Pith. sign in

REVIEW 5 major objections 5 minor 68 references

Locally Linear Embedding and fMRI feature selection in psychiatric classification

T0 review · 5 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Locally Linear Embedding of fMRI time series yields above-80% accuracy on eleven psychiatric datasets.

desk verdict Novel spatial-neighborhood LLE pipeline for fMRI classification, but the >80% claim rests on training accuracy after selection on the same folds; holdout numbers are weaker. read the letter →

arxiv 1908.06319 v9 pith:5JYVWTAK submitted 2019-08-17 eess.IV cs.LGstat.ML

Gagan Sidhu This is my paper · ORCID
classification eess.IVcs.LGstat.ML
keywords LocallyLinearEmbeddingfMRIdimensionalityreductionpsychiatricclassificationleave-one-outcross-validationfeatureselectionBOLDtime-seriesneuroimaging
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Locally Linear Embedding (LLE) is proposed as a preprocessing step that turns each voxel's BOLD time-series into a low-dimensional spatial embedding before psychiatric classification. The paper reports leave-one-out cross-validation accuracies above 80% on the training partitions of eleven public fMRI datasets covering schizophrenia, ADHD, and autism, and above-chance classification on ten of the eleven, where chance is the majority-class proportion. Raw fMRI and PCA-reduced fMRI perform markedly worse on the same comparison. The motivation is clinical: if the embedding reliably preserves disorder-relevant brain-activity patterns, it would make fMRI-based diagnosis support more feasible and point to interpretable time points and regions.

What carries the argument

The load-bearing object is the per-voxel local covariance matrix $G_i = C_i^T C_i$, built from the differences between a voxel's BOLD waveform and the waveforms of neighbouring voxels on a sphere; solving $G_i w_i = 1$ through a minimum-norm least-squares step gives reconstruction weights that encode local geometry. Modified LLE adds multiple weight vectors when the local covariance is ill-conditioned. A global alignment matrix $\Phi = \sum_i \hat{W}_i \hat{W}_i^T$ is assembled from these weights, and the bottom $d+1$ eigenvectors of $\Phi$ provide the low-dimensional embedding, with the last eigenvector discarded to centre the manifold. This machinery converts raw BOLD time series into low-dimensional spatial modes that the classifier then sees.

What would settle it

Hold out a set of subjects completely untouched by any selection step, choose the embedding dimension $d$ and diagnostic volumes using nested cross-validation on the training set only, then measure LLE accuracy once on the held-out subjects. If the held-out accuracy drops to the majority-class baseline or no longer beats PCA on the same held-out subjects, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a nonlinear, spatially-aware reconstruction of fMRI time series retains diagnostic information that the raw BOLD signal and its linear PCA projection discard. The pipeline embeds each subject's scan by LLE in a neighbourhood defined by the physical sphere around each voxel, solving the local reconstruction problem $G_i w_i = 1$ for each voxel and then finding global coordinates that preserve those local weights. A linear discriminant classifier, fed only the diagnostic volumes selected by sequential forward search, separates patients from controls. The paper reports this separation is above 80% accurate on all eleven training partitions, and significantly better than the majority-class baseline on ten of eleven, while raw and PCA versions often sit at chance.

Load-bearing premise

The headline accuracies come from the same leave-one-out cross-validation loop that chooses the embedding dimension and the diagnostic time points, so the numbers assume that selecting on the training folds does not inflate performance; an untouched test set for those choices is needed before the >80% figure can be taken as an unbiased estimate.

Editorial extensions

If this is right

  • If the embedding is as informative as reported, nonlinear manifold methods become a viable preprocessing alternative to PCA for fMRI-based diagnosis.
  • The selected diagnostic volumes map onto disorder-specific regions, giving concrete hypotheses about where and when patient and control brain activity differ.
  • The method's success on both resting-state and task-based scans suggests the preserved local geometry carries diagnostic signal beyond task-induced activation.
  • The contrast with PCA implies that simple linear preprocessing may discard the very structure that LLE retains for classification.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not isolate how much of the gain comes from LLE's nonlinear geometry versus the sequential-forward-selection step; rerunning SFS on raw and PCA data with identical classifier settings would quantify that split.
  • Because the method selects only a few time points, the same pipeline could be tested on event-related task decoding, predicting stimulus condition rather than diagnosis, where sparse temporal selection could reveal when information appears.
  • The reported regional differences use uncorrected $p<0.05$ thresholds, so those maps are best treated as hypotheses to be confirmed rather than stand-alone evidence.
  • Comparing the stress-tensor sphere neighbourhood against generic nonlinear embeddings on the same datasets would show whether the spatial neighbourhood is the active ingredient or whether any nonlinear reduction would perform similarly.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The manuscript proposes Locally Linear Embedding (LLE) as a nonlinear preprocessing step for fMRI-based psychiatric classification. The pipeline reconstructs each subject's BOLD time-series with LLE, selects the embedding dimension d and a set of diagnostic time points using LOOCV on the training partition together with sequential forward selection, and classifies with Fisher's linear discriminant. The authors report accuracies above 80% on eleven public datasets and claim significantly better-than-chance performance on ten of eleven datasets. The main text contains detailed results for five datasets, with six additional datasets deferred to a supplementary table.

Significance. If the advertised performance were supported, LLE would be a useful feature-extraction step for psychiatric fMRI classification, and the use of several public datasets is a strength. The paper makes falsifiable predictions and describes the pipeline in enough detail to be reimplemented. However, the central quantitative claim currently rests on training accuracies that are optimistically biased by model selection, and the independent holdout results are too few and too imprecise to support the abstract's assertions. The physical-conservation interpretation is not load-bearing for the classification claim and is not derived from the LLE equations.

major comments (5)
  1. [II-C and Table I] The training accuracies reported in Table I are not unbiased estimates of generalization. Section II-C states that 'LOOCV is used on the training data to find both the best reconstruction parameter d, and the diagnostic volumes that produce the highest accuracy for this d.' The same LOOCV accuracy is then reported as the method's performance. Because d and the diagnostic time points are selected by maximizing this same score, the reported training accuracy is optimistically biased. The statement in Section II-A that the selection 'does not induce any biased sampling' is therefore incorrect. The authors should either use a nested cross-validation scheme, where model selection is performed inside each training fold, or base all primary claims on the independent holdout partition.
  2. [Abstract and Table I] The headline '>80%' claim is contradicted by the independent holdout results in Table I: the LLE holdout accuracies are 75% (Beijing), 100% (COBRE), 68.8% (MRN), 70% (Stanford), and 80% (UM_2), so three of five fall below 80%. No confidence intervals, binomial tests, or comparisons against the majority-class baseline are provided for these holdout values, and the assertion of 'significantly better than chance classification' on ten of eleven datasets is therefore unsupported. The authors should report holdout accuracies with exact binomial confidence intervals and a test against the majority proportion (and, if relevant, against the PCA and original-data baselines), and revise the abstract to match the holdout evidence.
  3. [Section III (Evaluation Criteria)] The statement that 'discrimination performance on the holdout partition is determined in a one-time fashion, variance estimates are not applicable' is incorrect. A single holdout accuracy is a binomial proportion and has a well-defined sampling distribution. With holdout sizes of 10 to 20 subjects in Table I, the point estimates are highly variable; for example, 75% on 12 subjects has a 95% confidence interval of approximately 47% to 91%. The absence of holdout error bars is a major omission.
  4. [Tables I and II] The manuscript claims results on eleven datasets, but the main text reports results for only five, with the remaining six deferred to a supplementary table that is not included. The 'eleven datasets' claim cannot be verified from the manuscript. All dataset-level results and the supplementary tables and figures should be included, or the claim should be restricted to the datasets actually reported.
  5. [Table II] For all five main-text datasets, the selected embedding dimension d equals the number of time points T (Beijing d=236, COBRE d=150, MRN d=177, Stanford d=60, UM_2 d=222). Thus the reported experiments never actually reduce the temporal dimension. This is inconsistent with the stated goal of 'nonlinear dimensionality reduction' and with the description in Section II of reducing a time-series of length T to a smaller number of modes. The authors should explain why d=T is selected and how this affects the interpretation of the method.
minor comments (5)
  1. [II-B, Eqs. (2)-(4)] The notation for the Modified LLE multiple weight vectors is difficult to follow; in particular, the definition of the local alignment matrix is not clear about how the multiple weight vectors are aggregated into Phi. Please clarify.
  2. [Section III] The sentence 'Tables I and II demonstrate...' is misleading because Table II is a dataset summary, not a results table; the results sentence should refer only to Table I.
  3. [Section II-C] The procedure for constructing a 'representative sample' training partition is not specified. Please describe how the split was chosen, particularly for the unbalanced datasets.
  4. [Methods II-b and Appendix] The paper repeatedly invokes the Cauchy stress tensor, Gauss' Principle of Least Constraint, analytic capacity, and Theorema Egregium, but no derivation connects these concepts to the LLE equations. These interpretive claims are not needed for the classification pipeline and should either be substantiated with a formal argument or removed from the Methods section.
  5. [Throughout] The manuscript contains numerous typographical and OCR-like artifacts (e.g., 'Index T erms', 'One hundred and fourty-eight years', and broken math symbols in the equations) and would benefit from a careful proofreading pass.

Circularity Check

2 steps flagged · score 6.0 of 10

The abstract's >80% performance claim is the same LOOCV accuracy used to select d and the diagnostic volumes, so that headline reduces to the selection objective; only the sparse holdout rows are independent.

  1. fitted input called prediction [Section II-C (Evaluation Criteria), Table I; abstract Findings]
    "LOOCV is used on the training data to find both the best reconstruction parameter d, and the diagnostic volumes that produce the highest accuracy for this d."

    The headline finding of >80% accuracy on eleven datasets is taken from the LOOCV accuracy on the training partition (Table I, Results). However, the same LOOCV accuracy is the objective used to select the reconstruction dimension d (searched on a log scale from 1 to T) and the SFS-selected diagnostic volumes. The reported training accuracy is therefore the maximum of a selection search over model choices, not an independent estimate of generalization: the selection criterion and the reported performance metric are the same quantity by construction.

  2. other [Section II-A (Feature & Parameter Selection)]
    "Note that the selection of time points is based upon cross validation and does not induce any biased sampling."

    This sentence is the paper's explicit defense of treating the training LOOCV accuracy as unbiased evidence. It asserts away the selection effect that makes that accuracy circular: the time points and d are chosen by maximizing the same LOOCV accuracy that is later reported as 'highly diagnostic performance.' Selection on the same folds cannot remove the inflation; the only non-circular numbers are the holdout rows of Table I, where LLE accuracy is 75%, 100%, 68.8%, 70%, and 80%, with three values below the advertised >80%.

full rationale

The paper's central quantitative claim is its training LOOCV accuracy after the same LOOCV loop has selected both d and the SFS diagnostic volumes. That makes the headline performance a fitted quantity by construction, not a prediction. The paper does report independent holdout accuracies, and those are not circular; however, the abstract and Results section base the >80% claim on the eleven-dataset LOOCV figures rather than on the five holdout rows, and no significance test is supplied for the 'better than chance' comparison. The LLE reconstruction machinery itself is standard (Roweis and Saul), and the Cauchy-stress/Gauss-principle framing is interpretive rather than circular. There are no load-bearing self-citations or imported uniqueness theorems. Because one central claim reduces to its selection objective while some independent holdout content remains, the appropriate score is 6: partial circularity.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central empirical result is best interpreted as a pipeline evaluation, not a derivation. The formal axioms are mostly domain assumptions about BOLD and manifold structure, plus the ad hoc spatial-neighborhood choice. The main free parameters are the embedding dimension d, sphere radius r, and zero regularization; these are selected or fixed by hand rather than determined by the physics claims, so the claimed conservation story is not independently tested.

free parameters (3)
  • Embedding dimension d (LLE/PCA) = e.g., Beijing 236, COBRE 150, MRN 177, Stanford 60, UM_2 222
    Selected per dataset by LOOCV on training data over a log-scale from 1 to T; d is not fixed a priori and, for Beijing, equals T, meaning no temporal reduction.
  • Spatial neighbourhood radius r = 2 for all datasets
    The sphere radius defining K=(1+2r)^3-1 neighbours is a hand-set hyperparameter; the paper does not report sensitivity analysis.
  • LLE regularization xi = 0
    Set to zero in Eq. 2 to enforce positive-definiteness; no tuning or sensitivity analysis is reported.
assumptions (5)
  • domain assumption Every voxel waveform x in R^T is generated by a vector z in R^d lying on a low-dimensional manifold (Section II, second paragraph).
    Underlies the validity of LLE reconstruction; if BOLD time series do not lie near a locally linear low-dimensional manifold, the embedding may not preserve discriminative structure.
  • domain assumption BOLD signal is directly and monotonically related to local neural activity, with a linear relationship for short stimulus presentations (Introduction, citing Logothetis et al.).
    Justifies interpreting embedded spatial patterns as neural activity; the relationship is not established for all task and resting-state regimes used here.
  • domain assumption The class-conditional densities of the embedded diagnostic volumes are multivariate Gaussian with a common covariance matrix, justifying LDA (Section II-B).
    If covariances differ across classes or the densities are non-Gaussian, the linear decision boundary may be suboptimal, though classification could still work empirically.
  • ad hoc to paper Local neighbourhood geometry is captured by voxels on a sphere of radius r with Pythagorean distance (Eq. 1).
    This is a specific operationalization of 'local' that is not derived from data; different neighbourhood definitions could change the results.
  • ad hoc to paper Gauss' Principle of Least Constraint and the Cauchy stress tensor provide a physically meaningful model of BOLD conservation (Methods II-b and Appendix).
    Invoked to frame the method as physically motivated, but the equations shown are standard LLE/MLLE; no derivation connects the physical claims to the classification result.
invented entities (1)
  • Voxel-wise Cauchy stress tensor interpretation of LLE weights
    purpose: Claims conservation of spatiotemporal properties via Gauss' Principle and analytic capacity; used to frame the method as physically motivated.
    The tensor is a reinterpretation of the reconstruction weights, not a new measured quantity or a falsifiable prediction; standard LLE equations do not require it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Locally Linear Embedding and fMRI feature selection in psychiatric classification." pith.science (2026). https://pith.science/paper/5JYVWTAK

@misc{pith2026190806319,
  author       = {Pith},
  title        = {Pith review of: Locally Linear Embedding and fMRI feature selection in psychiatric classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5JYVWTAK}},
  note         = {Machine review of arXiv:1908.06319}
}
read the original abstract

Background: Functional magnetic resonance imaging (fMRI) provides non-invasive measures of neuronal activity using an endogenous Blood Oxygenation-Level Dependent (BOLD) contrast. This article introduces a nonlinear dimensionality reduction (Locally Linear Embedding) to extract informative measures of the underlying neuronal activity from BOLD time-series. The method is validated using the Leave-One-Out-Cross-Validation (LOOCV) accuracy of classifying psychiatric diagnoses using resting-state and task-related fMRI. Methods: Locally Linear Embedding of BOLD time-series (into each voxel's respective tensor) was used to optimise feature selection. This uses Gau\ss' Principle of Least Constraint to conserve quantities over both space and time. This conservation was assessed using LOOCV to greedily select time points in an incremental fashion on training data that was categorised in terms of psychiatric diagnoses. Findings: The embedded fMRI gave highly diagnostic performances (> 80%) on eleven publicly-available datasets containing healthy controls and patients with either Schizophrenia, Attention-Deficit Hyperactivity Disorder (ADHD), or Autism Spectrum Disorder (ASD). Furthermore, unlike the original fMRI data before or after using Principal Component Analysis (PCA) for artefact reduction, the embedded fMRI furnished significantly better than chance classification (defined as the majority class proportion) on ten of eleven datasets Interpretation: Locally Linear Embedding appears to be a useful feature extraction procedure that retains important information about patterns of brain activity distinguishing among psychiatric cohorts.

Figures

Figures reproduced from arXiv: 1908.06319 by the authors.

Figure 1
Figure 1. Statistical maps illustrating the individual differences in mental activity (schizophrenic patients versus healthy controls) for the discriminative time points determined on the training partition [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Statistical maps illustrating the individual differences in mental activity (ADHD patients versus healthy controls) for the discriminative time points determined on the training partition. distributed observations [33]. Given that discrimination perfor￾mance on the holdout partition is determined in a one-time fashion, variance estimates are not applicable. The Harvard-Oxford Subcortical/Cortical and Cerebellum at￾l… view at source ↗
Figure 3
Figure 3. The statistical maps illustrating the individual differences in mental activity (ASD patients versus healthy controls) for the discriminative time points determined on the training partition. cortex [66], was observed in autistic children. Furthermore, this temporal hypoperfusion was individually identifiable in 75% of autistic children [66] [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The local geometry of the Cauchy stress tensor and its relative spatiotemporal patterns on topology [7] [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Illustrating the reconstruction of the waveform at voxel i, located in the left cerebral cortex of a schizophrenic patient performing the Sternberg Item Recognition Paradigm (SIRP), as a linear combination of the Lebesgue measures [36] wi,1, . . . , wi,|N (i)| (defined…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 68 canonical work pages

  1. [1]

    H., G/a.sc/s.sc/h.sc, D

    A/n.sc/d.sc/e.sc/r.sc/s.sc/e.sc/n.sc, A. H., G/a.sc/s.sc/h.sc, D. M., /a.sc/n.sc/d.sc A/v.sc/i.sc/s.sc/o.sc/n.sc, M. J.Principal Component Analysis of the dynamic response measured by fMRI: a generalized linear systems framework. Magnetic Resonance Imaging , 17(6):795 – 815, 1999

    work page 1999

  2. [2]

    C.Time as a fourth dimension

    A/r.sc/c.sc/h.sc/i.sc/b.sc/a.sc/l.sc/d.sc, R. C.Time as a fourth dimension. Bulletin of the American Mathematical Society 20 , (8):409–412, 1914

    work page 1914

  3. [3]

    apud Joannem Billium, Typographum Regium, 1620

    B/a.sc/c.sc/o.sc/n.sc, F.Francisci de Verulamio, summi Angliae cancellarii, Instauratio magna. apud Joannem Billium, Typographum Regium, 1620

    work page

  4. [4]

    Impensis Thurnisiorum, Fraetum, 1713

    B/e.sc/r.sc/n.sc/o.sc/u.sc/l.sc/l.sc/i.sc, J.Ars Conjectandi . Impensis Thurnisiorum, Fraetum, 1713

    work page

  5. [5]

    M.Pattern Recognition and Machine Learning

    B/i.sc/s.sc/h.sc/o.sc/p.sc, C. M.Pattern Recognition and Machine Learning . Springer, 2006

    work page 2006

  6. [6]

    Wiley, 1979

    B/i.sc/l.sc/l.sc/i.sc/n.sc/g.sc/s.sc/l.sc/e.sc/y.sc, P .Probability and measure . Wiley, 1979

    work page 1979

  7. [7]

    Topologie algébrique

    B/o.sc/u.sc/r.sc/b.sc/a.sc/k.sc/i.sc, N.Éléments de mathématique. Topologie algébrique. Chapitres 1 à 4 . Springer-Verlag Berlin Heidelberg, Paris, 2016

    work page 2016

  8. [8]

    Mathematische Annalen , 67(3):355–386, 1909

    C/a.sc/r.sc/a.sc/t.sc/h.sc/eacute.sc/o.sc/d.sc/o.sc/r.sc/y.sc, C.Untersuchungen über die grundlagen der thermodynamik. Mathematische Annalen , 67(3):355–386, 1909

    work page 1909

Show all 68 references
  1. [9]

    L.De la pression ou tension dans un corps solide

    C/a.sc/u.sc/c.sc/h.sc/y.sc, A. L.De la pression ou tension dans un corps solide. Exercices de Mathèmatiques 2 (1827), 42–56

  2. [10]

    C/h.sc/u.sc/n.sc/g.sc, F. R. K.Spectral Graph Theory . American Mathematical Society, 1997

    1997

  3. [11]

    H., L/e.sc/i.sc/s.sc/e.sc/r.sc/s.sc/o.sc/n.sc, C

    C/o.sc/r.sc/m.sc/e.sc/n.sc, T. H., L/e.sc/i.sc/s.sc/e.sc/r.sc/s.sc/o.sc/n.sc, C. E., R/i.sc/v.sc/e.sc/s.sc/t.sc, R. L., /a.sc/n.sc/d.sc S/t.sc/e.sc/i.sc/n.sc, C.Introduction to Algorithms (2. ed.) . MIT Press, 2001

    2001

  4. [12]

    time points

    enables calculation of these bottom (/d.math+ 1) eigenvectors. Each eigenvector represents a degree of freedom in space and time, where the (/d.math+1) eigenvector is the global unit vector that fills three-dimensional space. The global unit vector is discarded to enforce the c...

    1972

  5. [13]

    Bulletin of the American Mathematical Society 49 , (1):1–23, 1943

    C/o.sc/u.sc/r.sc/a.sc/n.sc/t.sc, R.Variational methods for the solution of problems of equilibrium and vibrations. Bulletin of the American Mathematical Society 49 , (1):1–23, 1943

    1943

  6. [14]

    C/o.sc/x.sc/e.sc/t.sc/e.sc/r.sc, H. S. M.Introduction to geometry (2. ed.) . Wiley, 1989

    1989

  7. [15]

    London: Murray, 1871

    D/a.sc/r.sc/w.sc/i.sc/n.sc, C.The Descent of Man . London: Murray, 1871

  8. [16]

    D/e.sc/e.sc/l.sc/e.sc/y.sc, E. M. /a.sc/n.sc/d.sc M/a.sc/c.scK/a.sc/y.sc, D. M.Multiplication and division by electronic- analogue methods Nature 163 , 4147 (4 1949), 650–650

    1949

  9. [17]

    P ., M/a.sc/g.sc/n.sc/o.sc/t.sc/t.sc/a.sc, V

    D/e.sc/m.sc/i.sc/r.sc/c.sc/i.sc, O., C/l.sc/a.sc/r.sc/k.sc, V. P ., M/a.sc/g.sc/n.sc/o.sc/t.sc/t.sc/a.sc, V. A., A/n.sc/d.sc/r.sc/e.sc/a.sc/s.sc/e.sc/n.sc, N. C., L/a.sc/u.sc/r.sc/i.sc/e.sc/l.sc/l.sc/o.sc, J., K/i.sc/e.sc/h.sc/l.sc, K. A., P/e.sc/a.sc/r.sc/l.sc/s.sc/o.sc/n.sc,...

    2008

  10. [18]

    A.The Use of Multiple Measurements in T axonomic Problems

    F/i.sc/s.sc/h.sc/e.sc/r.sc, R. A.The Use of Multiple Measurements in T axonomic Problems. Annals of Eugenics 7 , 2 (1936), 179–188

    1936

  11. [19]

    A.The Statistical Utilization of Multiple Measurements

    F/i.sc/s.sc/h.sc/e.sc/r.sc, R. A.The Statistical Utilization of Multiple Measurements. Annals of Eugenics 8 , 4 (1938), 376–386

    1938

  12. [20]

    J.The free-energy principle: a unified brain theory? Nature Reviews Neuroscience 11 , 2 (2010), 127–138

    F/r.sc/i.sc/s.sc/t.sc/o.sc/n.sc, K. J.The free-energy principle: a unified brain theory? Nature Reviews Neuroscience 11 , 2 (2010), 127–138

    2010

  13. [21]

    R., S/c.sc/h.sc/i.sc/f.sc/f.sc/e.sc/r.sc, M.On existence theorems of potential theory and conformal mapping

    G/a.sc/r.sc/a.sc/b.sc/e.sc/d.sc/i.sc/a.sc/n.sc, P . R., S/c.sc/h.sc/i.sc/f.sc/f.sc/e.sc/r.sc, M.On existence theorems of potential theory and conformal mapping. Annals of Mathematics , 52(1):164–187, 1950

    1950

  14. [22]

    F.Disquisitiones generales circa superficies curvas

    G/a.sc/u.sc/germandbls.sc, C. F.Disquisitiones generales circa superficies curvas . Typis Ditericianis, Germany, 1828

  15. [23]

    F.Theoria motus corporum coelestium in sectionibus conicis solem ambientium

    G/a.sc/u.sc/germandbls.sc, C. F.Theoria motus corporum coelestium in sectionibus conicis solem ambientium. Perthes et Besser, 1809

  16. [24]

    F.Ueber ein allgemeines grundgesetz der mechanik

    G/a.sc/u.sc/germandbls.sc, C. F.Ueber ein allgemeines grundgesetz der mechanik. Journal für die reine und angewandte Mathematik (1829), 232–235

  17. [25]

    J.Fifty years of the vsepr model

    G/i.sc/l.sc/l.sc/e.sc/s.sc/p.sc/i.sc/e.sc, R. J.Fifty years of the vsepr model. Coordination Chemistry Reviews, 252(12–14):1315–1327, 7 2008

    2008

  18. [26]

    Patavii: Typis heredum Pauli Frambotti, Italy, 1668

    G/r.sc/e.sc/g.sc/o.sc/r.sc/y.sc, J.Geometriae pars universalis. Patavii: Typis heredum Pauli Frambotti, Italy, 1668

  19. [27]

    H/a.sc/s.sc/t.sc/i.sc/e.sc, T., T/i.sc/b.sc/s.sc/h.sc/i.sc/r.sc/a.sc/n.sc/i.sc, R., /a.sc/n.sc/d.sc F/r.sc/i.sc/e.sc/d.sc/m.sc/a.sc/n.sc, J.The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition , 2nd ed

  20. [28]

    H., /a.sc/n.sc/d.sc F/r.sc/a.sc/n.sc/z.sc/eacute.sc/n.sc, G.Distribution of cerebral activity in chronic schizophrenia

    I/n.sc/g.sc/v.sc/a.sc/r.sc, D. H., /a.sc/n.sc/d.sc F/r.sc/a.sc/n.sc/z.sc/eacute.sc/n.sc, G.Distribution of cerebral activity in chronic schizophrenia. The Lancet 304 , 7895 (12 1974), 1484–1486

    1974

  21. [29]

    H/i.sc/l.sc/b.sc/e.sc/r.sc/t.sc, D.Ueber die stetige Abbildung einer Line auf ein Flächenstück Mathematische Annalen 3 38 (1891), 459–460

  22. [30]

    F., B/e.sc/h.sc/r.sc/e.sc/n.sc/s.sc, T

    J/e.sc/n.sc/k.sc/i.sc/n.sc/s.sc/o.sc/n.sc, M., B/e.sc/c.sc/k.sc/m.sc/a.sc/n.sc/n.sc, C. F., B/e.sc/h.sc/r.sc/e.sc/n.sc/s.sc, T. E. J., W/o.sc/o.sc/l.sc/r.sc/i.sc/c.sc/h.sc, M. W., /a.sc/n.sc/d.sc S/m.sc/i.sc/t.sc/h.sc, S. M. Fsl. Neuroimage 62 , 2 (Aug 2012), 782–790

    2012

  23. [31]

    H., /a.sc/n.sc/d.sc L/a.sc/s.sc/s.sc/e.sc/n.sc, N

    I/n.sc/g.sc/v.sc/a.sc/r.sc, D. H., /a.sc/n.sc/d.sc L/a.sc/s.sc/s.sc/e.sc/n.sc, N. A.Quantitative determination of regional cerebral blood-flow in man. The Lancet 278 , 7206 (10 1961), 806–807

    1961

  24. [32]

    K/o.sc/c.sc/h.sc, C., /a.sc/n.sc/d.sc H/e.sc/p.sc/p.sc, K.Quantum mechanics in the brain Nature 440 , 7084 (2006), 611-612

    2006

  25. [33]

    Springer Verlag, New Y ork, 2002

    J/o.sc/l.sc/l.sc/i.sc/f.sc/f.sc/e.sc, I.Principal component analysis . Springer Verlag, New Y ork, 2002

    2002

  26. [34]

    A., /a.sc/n.sc/d.sc I/n.sc/g.sc/v.sc/a.sc/r.sc, D

    L/a.sc/s.sc/s.sc/e.sc/n.sc, N. A., /a.sc/n.sc/d.sc I/n.sc/g.sc/v.sc/a.sc/r.sc, D. H.; P/o.sc/t.sc/c.sc/h.sc/e.sc/n.sc, E. J., M/c.scC/r.sc/e.sc/a.sc/d.sc/y.sc, V. R.(Eds.) Radioisotopic assessment of regional cerebral blood flow. Progress in Nuclear Medicine 1: Neuro-Nuclear Me...

    1972

  27. [35]

    Courcier, Paris, 1812

    L/a.sc/p.sc/l.sc/a.sc/c.sc/e.sc, P .Théorie analytique des probabilités . Courcier, Paris, 1812

  28. [36]

    Collection de monographies sur la théorie des fonctions

    L/e.sc/b.sc/e.sc/s.sc/g.sc/u.sc/e.sc, H.Leçons sur l’intégration et la recherche des fonctions primitives: professées au Collège de France . Collection de monographies sur la théorie des fonctions. Gauthier-Villars, 1904

    1904

  29. [37]

    D.A unifying probabilistic perspective for spectral dimen- sionality reduction: Insights and new models

    L/a.sc/w.sc/r.sc/e.sc/n.sc/c.sc/e.sc, N. D.A unifying probabilistic perspective for spectral dimen- sionality reduction: Insights and new models. Journal of Machine Learning Research 13 (June 2012), 1609–1638

    2012

  30. [38]

    Thomas Basset, 1690

    L/o.sc/c.sc/k.sc/e.sc, J.An Essay Concerning Humane Understanding . Thomas Basset, 1690

  31. [39]

    F., F/r.sc/i.sc/s.sc/t.sc/o.sc/n.sc, K

    L/i.sc/d.sc/d.sc/l.sc/e.sc, P . F., F/r.sc/i.sc/s.sc/t.sc/o.sc/n.sc, K. J., F/r.sc/i.sc/t.sc/h.sc, C. D., H/i.sc/r.sc/s.sc/c.sc/h.sc, S. R., J/o.sc/n.sc/e.sc/s.sc, T., /a.sc/n.sc/d.sc F/r.sc/a.sc/c.sc/k.sc/o.sc/w.sc/i.sc/a.sc/k.sc, R. S.Patterns of cerebral blood flow in schizo...

    1992

  32. [40]

    K.What we can do and what we cannot do with fMRI

    L/o.sc/g.sc/o.sc/t.sc/h.sc/e.sc/t.sc/i.sc/s.sc, N. K.What we can do and what we cannot do with fMRI. Nature 453 , 7197 (6 2008), 869–878

    2008

  33. [41]

    Neurophysiological investigation of the basis of the fMRI signal

    L/o.sc/g.sc/o.sc/t.sc/h.sc/e.sc/t.sc/i.sc/s.sc, N., P/a.sc/u.sc/l.sc/s.sc, J., A/u.sc/g.sc/a.sc/t.sc/h.sc, M., T/r.sc/i.sc/n.sc/a.sc/t.sc/h.sc, T., /a.sc/n.sc/d.sc O/e.sc/l.sc/t.sc/e.sc/r.sc/m.sc/a.sc/n.sc/n.sc, A.. Neurophysiological investigation of the basis of the fMRI sig...

    2001

  34. [42]

    M/a.sc/c.scK/a.sc/y.sc, D. J. C.Information Theory, Inference, and Learning Algorithms Cambridge University Press, 2003

    2003

  35. [43]

    C., H/e.sc/n.sc/r.sc/i.sc/k.sc/s.sc/e.sc/n.sc, L., /a.sc/n.sc/d.sc B/r.sc/u.sc/h.sc/n.sc, P .Focal cerebral dysfunction in developmental learning disabilities

    L/o.sc/u.sc, H. C., H/e.sc/n.sc/r.sc/i.sc/k.sc/s.sc/e.sc/n.sc, L., /a.sc/n.sc/d.sc B/r.sc/u.sc/h.sc/n.sc, P .Focal cerebral dysfunction in developmental learning disabilities. The Lancet 335 , 8680 (1 1990), 8–11

    1990

  36. [44]

    Dimensionality reduction of fMRI time series data using locally linear embedding

    M/a.sc/n.sc/n.sc/f.sc/o.sc/l.sc/k.sc, P ., W/i.sc/r.sc/e.sc/s.sc/t.sc/a.sc/m.sc, R., N/i.sc/l.sc/s.sc/s.sc/o.sc/n.sc, M., S/t.sc/a.sc/h.sc/l.sc/b.sc/e.sc/r.sc/g.sc, F., /a.sc/n.sc/d.sc O/l.sc/s.sc/r.sc/u.sc/d.sc, J. Dimensionality reduction of fMRI time series data using local...

    2010

  37. [45]

    M.A high-speed electronic function generator Nature 159, 4038 (3 1947), 406–407

    M/a.sc/c.scK/a.sc/y.sc, D. M.A high-speed electronic function generator Nature 159, 4038 (3 1947), 406–407

    1947

  38. [46]

    S.A System of Logic, Ratiocinative and Inductive: Being a Connected View of the Principles of Evidence and the Methods of Scientific Investigation, Volume I

    M/i.sc/l.sc/l.sc, J. S.A System of Logic, Ratiocinative and Inductive: Being a Connected View of the Principles of Evidence and the Methods of Scientific Investigation, Volume I. John W. Parker, 1843

  39. [47]

    S.Analytic capacity: discrete approach and curvature of measure Sbornik: Mathematics 186 , (6):827-846, 1995

    M/e.sc/l.sc’/n.sc/i.sc/k.sc/o.sc/v.sc, M. S.Analytic capacity: discrete approach and curvature of measure Sbornik: Mathematics 186 , (6):827-846, 1995

    1995

  40. [48]

    N/e.sc/w.sc/t.sc/o.sc/n.sc, I.Philosophiae naturalis principia mathematica . J. Societatis Regiae ac Typis J. Streater, 1687

  41. [49]

    Physikalische Zeitschrift 10 (1908), 75–88

    M/i.sc/n.sc/k.sc/o.sc/w.sc/s.sc/k.sc/i.sc, H.Raum und Zeit. Physikalische Zeitschrift 10 (1908), 75–88

    1908

  42. [50]

    Mathematical Proceedings of the Cambridge Philosophical Society , 51:406–413, 7 1955

    P/e.sc/n.sc/r.sc/o.sc/s.sc/e.sc, R.A generalized inverse for matrices. Mathematical Proceedings of the Cambridge Philosophical Society , 51:406–413, 7 1955

    1955

  43. [51]

    M., K/a.sc/y.sc, A

    O/g.sc/a.sc/w.sc/a.sc, S., L/e.sc/e.sc, T. M., K/a.sc/y.sc, A. R., /a.sc/n.sc/d.sc T/a.sc/n.sc/k.sc, D. W.Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proceedings of the National Academy of Sciences 87 , 24 (1990), 9868–9872

    1990

  44. [52]

    R/a.sc/y.sc/l.sc/e.sc/i.sc/g.sc/h.sc, J. W. S.The theory of sound . MacMillan and Co. London, 1877

  45. [53]

    G.; P/o.sc/t.sc/c.sc/h.sc/e.sc/n.sc, E

    P/o.sc/p.sc/h.sc/a.sc/m.sc, M. G.; P/o.sc/t.sc/c.sc/h.sc/e.sc/n.sc, E. J., M/c.scC/r.sc/e.sc/a.sc/d.sc/y.sc, V. R.(Eds.) Numerical Methods for the Detection of Abnormalities in Radionuclide Brain Scans. Progress in Nuclear Medicine 1: Neuro-Nuclear Medicine . University Park P...

    1972

  46. [54]

    T., /a.sc/n.sc/d.sc S/a.sc/u.sc/l.sc, L

    R/o.sc/w.sc/e.sc/i.sc/s.sc, S. T., /a.sc/n.sc/d.sc S/a.sc/u.sc/l.sc, L. K.Nonlinear dimensionality reduction by locally linear embedding. Science 290 , 5500 (2000), 2323–2326

    2000

  47. [55]

    N., /a.sc/n.sc/d.sc H/e.sc/l.sc/z.sc/e.sc/r.sc, J

    R/o.sc/b.sc/i.sc/n.sc/s.sc, L. N., /a.sc/n.sc/d.sc H/e.sc/l.sc/z.sc/e.sc/r.sc, J. E.Diagnosis and clinical assessment: The current state of psychiatric diagnosis. Annual Review of Psychology 37 , 1 (1986), 409 – 432

    1986

  48. [56]

    K., /a.sc/n.sc/d.sc R/o.sc/w.sc/e.sc/i.sc/s.sc, S

    S/a.sc/u.sc/l.sc, L. K., /a.sc/n.sc/d.sc R/o.sc/w.sc/e.sc/i.sc/s.sc, S. T.Think globally, fit locally: unsupervised learning of low dimensional manifolds. Journal of Machine Learning Research 4 (Dec. 2003), 119–155

    2003

  49. [57]

    S., /a.sc/n.sc/d.sc S/h.sc/e.sc/r.sc/r.sc/i.sc/n.sc/g.sc/t.sc/o.sc/n.sc, C

    R/o.sc/y.sc, C. S., /a.sc/n.sc/d.sc S/h.sc/e.sc/r.sc/r.sc/i.sc/n.sc/g.sc/t.sc/o.sc/n.sc, C. S.On the regulation of the blood-supply of the brain. Journal of Physiology 11 , 1-2 (Jan 1890), 85–158

  50. [58]

    S/c.sc/h.sc/m.sc/i.sc/d.sc/t.sc, E.Über die Auflösung linearer Gleichungen mit Unendlich vielen unbekannten Rendiconti del Circolo Matematico di Palermo 25 1 (1908), 53–77

    1908

  51. [59]

    S., /a.sc/n.sc/d.sc L/e.sc/e.sc, D

    S/e.sc/u.sc/n.sc/g.sc, H. S., /a.sc/n.sc/d.sc L/e.sc/e.sc, D. D.The manifold ways of perception. Science 290 , 5500 (2000), 2268–2269

    2000

  52. [60]

    S., /a.sc/n.sc/d.sc S/p.sc/i.sc/e.sc/r.sc/s.sc, H

    S/i.sc/m.sc/o.sc/n.sc/s.sc, J. S., /a.sc/n.sc/d.sc S/p.sc/i.sc/e.sc/r.sc/s.sc, H. J.Prefrontal and medial temporal lobe interactions in long-term memory. Nature Reviews Neuroscience 4 , 8 (08 2003), 637–648

    2003

  53. [61]

    S.Man on His Nature

    S/h.sc/e.sc/r.sc/r.sc/i.sc/n.sc/g.sc/t.sc/o.sc/n.sc, C. S.Man on His Nature . Gifford lectures, Edinburgh. University Press, 1951

    1951

  54. [62]

    S/y.sc/l.sc/v.sc/e.sc/s.sc/t.sc/e.sc/r.sc, J. J.A demonstration of the theorem that every homogeneous quadratic polynomial is reducible by real orthogonal substitutions to the form of a sum of positive and negative squares Philosophical Magazine Series 4 (4) 23 (1852), 138–142

  55. [63]

    Science 153 , 3736 (1966), 652–654

    S/t.sc/e.sc/r.sc/n.sc/b.sc/e.sc/r.sc/g.sc, S.High-speed scanning in human memory. Science 153 , 3736 (1966), 652–654

    1966

  56. [64]

    W/i.sc/l.sc/l.sc/i.sc/a.sc/m.sc/s.sc, J., G/i.sc/b.sc/b.sc/o.sc/n.sc, M., F/i.sc/r.sc/s.sc/t.sc, M., S/p.sc/i.sc/t.sc/z.sc/e.sc/r.sc, R., D/a.sc/v.sc/i.sc/e.sc/s.sc, M., B/o.sc/r.sc/u.sc/s.sc, J., H/o.sc/w.sc/e.sc/s.sc, M., K/a.sc/n.sc/e.sc, J., P/o.sc/p.sc/e.sc, H., /a.sc/n.s...

    1992

  57. [65]

    Computers, IEEE Transactions on C-20 , 9 (sept

    W/h.sc/i.sc/t.sc/n.sc/e.sc/y.sc, A.A direct method of nonparametric measurement selection. Computers, IEEE Transactions on C-20 , 9 (sept. 1971), 1100 – 1103

    1971

  58. [66]

    American Journal of Psychiatry 157 , 12 (2000), 1988– 1993

    Z/i.sc/l.sc/b.sc/o.sc/v.sc/i.sc/c.sc/i.sc/u.sc/s.sc, M., B/o.sc/d.sc/d.sc/a.sc/e.sc/r.sc/t.sc, N., B/e.sc/l.sc/i.sc/n.sc, P ., P/o.sc/l.sc/i.sc/n.sc/e.sc, J., R/e.sc/m.sc/y.sc, P ., M/a.sc/n.sc/g.sc/i.sc/n.sc, J., T/h.sc/i.sc/v.sc/a.sc/r.sc/d.sc, L., B/a.sc/r.sc/t.sc/h.sc/eacu...

    2000

  59. [67]

    /a.sc/n.sc/d.sc W/a.sc/n.sc/g.sc, J.MLLE: Modified Locally Linear Embedding Using Multiple Weights Advances in Neural Information Processing Systems 19 (2007): 1593

    Z/h.sc/a.sc/n.sc/g.sc, Z. /a.sc/n.sc/d.sc W/a.sc/n.sc/g.sc, J.MLLE: Modified Locally Linear Embedding Using Multiple Weights Advances in Neural Information Processing Systems 19 (2007): 1593

    2007

  60. [2009]

    3rd printing 5th printing

    corr. 3rd printing 5th printing. ed. Springer Series in Statistics. Springer, Sept. 2009

    2009

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.