Modeling Shared Responses in Neuroimaging Studies through MultiView ICA

NIPS 2020, 2020.

Cited by: 2|Views42
EI
Weibo:
We demonstrated the usefulness of MultiView Independent Component Analysis for neuroimaging group studies both on functional MRI and MEG data, where it outperforms other methods

Abstract:

Group studies involving large cohorts of subjects are important to draw general conclusions about brain functional organization. However, the aggregation of data coming from multiple subjects is challenging, since it requires accounting for large variability in anatomy, functional topography and stimulus response across individuals. Dat...More
0
Full Text
Bibtex
Weibo
Introduction
  • The past decade has seen the emergence of two trends in neuroimaging: the collection of massive neuroimaging datasets, containing data from hundreds of participants [57, 59, 55], and the use of naturalistic stimuli to move closer to a real life experience with dynamic and multimodal stimuli [54].
  • The statistical analysis of the data using supervised regression-based approaches is difficult
  • This has motivated the use of unsupervised learning methods that leverage the availability of data from multiple subjects performing the same experiment; analysis on such large groups boosts statistical power.
  • Independent component analysis [38] (ICA) is a widely used unsupervised method for neuroimaging studies
  • It is routinely applied on individual subject electroencephalography (EEG) [41], magnetoencephalography (MEG) [61] or functional MRI [43] data.
  • The identifiability theory of ICA states that having non-Gaussian independent sources is a strong enough condition to recover the
Highlights
  • The past decade has seen the emergence of two trends in neuroimaging: the collection of massive neuroimaging datasets, containing data from hundreds of participants [57, 59, 55], and the use of naturalistic stimuli to move closer to a real life experience with dynamic and multimodal stimuli [54]
  • The statistical analysis of the data using supervised regression-based approaches is difficult. This has motivated the use of unsupervised learning methods that leverage the availability of data from multiple subjects performing the same experiment; analysis on such large groups boosts statistical power
  • In contrast to previous approaches, the proposed model leads to a closed-form likelihood, which we optimize efficiently using a dedicated alternate quasi-Newton approach
  • Our approach enjoys the statistical guarantees of maximum-likelihood theory, while still being tractable
  • We demonstrated the usefulness of MultiView Independent Component Analysis (ICA) for neuroimaging group studies both on functional MRI (fMRI) and MEG data, where it outperforms other methods
  • Our method is not specific to neuroimaging data and could be relevant to other observational sciences like genomics or astrophysics where ICA is already widely used
Methods
  • Find Exp Clin Pharmacol, 24(Suppl D):5–12, 2002.

    [47] Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12:2825–2830, 2011.

    [48] Kaare Brandt Petersen, Ole Winther, and Lars Kai Hansen.
  • [47] Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al Scikit-learn: Machine learning in Python.
  • On the slow convergence of EM and VBEM in low-noise linear models.
  • [49] Niklas Pfister, Sebastian Weichwald, Peter Bühlmann, and Bernhard Schölkopf.
  • Robustifying independent component analysis by adjusting for group-wise stationary noise.
  • Journal of Machine Learning Research, 20(147):1–50, 2019.
Conclusion
  • Discussion on ROIs choice

    The quality of the reconstructed BOLD signal varies depending on the choice of the region of interest.
  • In Figure 4, the authors plot for GroupICA, SRM and MultiViewICA, the R2 score per voxel using 50 components for datasets sherlock, forrest, raiders and clips.
  • The authors see visually see that data reconstructed by MultiViewICA are a better approximation of the original data than other methods
  • This is obvious for the clips datasets where it is clear that voxels in the posterior part of the superior temporal sulcus are better recovered by MultiViewICA than by SRM or GroupICA.
  • The authors' method is not specific to neuroimaging data and could be relevant to other observational sciences like genomics or astrophysics where ICA is already widely used
Summary
  • Introduction:

    The past decade has seen the emergence of two trends in neuroimaging: the collection of massive neuroimaging datasets, containing data from hundreds of participants [57, 59, 55], and the use of naturalistic stimuli to move closer to a real life experience with dynamic and multimodal stimuli [54].
  • The statistical analysis of the data using supervised regression-based approaches is difficult
  • This has motivated the use of unsupervised learning methods that leverage the availability of data from multiple subjects performing the same experiment; analysis on such large groups boosts statistical power.
  • Independent component analysis [38] (ICA) is a widely used unsupervised method for neuroimaging studies
  • It is routinely applied on individual subject electroencephalography (EEG) [41], magnetoencephalography (MEG) [61] or functional MRI [43] data.
  • The identifiability theory of ICA states that having non-Gaussian independent sources is a strong enough condition to recover the
  • Methods:

    Find Exp Clin Pharmacol, 24(Suppl D):5–12, 2002.

    [47] Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12:2825–2830, 2011.

    [48] Kaare Brandt Petersen, Ole Winther, and Lars Kai Hansen.
  • [47] Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al Scikit-learn: Machine learning in Python.
  • On the slow convergence of EM and VBEM in low-noise linear models.
  • [49] Niklas Pfister, Sebastian Weichwald, Peter Bühlmann, and Bernhard Schölkopf.
  • Robustifying independent component analysis by adjusting for group-wise stationary noise.
  • Journal of Machine Learning Research, 20(147):1–50, 2019.
  • Conclusion:

    Discussion on ROIs choice

    The quality of the reconstructed BOLD signal varies depending on the choice of the region of interest.
  • In Figure 4, the authors plot for GroupICA, SRM and MultiViewICA, the R2 score per voxel using 50 components for datasets sherlock, forrest, raiders and clips.
  • The authors see visually see that data reconstructed by MultiViewICA are a better approximation of the original data than other methods
  • This is obvious for the clips datasets where it is clear that voxels in the posterior part of the superior temporal sulcus are better recovered by MultiViewICA than by SRM or GroupICA.
  • The authors' method is not specific to neuroimaging data and could be relevant to other observational sciences like genomics or astrophysics where ICA is already widely used
Related work
  • Many methods for data-driven multivariate analysis of neuroimaging group studies have been proposed. We summarize the characteristics of some of the most commonly used ones. A more thorough description of these methods can be found in appendix F. For completeness, we start by describing PCA. For a zero-mean data matrix X of size p×n with p ≤ n, we denote X = U DV the singular value decomposition of X where U ∈ Rp×p, V ∈ Rn×p are orthogonal and D the diagonal matrix of singular values ordered in decreasing order. The PCA of X with k components is Y ∈ Rk×n containing the first k rows of DV , and in general we do not have Y Y = Ik: in the rest of the paper, what we call PCA does not whiten signals.
Reference
  • Pierre Ablin, Jean-François Cardoso, and Alexandre Gramfort. Faster independent component analysis by preconditioning with Hessian approximations. IEEE Transactions on Signal Processing, 66(15):4040–4049, 2018.
    Google ScholarLocate open access versionFindings
  • Alexandre Abraham, Fabian Pedregosa, Michael Eickenberg, Philippe Gervais, Andreas Mueller, Jean Kossaifi, Alexandre Gramfort, Bertrand Thirion, and Gaël Varoquaux. Machine learning for neuroimaging with scikit-learn. Frontiers in neuroinformatics, 8:14, 2014.
    Google ScholarLocate open access versionFindings
  • Shun-ichi Amari, Andrzej Cichocki, and Howard H Yang. A new learning algorithm for blind signal separation. In Advances in neural information processing systems, pages 757–763, 1996.
    Google ScholarLocate open access versionFindings
  • Andreas Bartels and Semir Zeki. Brain dynamics during natural viewing conditions—a new guide for mapping connectivity in vivo. Neuroimage, 24(2):339–349, 2005.
    Google ScholarLocate open access versionFindings
  • Christian F Beckmann, Marilena DeLuca, Joseph T Devlin, and Stephen M Smith. Investigations into resting-state connectivity using independent component analysis. Philosophical Transactions of the Royal Society B: Biological Sciences, 360(1457):1001–1013, 2005.
    Google ScholarLocate open access versionFindings
  • Christian F Beckmann, Clare E Mackay, Nicola Filippini, and Stephen M Smith. Group comparison of resting-state fMRI data using multi-subject ICA and dual regression. Neuroimage, 47(Suppl 1):S148, 2009.
    Google ScholarLocate open access versionFindings
  • Christian F Beckmann and Stephen M Smith. Tensorial extensions of independent component analysis for multisubject fMRI analysis. Neuroimage, 25(1):294–311, 2005.
    Google ScholarLocate open access versionFindings
  • Anthony J Bell and Terrence J Sejnowski. An information-maximization approach to blind separation and blind deconvolution. Neural computation, 7(6):1129–1159, 1995.
    Google ScholarLocate open access versionFindings
  • Olivier Bermond and Jean-François Cardoso. Approximate likelihood for noisy mixtures. In Proc. ICA, volume 99, pages 325–330.
    Google ScholarLocate open access versionFindings
  • Nima Bigdely-Shamlo, Tim Mullen, Kenneth Kreutz-Delgado, and Scott Makeig. Measure projection analysis: a probabilistic approach to EEG source comparison and multi-subject inference. Neuroimage, 72:287–303, 2013.
    Google ScholarLocate open access versionFindings
  • Vince D Calhoun, Tülay Adali, Vince B McGinty, James J Pekar, Todd D Watson, and Godfrey D Pearlson. fMRI activation in a visual-perception task: network of areas detected using the general linear model and independent components analysis. NeuroImage, 14(5):1080–1088, 2001.
    Google ScholarLocate open access versionFindings
  • Vince D Calhoun, Tülay Adali, Godfrey D Pearlson, and James J Pekar. A method for making group inferences from functional MRI data using independent component analysis. Human brain mapping, 14(3):140–151, 2001.
    Google ScholarLocate open access versionFindings
  • Vince D Calhoun, Jingyu Liu, and Tülay Adalı. A review of group ICA for fMRI data and ICA for joint inference of imaging, genetic, and erp data. Neuroimage, 45(1):S163–S172, 2009.
    Google ScholarLocate open access versionFindings
  • Vince D Calhoun, James J Pekar, Vince B McGinty, Tülay Adali, Todd D Watson, and Godfrey D Pearlson. Different activation dynamics in multiple neural systems during simulated driving. Human brain mapping, 16(3):158–167, 2002.
    Google ScholarLocate open access versionFindings
  • Jean-François Cardoso. Infomax and maximum likelihood for blind source separation. IEEE Signal processing letters, 4(4):112–114, 1997.
    Google ScholarLocate open access versionFindings
  • Jean-François Cardoso. Blind signal separation: statistical principles. Proceedings of the IEEE, 86(10):2009–2025, 1998.
    Google ScholarLocate open access versionFindings
  • Jean-François Cardoso and Beate H Laheld. Equivariant adaptive source separation. IEEE Transactions on signal processing, 44(12):3017–3030, 1996.
    Google ScholarLocate open access versionFindings
  • Janice Chen, Yuan C Leong, Christopher J Honey, Chung H Yong, Kenneth A Norman, and Uri Hasson. Shared memories reveal shared structure in neural activity across individuals. Nature neuroscience, 20(1):115–125, 2017.
    Google ScholarLocate open access versionFindings
  • Po-Hsuan Chen, Janice Chen, Yaara Yeshurun, Uri Hasson, James Haxby, and Peter J Ramadge. A reduced-dimension fMRI shared response model. In Advances in Neural Information Processing Systems, pages 460–468, 2015.
    Google ScholarLocate open access versionFindings
  • Po-Hsuan Chen, Xia Zhu, Hejia Zhang, Javier S Turek, Janice Chen, Theodore L Willke, Uri Hasson, and Peter J Ramadge. A convolutional autoencoder for multi-subject fMRI data aggregation. arXiv preprint arXiv:1608.04846, 2016.
    Findings
  • Pierre Comon. Independent component analysis, a new concept? Signal processing, 36(3):287–314, 1994.
    Google ScholarLocate open access versionFindings
  • Fengyu Cong, Zhaoshui He, Jarmo Hämäläinen, Paavo HT Leppänen, Heikki Lyytinen, Andrzej Cichocki, and Tapani Ristaniemi. Validating rationale of group-level component analysis based on estimating number of sources in EEG through model order selection. Journal of neuroscience methods, 212(1):165–172, 2013.
    Google ScholarLocate open access versionFindings
  • Arnaud Delorme, Jason Palmer, Julie Onton, Robert Oostenveld, and Scott Makeig. Independent EEG sources are dipolar. PloS one, 7(2), 2012.
    Google ScholarLocate open access versionFindings
  • Tom Eichele, Srinivas Rachakonda, Brage Brakedal, Rune Eikeland, and Vince D Calhoun. EEGIFT: group independent component analysis for event-related EEG data. Computational intelligence and neuroscience, 2011, 2011.
    Google ScholarLocate open access versionFindings
  • Fabrizio Esposito, Tommaso Scarabino, Aapo Hyvärinen, Johan Himberg, Elia Formisano, Silvia Comani, Gioacchino Tedeschi, Rainer Goebel, Erich Seifritz, and Francesco Di Salle. Independent component analysis of fMRI group studies by self-organizing clustering. Neuroimage, 25(1):193–205, 2005.
    Google ScholarLocate open access versionFindings
  • Karl J Friston, Andrew P Holmes, Keith J Worsley, J-P Poline, Chris D Frith, and Richard SJ Frackowiak. Statistical parametric maps in functional imaging: a general linear approach. Human brain mapping, 2(4):189–210, 1994.
    Google ScholarLocate open access versionFindings
  • Alexandre Gramfort, Martin Luessi, Eric Larson, Denis A Engemann, Daniel Strohmeier, Christian Brodbeck, Roman Goj, Mainak Jas, Teon Brooks, Lauri Parkkonen, et al. MEG and EEG data analysis with MNE-Python. Frontiers in neuroscience, 7:267, 2013.
    Google ScholarLocate open access versionFindings
  • Vera A Grin-Yatsenko, Ineke Baas, Valery A Ponomarev, and Juri D Kropotov. Independent component approach to the analysis of EEG recordings at early stages of depressive disorders. Clinical Neurophysiology, 121(3):281–289, 2010.
    Google ScholarLocate open access versionFindings
  • J Swaroop Guntupalli, Ma Feilong, and James V Haxby. A computational model of shared fine-scale structure in the human connectome. PLoS computational biology, 14(4):e1006120, 2018.
    Google ScholarLocate open access versionFindings
  • Ying Guo and Giuseppe Pagnoni. A unified framework for group independent component analysis for multi-subject fMRI data. NeuroImage, 42(3):1078–1093, 2008.
    Google ScholarLocate open access versionFindings
  • Michael Hanke, Florian J Baumgartner, Pierre Ibe, Falko R Kaule, Stefan Pollmann, Oliver Speck, Wolf Zinke, and Jörg Stadler. A high-resolution 7-Tesla fMRI dataset from complex natural stimulation with an audio movie. Scientific data, 1:140003, 2014.
    Google ScholarFindings
  • James V Haxby, J Swaroop Guntupalli, Andrew C Connolly, Yaroslav O Halchenko, Bryan R Conroy, M Ida Gobbini, Michael Hanke, and Peter J Ramadge. A common, high-dimensional model of the representational space in human ventral temporal cortex. Neuron, 72(2):404–416, 2011.
    Google ScholarLocate open access versionFindings
  • John D Hunter. Matplotlib: A 2d graphics environment. Computing in science & engineering, 9(3):90–95, 2007.
    Google ScholarLocate open access versionFindings
  • Rene Huster, Sergey M Plis, and Vince D Calhoun. Group-level component analyses of EEG: validation and evaluation. Frontiers in neuroscience, 9:254, 2015.
    Google ScholarLocate open access versionFindings
  • Alexander G Huth, Shinji Nishimoto, An T Vu, and Jack L Gallant. A continuous semantic space describes the representation of thousands of object and action categories across the human brain. Neuron, 76(6):1210–1224, 2012.
    Google ScholarLocate open access versionFindings
  • Aapo Hyvarinen. Fast and robust fixed-point algorithms for independent component analysis. IEEE transactions on Neural Networks, 10(3):626–634, 1999.
    Google ScholarLocate open access versionFindings
  • Aapo Hyvärinen. Independent component analysis: recent advances. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 371(1984):20110534, 2013.
    Google ScholarLocate open access versionFindings
  • Aapo Hyvärinen and Erkki Oja. Independent component analysis: algorithms and applications. Neural networks, 13(4-5):411–430, 2000.
    Google ScholarLocate open access versionFindings
  • Tzyy-Ping Jung, Colin Humphries, Te-Won Lee, Scott Makeig, Martin J McKeown, Vicente Iragui, and Terrence J Sejnowski. Extended ica removes artifacts from electroencephalographic recordings. In Advances in neural information processing systems, pages 894–900, 1998.
    Google ScholarLocate open access versionFindings
  • Manoj Kumar, Cameron T Ellis, Qihong Lu, Hejia Zhang, Mihai Capotă, Theodore L Willke, Peter J Ramadge, Nicholas B Turk-Browne, and Kenneth A Norman. Brainiak tutorials: User-friendly learning materials for advanced fmri analysis. PLoS computational biology, 16(1):e1007549, 2020.
    Google ScholarLocate open access versionFindings
  • Scott Makeig, Anthony J Bell, Tzyy-Ping Jung, and Terrence J Sejnowski. Independent component analysis of electroencephalographic data. In Advances in neural information processing systems, pages 145–151, 1996.
    Google ScholarLocate open access versionFindings
  • Sanna Malinen, Yevhen Hlushchuk, and Riitta Hari. Towards natural stimulation in fMRI—issues of data analysis. Neuroimage, 35(1):131–139, 2007.
    Google ScholarLocate open access versionFindings
  • Martin J McKeown and Terrence J Sejnowski. Independent component analysis of fMRI data: examining the assumptions. Human brain mapping, 6(5-6):368–372, 1998.
    Google ScholarLocate open access versionFindings
  • Samuel A Nastase, Yun-Fei Liu, Hanna Hillman, Kenneth A. Norman, and Uri Hasson. Leveraging shared connectivity to aggregate heterogeneous datasets into a common response space. bioRxiv, 2019.
    Google ScholarFindings
  • S. Nishimoto, A.T. Vu, T. Naselaris, Y. Benjamini, B. Yu, and J. L Gallant. Reconstructing visual experiences from brain activity evoked by natural movies. Current Biology, 21(19):1641–1646, 2011.
    Google ScholarLocate open access versionFindings
  • Roberto Domingo Pascual-Marqui et al. Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details. Methods Find Exp Clin Pharmacol, 24(Suppl D):5–12, 2002.
    Google ScholarLocate open access versionFindings
  • Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al. Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12:2825–2830, 2011.
    Google ScholarLocate open access versionFindings
  • Kaare Brandt Petersen, Ole Winther, and Lars Kai Hansen. On the slow convergence of EM and VBEM in low-noise linear models. Neural computation, 17(9):1921–1926, 2005.
    Google ScholarLocate open access versionFindings
  • Niklas Pfister, Sebastian Weichwald, Peter Bühlmann, and Bernhard Schölkopf. Robustifying independent component analysis by adjusting for group-wise stationary noise. Journal of Machine Learning Research, 20(147):1–50, 2019.
    Google ScholarLocate open access versionFindings
  • Ana Luísa Pinho, Alexis Amadon, Torsten Ruest, Murielle Fabre, Elvis Dohmatob, Isabelle Denghien, Chantal Ginisty, Séverine Becuwe-Desmidt, Séverine Roger, Laurence Laurier, et al. Individual brain charting, a high-resolution fMRI dataset for cognitive mapping. Scientific data, 5, 2018.
    Google ScholarLocate open access versionFindings
  • Russell A Poldrack, Deanna M Barch, Jason Mitchell, Tor Wager, Anthony D Wagner, Joseph T Devlin, Chad Cumba, Oluwasanmi Koyejo, and Michael Milham. Toward open sharing of task-based fMRI data: the OpenfMRI project. Frontiers in neuroinformatics, 7:12, 2013.
    Google ScholarLocate open access versionFindings
  • Jean-Baptiste Poline and Matthew Brett. The general linear model and fMRI: does love last forever? Neuroimage, 62(2):871–880, 2012.
    Google ScholarLocate open access versionFindings
  • Stephen M Smith, Aapo Hyvärinen, Gaël Varoquaux, Karla L Miller, and Christian F Beckmann. Group-pca for very large fMRI datasets. Neuroimage, 101:738–749, 2014.
    Google ScholarLocate open access versionFindings
  • Saurabh Sonkusare, Michael Breakspear, and Christine Guo. Naturalistic stimuli in neuroscience: Critically acclaimed. Trends in Cognitive Sciences, 23, 06 2019.
    Google ScholarLocate open access versionFindings
  • Cathie Sudlow, John Gallacher, Naomi Allen, Valerie Beral, Paul Burton, John Danesh, Paul Downey, Paul Elliott, Jane Green, Martin Landray, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS medicine, 12(3), 2015.
    Google ScholarLocate open access versionFindings
  • Markus Svensén, Frithjof Kruggel, and Habib Benali. ICA of fMRI group study data. NeuroImage, 16(3):551–563, 2002.
    Google ScholarLocate open access versionFindings
  • Jason R Taylor, Nitin Williams, Rhodri Cusack, Tibor Auer, Meredith A Shafto, Marie Dixon, Lorraine K Tyler, Richard N Henson, et al. The Cambridge Centre for Ageing and Neuroscience (Cam-CAN) data repository: structural and functional MRI, MEG, and cognitive data from a cross-sectional adult lifespan sample. Neuroimage, 144:262–269, 2017.
    Google ScholarLocate open access versionFindings
  • Petr Tichavsky and Zbynek Koldovsky. Optimal pairing of signal components separated by blind techniques. IEEE Signal Processing Letters, 11(2):119–122, 2004.
    Google ScholarLocate open access versionFindings
  • David C Van Essen, Stephen M Smith, Deanna M Barch, Timothy EJ Behrens, Essa Yacoub, Kamil Ugurbil, Wu-Minn HCP Consortium, et al. The WU-Minn human connectome project: an overview. Neuroimage, 80:62–79, 2013.
    Google ScholarLocate open access versionFindings
  • Gaël Varoquaux, Sepideh Sadaghiani, Jean-Baptiste Poline, and Bertrand Thirion. CanICA: Modelbased extraction of reproducible group-level ica patterns from fMRI time series. arXiv preprint arXiv:0911.4650, 2009.
    Findings
  • Ricardo Vigário, Veikko Jousmäki, Matti Hämäläinen, Riitta Hari, and Erkki Oja. Independent component analysis for identification of artifacts in magnetoencephalographic recordings. In Advances in neural information processing systems, pages 229–235, 1998.
    Google ScholarLocate open access versionFindings
  • Ricardo Vigário, Jaakko Sarela, Veikko Jousmiki, Matti Hamalainen, and Erkki Oja. Independent component approach to the analysis of EEG and MEG recordings. IEEE transactions on biomedical engineering, 47(5):589–593, 2000.
    Google ScholarLocate open access versionFindings
  • Hejia Zhang, Po-Hsuan Chen, Janice Chen, Xia Zhu, Javier S Turek, Theodore L Willke, Uri Hasson, and Peter J Ramadge. A searchlight factor model approach for locating shared information in multi-subject fMRI analysis. arXiv preprint arXiv:1609.09432, 2016.
    Findings
  • Hejia Zhang, Po-Hsuan Chen, and Peter Ramadge. Transfer learning on fMRI datasets. In International Conference on Artificial Intelligence and Statistics, pages 595–603, 2018.
    Google ScholarLocate open access versionFindings
  • Michael Zibulevsky. Blind source separation with relative newton method. In Proc. ICA, volume 2003, pages 897–902, 2003.
    Google ScholarLocate open access versionFindings
  • [63] Mean classification accuracy - error bars represent 95% confidence interval
    Google ScholarFindings
Your rating :
0

 

Tags
Comments