computational Statistics & Data Analysis, vol. 57, pp. 661–671, jan 2013M. G. b. P. L. a. M. L. a. GuoJun Liu a, Yang Liu a, “Variational inference with gaussian mixture model and householder flow,” Neural Networks, vol. 109, pp. 43–55, jan 2019.D. M.Pavlovic, B. R.L.Guillaume, E. K.Towlson, N. M.Y.Kuek, S. Afyouni, ´ P. E.Vértes, B. Yeo, E. T.Bullmore, and T. E.Nichols, “Multi-subject stochas tic blockmodels for adaptive analysis of individual differences in human brain network cluster structure,” NeuroImage, 2020.C. R. Madan, “Advances in studying brain morphology: The benefits of open access data,” Frontier in Human Neuroscience, 2017.T. Shiohama, J. Levman, N. Baumer, and E. Takahashi, “Structural magnetic resonance imaging-based brain morphology study in infants and toddlers with down syndrome: The effect of comorbidities,” Pediatric Neurobiology, pp. 67–73, 2019.J. Corps and I. Rekik, “Morphological brain age prediction using multi-view brain networks derived from cortical morphology in healthy and disordered participants,” Scientific Reports, 2019.P. M. C. F. B. Alberto Llera, Thomas Wolfers, “Inter-individual differences in human brain structure and morphology link to variation in demographics and behavior,” eLife, 2019.V. Chouvatut and E. Boonchieng, “Brain tumor’s approximate corresponden ce and area with interior holes filled,” in 2017 14th International Joint Con ference on Computer Science and Software Engineering (JCSSE), 2017, pp. 1–5.J. V. Buren and R. Borke, “Variations and connections of the human thala mus,” in Springer Berlin, 1E. Sirnes, L. Oltedal, H. Bartsch, G. E. Eide, I. B. Elgen, and S. M. Aukland, “Brain morphology in school-aged children with prenatal opioid exposure: A structural mri study,” Early Human Development, pp. 33–39, 2017.O. van Kaick Hao Zhang Ghassan Hamarneh Daniel Cohen-Or, “A survey on shape correspondence,” Computer Graphics Forum, jul 2011.Z. Lahner, M. Vestner, A. Boyarski, O. Litany, R. Slossberg, T. Remez, E. Ro dolà, A. Bronstein, M. Bronstein, R. Kimmel, and D. Cremers, “Efficient deformable shape correspondence via kernel matching,” arXiv, Cornell Uni versity, sep 2017J. Krüger, S. Schultz, and H. Handels, “Registration with probabilistic corres pondences – accurate and robust registration for pathological and inhomoge neous medical data,” Computer Vision and Image Understanding, 2019.M. Ovsjanikov, Handbook of Numerical Analysis. LIX, Ecole Polytechni que, CNRS, Palaiseau, France, 2018, ch.T. R. Song Wang, Brent Munsell, Statistical shape and Deformation Analysis. Elsevier Ltd., 2017, ch. 3.H. F. García, . A. Orozco, and M. A. Álvarez, “Nonlinear probabilistic latent variable models for groupwise correspondence analysis in brain structures,” in 2018 IEEE 28th International Workshop on Machine Learning for Signal Processing (MLSP), 2018, pp. 1–6.D. Boscaini, J. Masci, E. Rodolà, and M. M. Bronstein, “Learning shape co rrespondence with anisotropic convolutional neural networks,” arXiv, Cornell University, may 2016.T. B. Alan Brunton, Augusto Salazar and StefanieWuhrer, “Review of statis tical shape spaces for 3d data with comparative analysis for human faces,” Computer Vision and Image Understanding, vol. 128, pp. 1–6, 2014.U. S. M. Aubry and D. Cremers, “The wave kernel signature: A quantum mechanical approach to shape analysis,” In Proc. 4DMOD, 2011.M. M. Bronstein and I. Kokkinos, “Scale-invariant heat kernel signatures for non-rigid shape recognition,” In Proc. CVPR, 2010.R. M. Rustamov, “Laplace-beltrami eigenfunctions for deformation invariant shape representation,” In Proc. SGP, 2007.Y. Aflalo, A. Bronstein, and R. Kimmel, “On convex relaxation of graph iso morphism,” PNAS, 2015.M. M. B. A. M. Bronstein and R. Kimmel, “Generalized multidimensional scaling: a framework for isometryinvariant partial surface matching,” PNAS,, 2006.Q. Chen and V. Koltun., “Robust nonrigid registration by convex optimiza tion,” In Proc. ICCV, 2015.I. Kezurer, S. Z. Kovalsky, R. Basri, , and Y. Lipman, “Tight relaxation of quadratic matching,” In Computer Graphics Forum, 2015.K. A. Sidorov, S. Richmond, and D. Marshall, “Efficient groupwise non-rigid registration of textured surfaces,” in Proceedings of the 2011 IEEE Conferen ce on Computer Vision and Pattern Recognition, Washington, DC, pp. 2401– 2408, 2011.T. Iwata, T. Hirao, and N. Ueda, “Probabilistic latent variable models for unsupervised many-to-many object matching,” Information Processing and Management, pp. 682–697, 2016.J. Masci, D. Boscaini, M. M. Bronstein, and P. Vandergheynst, “Geodesic convolutional neural networks on riemannian manifolds,” In Proc. 3dRR,, pp. 682–697, 2015.D. C. E. V. L. Wei, Q. Huang and H. Li, “Dense human body correspondences using convolutional networks,” In Proc. CVPR, 2016.A. K. Z. Wu, S. Song, “3d shapenets: A deep representation for volumetric shapes,” In Proc. CVPR, pp. 682–697, 2015.L. Z. F. a. I. Z. H. X. J. W. R. . E. L. G. S. P. a. X. H. R. W.-R. Tuo Zhang Methodology; Conceptualization; Writing-Original draft preparation, Ying Huang Formal analysis; Investigation and T. L. C. Editing, “Identifying cross individual correspondences of 3-hinge gyri,” Medical Image Analysis, 2020.J. A. Claudia Blaiottaa, M. Jorge Cardosob, “Variational inference for me dical image segmentation,” Computer Vision and Image Understanding, vol. 151, pp. 14–28, oct 2016.Z. G. Tomoharu Iwata, David duvenaud, “Warped mixtures for nonparametric cluster shapes,” in UAI’13: Proceedings of the Twenty-Ninth Conference on Uncertainty in Artificial Intelligence, aug 2013, pp. 311–320.M. Cootes T. F., Taylor C.J., “Statistical models of appearance for computer vision,” 2004.X. X. S. S. Jing Zhao, Xijiong Xie, “Multi-view learning overview: Recent progress and new challenges,” Information Fusion, vol. 38, pp. 43–54, feb 2017.N. Lawrence, “Gaussian process latent variable models for visualisation of high dimensional data,” Advances in Neural Information Processing Systems, vol. 16, pp. 329–336, 2004Q. L. Zhenglong Li and H. Lu, “A variational multi-view learning framework and its application to image segmentation,” IEEE International Conference on Multimedia and Expo, pp. 1516–1519, jul 2009.L. H. Changde Du, Changying Du and H. He, “Reconstructing perceived images from human brain activities with bayesian deep multiview learning,” IEEE Transactions on Neural Networks and Learning Systems, vol. 30, no. 8, pp. 2310–2323, aug 2019.Y. Q. C. L. X. L. Bingqian Lin, Yuan Xie, “Jointly deep multi-view learning for clustering analysis,” ArXiv, nov 2018.Z. C. Liang Zhao and Z. J. Wang, “Unsupervised multiview nonnegative co rrelated feature learning for data clustering,” IEEE Signal Processing Letters, vol. 25, no. 1, pp. 60–64, nov 2017.R. R. Shashini De Silva, Jinsub Kim, “Unsupervised multiview learning with partial distribution information,” in 2017 IEEE 27th International Workshop on Machine Learning for Signal Processing (MLSP), 2017, pp. 1–6.P. H. Y. S. Xu Wang, Dezhong Peng, “Adversarial correlated autoencoder for unsupervised multi-view representation learning,” Knowledge-Based Sys tems, vol. 168, pp. 109–120, mar 2019.D. Z. Wentao Fan, Nizar Bouguila, “Unsupervised hybrid feature extraction selection for high-dimensional non-gaussian data clustering with variational inference,” IEEE Transactions on Knowledge and Data Engineering, vol. 25, no. 7, pp. 1670–1685, may 2013X. C. P. A. W.-C. L. Tien Thanh Nguyen, Thi Thu Thuy Nguyen, “A novel combining classifier method based on variational inference,” Pattern Recog nition, vol. 49, pp. 198–212, jan 2016.N. B. Hieu Nguyen, Muhammad Azam, “Data clustering using variational learning of finite scaled dirichlet mixture models,” in 2019 IEEE 28th Inter national Symposium on Industrial Electronics (ISIE), 2019, pp. 1391–1R. K. N. W. C. B. D. Z. L. M. M. B. A. L. Klaus Greff, Raphaël Lopez Kauf man, “Multi-object representation learning with iterative variational inferen ce,” ArXiv, may 2019M. A. l. Hernán Darío Vargas Cardona, Álvaro Ángel Orozco, “Unsupervi sed learning applied in mer and ecg signals through gaussians mixtures with the expectation-maximization algorithm and variational bayesian inference,” 2013 35th Annual International Conference of the IEEE Engineering in Me dicine and Biology Society (EMBC), pp. 4326–4329, jul 2013Q. Zheng, A. Sharf, A. Tagliasacchi, H. Z. Baoquan Chen, A. Sheffer, and D. C. Or., “Consensus skeleton for non-rigid space-time registration.” Com puter Graphcis Forum (Special Issue of Eurographics), vol. 29, no. 2, pp. 635–644, 2010.A. A. Salama, R. A. Alarabawy, W. El-shehaby, D. El-amrousy, M. S. Bagh dadi, and M. F. Rizkallah., “Consensus skeleton for non-rigid space-time re gistration.” The Egyptian Journal of Radiology and NucleF. Kanavati, T. Tong, K. Misawa, K. M. Michitaka Fujiwara, D. Rueckert, , and B. Glocker., “Supervoxel classification forests for estimating pairwise image correspondences.” Pattern Recognition., vol. 63, pp. 561–569, 2017.M. Cabezas, A. Oliver, X. Lladó, J. Freixenet, and M. B. Cuadra., “A review of atlas-based segmentation for magnetic resonance brain images.” Computer Methods and Programs in Biomedicine., vol. 104, no. 3, pp. 158–177, 2011.D. Aiger, N. J. Mitra, and D. Cohen-Or., “4pointss congruent sets for ro bust pairwise surface registration.” ACM SIGGRAPH 2008 Papers., pp. 84–1, 2008.R. Panda, S. Agrawal, M. Sahoo, and R. Nayak, “novel evolutionary rigid body docking algorithm for medical image registration.” Swarm and Evolu tionary Computation., vol. 33, pp. 108–118, 2017.C. P. Weingarten, M. H. Sundman, P. Hickey, and N. kuei Chen., “Neuroima ging of parkinson’s disease: Expanding views.” Neuroscience and Biobeha vioral Reviews., vol. 59, pp. 16–52, 2015L. Wang and C. Pan., “Nonrigid medical image registration with locally linear reconstruction.” Neurocomputing., vol. 145, pp. 303–315, 2014.A. S. Lundervold and A. Lundervold., “An overview of deep learning in me dical imaging focusing on mri.” Zeitschrift für Medizinische Physik., vol. 145, pp. 102–127, 2019.C.-Y. Wee, C. Liu, A. Lee, J. S. Poh, H. Ji, and A. Qiu., “Cortical graph neural network for ad and mci diagnosis and transfer learning across populations.” NeuroImage: Clinical., vol. 23, 2019.K. A. Sidorov, S. Richmond, and D. Marshall., “Efficient groupwise non-rigid registration of textured surfaces.” In Proceedings of the 2011 IEEE Conferen ce on Computer Vision and Pattern Recognition., pp. 2401–2408, 20M. M. Bronstein and I. Kokkinos, “Scale-invariant heat kernel signatures for non-rigid shape recognition,” In Proc. CVPR, 2010.K. Cutajar, E. V. Bonilla, P. Michiardi, and M. Filippone, “Random feature expansions for deep gaussian processes,” Proceedings of the 34th Internatio nal Conference on Machine Learning, vol. 70, pp. 884–893, 2017.A. M. Bronstein, M. M. Bronstein, L. J. Guibas, and M. Ovsjanikov, “Shape google: Geometric words and expressions for invariant shape retrieval,” ACM Trans. Graph., vol. 30, pp. 1–20, 2011.A. Rahimi and B. Recht, “Random features for largescale kernel machines,” In Neural Infomration Processing Systems., 2007.A. Corduneanu and C. M. Bishop, “Variational bayesian model selection for mixture distributions,” Artificial Intelligence and Statistics, 2001.K. V. G., L. Y., and funkhouser T. A., “Blended intrinsic maps.” TOG 30, 2011.A. M. Bronstein, M. M. Bronstein, and R. Kimmel, “Calculus of nonrigid surfaces for geometry and texture manipulation,” IEEE Trans. Vis. Comput. Graph., 2007.S. Y. and A. Y., “Coarse-to-fine com-binatorial matching for dense isometric shape correspondence.” Computer Graphics Forum, 2011.M. Vestner, Z. Lähner, A. Boyarski, O. Litany, R. Slossberg, T. Remez, E. Ro dola, A. Bronstein, M. Bronstein, R. Kimmel, and D. Cremers, “Efficient de formable shape correspondence via kernel matching.” 017 International Con ference on 3D Vision (3DV), pp. 517–526,Y. Aflalo, A. Dubrovina, and R. Kimmel., “Spectral generalized multidimen sional scaling.” Int. J. Comput. Vision, pp. 380–392, 2016.E. Rodolà, S. R. Bulò, T. Windheuser, M. Vestner, and D. Cremers., “Dense non-rigid shape correspondence using random forests.” Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, CVPR ’14, pp. 4177–4184, 2014.L. Cosmo, E. Rodolà, M. M. Bronstein, A. Torsello, D. Cremers, and Y. Sahi llio glu, “Shrec’16: Partial matching of deformable shapes,” Eurographics Workshop on 3D Object Retrieval, 2016.E. Rodola, L. Cosmo, M. M. Bronstein, A. Torsello, and D. Cremers., “Partial functional correspondence.” Computer Graphics Forum, 2016.I. S. Gousias, A. D. Edwards, M. A. Rutherford, S. J. Counsell, J. V. Hajnal, D. Rueckert, and A. Hammers., “Magnetic resonance imaging of the newborn brain: Manual segmentation of labelled atlases in term-born and preterm in fants.” NeuroImage, pp. 1499–1509, 201A. C. Damianou, M. K. Titsias, and N. D. Lawrence, “Variational inference for latent variables and uncertain inputs in gaussian processes,” Journal of Machine Learning Research, 2016.P. WD, “Kl-divergences of normal, gamma, dirichlet and wishart densities,” University College, London, 2001.A. G. d. G. Matthews, M. van der Wilk, T. Nickson, K. Fujii, A. Boukouvalas, P. León-Villagrá, Z. Ghahramani, and J. Hensman, “GPflow: A Gaussian process library using TensorFlow,” Journal of Machine Learning Research, vol. 18, no. 40, pp. 1–6, apr 2017. [Online]. Available: http://jmlr.org/papers/v18/16-537.htmlJ. Hensman, F. Nicolo, and N. D. Lawrence, “Gaussian processes for big data.” Uncertainty in Artificial Intelligence, 2013.M. Johnson, D. K. Duvenaud, A. Wiltschko, R. P. Adams, , and S. R. Dat ta., “Composing graphical models with neural networks for structured repre sentations and fast inference.” Advances in Neural Information Processing Systems, vol. 29, pp. 2946–2954, 2016.A. Hebbal, L. Brevault, M. Balesdent, E.-G. Talbi, and N. Melab., “Bayesian optimization using deep gaussian processes.” hal-02924230ff, 2020.PublicationORIGINALTRABAJO DE GRADO.pdfTRABAJO DE GRADO.pdfapplication/pdf6701289https://dspace7-utp.metabuscador.org/bitstreams/85695f8e-9e74-4841-99f4-1488cfc0d090/download4f19f4b4a0844163025fcb8cadd1ef85MD51LICENSElicense.txtlicense.txttext/plain; charset=utf-814828https://dspace7-utp.metabuscador.org/bitstreams/90452e11-b733-430f-a151-94159f0d98bd/download2f9959eaf5b71fae44bbf9ec84150c7aMD52TEXTTRABAJO DE GRADO.pdf.txtTRABAJO DE GRADO.pdf.txtExtracted texttext/plain162815https://dspace7-utp.metabuscador.org/bitstreams/17df1aa1-b68b-4fe3-ae08-70f3a16038c1/downloadf3455e61e5da9b76c168ab1f0fef031aMD53THUMBNAILTRABAJO DE GRADO.pdf.jpgTRABAJO DE GRADO.pdf.jpgGenerated Thumbnailimage/jpeg7719https://dspace7-utp.metabuscador.org/bitstreams/e92cd28c-37bc-424a-bbb1-f2f8698e2afb/download155e157c60bf79c1df340474bf7f26b7MD5411059/14267oai:dspace7-utp.metabuscador.org:11059/142672024-09-05 17:07:50.055https://creativecommons.org/licenses/by-nc-nd/4.0/Manifiesto (Manifestamos) en este documento la voluntad de autorizar a la Biblioteca Jorge Roa Martínez de la Universidad Tecnológica de Pereira la publicación en el Repositorio institucional (http://biblioteca.utp.edu.co), la versión electrónica de la OBRA titulada: ________________________________________________________________________________________________ ________________________________________________________________________________________________ ________________________________________________________________________________________________ La Universidad Tecnológica de Pereira, entidad académica sin ánimo de lucro, queda por lo tanto facultada para ejercer plenamente la autorización anteriormente descrita en su actividad ordinaria de investigación, docencia y publicación. La autorización otorgada se ajusta a lo que establece la Ley 23 de 1982. Con todo, en mi (nuestra) condición de autor (es) me (nos) reservo (reservamos) los derechos morales de la OBRA antes citada con arreglo al artículo 30 deopen.accesshttps://dspace7-utp.metabuscador.orgRepositorio de la Universidad Tecnológica de Pereirabdigital@metabiblioteca.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 |