pith. sign in

arxiv: 2207.06229 · v3 · submitted 2022-07-11 · 📊 stat.ML · cs.LG· math.FA· math.PR· math.ST· stat.CO· stat.TH

Distribution-Free Stochastic Analysis and Robust Multilevel Vector Field Anomaly Detection

Julio E Castrillon-Candas , Michael Rosenbaum , Mark Kon This is my paper

Pith reviewed 2026-05-24 11:56 UTC · model grok-4.3

classification 📊 stat.ML cs.LGmath.FAmath.PRmath.STstat.COstat.TH
keywords anomaly detectionvector fieldsdistribution-free methodsKarhunen-Loeve expansionmultilevel subspacesstochastic analysisremote sensingsatellite imagery
0
0 comments X

The pith

Covariance-based multilevel subspaces enable distribution-free hypothesis tests for anomalies in vector field data.

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

The paper establishes a stochastic functional analysis method that detects anomalies in high-dimensional vector field datasets by building on the covariance structure of normal behavior. An optimal vector field Karhunen-Loeve expansion is formed, followed by multilevel orthogonal functional subspaces derived from domain geometry. Projections onto these bases support hypothesis tests that require no assumptions about underlying probability distributions. This matters for applications like multi-spectral satellite imagery where distributions cannot be estimated, and the approach shows improved detection of subtle changes compared to PCA in simulations while incorporating multiple data bands.

Core claim

The paper claims that applying an optimal vector field Karhunen-Loeve expansion to random field data and constructing a series of multilevel orthogonal functional subspaces adapted from the KL expansion using the geometry of the domain allows anomaly detection via projection of the random field onto the multilevel basis. This produces reliable hypothesis tests without any prior assumptions on the probability distributions of the data. The method is demonstrated on Amazon forest degradation using vectorized multi-band imagery and on simulated data where it identifies subtle anomalies that PCA-based methods cannot detect.

What carries the argument

optimal vector field Karhunen-Loeve expansion and adapted multilevel orthogonal functional subspaces from domain geometry for projection-based detection

If this is right

  • Reliable hypothesis tests become possible for anomaly detection without needing to assume or estimate probability distributions.
  • Multiple bands of vector field data can be combined in a single complex-valued representation for stronger detection performance.
  • Subtle anomalies that remain invisible to PCA methods can be recovered through the multilevel subspace projections in controlled simulations.
  • The approach remains applicable to high-dimensional remote sensing problems where distributional assumptions are unrealistic.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same covariance-to-subspace construction could be applied to vector fields arising in fluid flow or medical imaging if stable nominal covariance estimates are available.
  • The multilevel structure may permit anomaly localization at varying spatial scales by examining projections at different subspace levels.
  • Computational cost could be further reduced by combining the KL-derived bases with sparse sampling techniques for very large domains.

Load-bearing premise

The covariance structure of nominal stochastic behavior across a domain can be used to construct an optimal vector field Karhunen-Loeve expansion and adapted multilevel orthogonal functional subspaces that enable distribution-free detection.

What would settle it

A collection of vector field observations containing known anomalies where the covariance-derived multilevel projections produce hypothesis test results no better than random guessing or standard PCA at identifying the anomalies.

Figures

Figures reproduced from arXiv: 2207.06229 by Julio E Castrillon-Candas, Mark Kon, Michael Rosenbaum.

Figure 1
Figure 1. Figure 1: Surface domain U constructed from 2-simplicies (triangles) in R 3 . The vector random field v(x, ω) ∈ L 2 P (Ω;L 2 (U; R q )) is defined over the domain U. As an example this could be satellite multi-spectral data over land. iv) Let Vn+1 = P(E) := span{χ l i }. We assume that Karhunen-Lo`eve eigenfunctions φi ∈ P(E) for all i = 1, . . . , M where N > M. From the set of of indicator functions in E, a multil… view at source ↗
Figure 2
Figure 2. Figure 2: Binary tree construction algorithm from the simplicies in S used to construct the domain U. Algorithm 2 is used to decide how the barycenters in S˜ (Bk l ) are split. The tree structure T is built from Algorithm 1 and 2 and is constructed recursively until there are at most n0 − 1 barycenters left in the cell Bl k . Once all the leaves are reached the algorithm stops. Notice that the list of barycenters S˜… view at source ↗
Figure 3
Figure 3. Figure 3: Cartoon example of the construction of a kd-tree from the triangular simplicies in T [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Deforestation sequence from Sentinel 2 satellite data. Four frames of Amazon forest at days 3335, 3484, 3774 and 3909 showing the clearing and regrowth of forest. Coordinate pixel at (108,66) is marked with a small blue box. Note that from the brown discoloration at day 3484 swaths of the trees are cleared. By day 3774 the forest vegetation grows back from nearby trees. However, as we will see, it does not… view at source ↗
Figure 5
Figure 5. Figure 5: Multilevel anomaly map for day 1304 (small anomalies). There are very few statistical changes relative to the training data set of the multilevel filter. The top image corresponds to the multilevel cells Bl k ∈ Bl on the patch of terrain for each level l = 0, 1, . . . , 3 overlain on the RGB map from the Landsat optical measurements. The corresponding multilevel spaces W1304 3 , . . . , W1304 0 are given f… view at source ↗
Figure 6
Figure 6. Figure 6: Multilevel anomaly map for day 1336 (large changes). In contrast to day 1304 we observe changes in large regions of the forest. This is reflected in the magnitude of of E l,k p in each cell Bl k and from the corresponding hypothesis tests. We observe that at level l = 2 many p-values of the coefficients d k l are below the significance level 10−2 . This becomes more pronounced for levels l = 1 and l = 0. d… view at source ↗
Figure 7
Figure 7. Figure 7: Multilevel projection coefficient anomaly map for scalar EVI Sentinel 2 data on a 75 × 75 land cover patch for days 3344, 3484 and 3704. For day 3344 the sizes of the anomalies E l,k p are shown for each cell B l,k p . On day 3484 a part of the tree cover in the forest is removed, so that magnitudes of the anomalies increase. By day 3704 the forest has largely recovered, but sizes of the anomalies are stil… view at source ↗
Figure 8
Figure 8. Figure 8: Multilevel coefficient probabilities for scalar EVI Sentinel 2 data for days 3344, 3484 and 3704. For day 3344 the p-values of the hypothesis test for each coefficient are largely above the significance level of α = 0.05. This is consistent with [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: Multilevel projection coefficient anomaly map for vector field Sentinel 2 data on a 75 × 75 landcover patch for days 3344, 3484 and 3704. Six spectral bands are used that include blue, green, red, near-infrared, shortwave infrared 1 and shortwave infrared 2. For visualization purposes the anomaly sizes are divided by 1000 and only printed in yellow if they are greater than 3. For day 3344 the size of the … view at source ↗
read the original abstract

Massive vector field datasets are common in multi-spectral optical and radar sensors, among many other emerging areas of application. We develop a novel stochastic functional (data) analysis approach for detecting anomalies based on the covariance structure of nominal stochastic behavior across a domain. An optimal vector field Karhunen-Loeve expansion is applied to such random field data. A series of multilevel orthogonal functional subspaces is constructed from the geometry of the domain, adapted from the KL expansion. Detection is achieved by examining the projection of the random field on the multilevel basis. A critical feature of this approach is that reliable hypothesis tests are formed, which do not require prior assumptions on probability distributions of the data. The method is applied to the important problem of degradation in the Amazon forest. Due to the complexity and high dimensionality of satellite imagery, it is not feasible to assume known distributions, nor to estimate them. In addition to providing reliable hypothesis tests, our approach shows the advantage of using multiple bands of data in a vectorized complex, leading to better anomaly detection. Furthermore, using simulated data, our approach is capable of detecting subtle anomalies that are impossible to detect with PCA-based methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript develops a stochastic functional analysis framework for anomaly detection in high-dimensional vector field data (e.g., multi-spectral satellite imagery). An optimal vector-field Karhunen-Loève expansion is constructed from the covariance operator of nominal data; multilevel orthogonal functional subspaces are then adapted from the domain geometry. Anomaly detection proceeds by examining projections of observed fields onto these subspaces, with the central claim that the resulting hypothesis tests are reliable and distribution-free. The method is illustrated on Amazon forest degradation and on simulated data, where it is reported to detect subtle anomalies undetectable by PCA while benefiting from multi-band vectorization.

Significance. If the distribution-free property can be rigorously established, the approach would offer a practical advance for anomaly detection in settings where distributional assumptions are untenable and dimensionality precludes density estimation. The emphasis on vector-valued multi-band data and the reported simulation advantage over PCA constitute concrete strengths.

major comments (1)
  1. [Abstract] Abstract (and §3–4, judging from the described construction): the claim that projections onto the adapted multilevel subspaces yield reliable, distribution-free hypothesis tests is load-bearing yet unsupported by any explicit device (permutation test, rank statistic, or pivotal quantity independent of the unknown law). The KL expansion produces uncorrelated coefficients, but for a general non-Gaussian random field their joint distribution still depends on higher-order moments; the geometric adaptation step alone does not remove this dependence, so type-I error control under arbitrary distributions is not guaranteed.
minor comments (1)
  1. Notation for the multilevel subspaces and the precise definition of the projection statistic should be introduced earlier and used consistently; a small simulation table comparing detection power versus PCA at fixed false-alarm rates would strengthen the empirical claim.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the thoughtful review and for identifying the central claim requiring stronger support. We address the major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract (and §3–4, judging from the described construction): the claim that projections onto the adapted multilevel subspaces yield reliable, distribution-free hypothesis tests is load-bearing yet unsupported by any explicit device (permutation test, rank statistic, or pivotal quantity independent of the unknown law). The KL expansion produces uncorrelated coefficients, but for a general non-Gaussian random field their joint distribution still depends on higher-order moments; the geometric adaptation step alone does not remove this dependence, so type-I error control under arbitrary distributions is not guaranteed.

    Authors: We agree that the manuscript does not supply an explicit device (e.g., permutation test, rank statistic, or distribution-free pivotal quantity) that would guarantee finite-sample type-I error control for arbitrary laws. The current text relies on the uncorrelated KL coefficients together with the geometric multilevel construction to assert distribution-freeness, but this is insufficient to control the joint distribution of the projections when higher-order moments are unknown. In the revision we will either (i) introduce a concrete, distribution-free testing procedure (such as a permutation test on the multilevel coefficients or a rank-based statistic) with a proof of validity, or (ii) revise the abstract and §§3–4 to state precisely the weaker guarantees that the method actually provides. We view this as a necessary clarification rather than a change in the core methodology. revision: yes

Circularity Check

0 steps flagged

No circularity; distribution-free claim rests on unshown but non-self-referential construction of pivotal statistics

full rationale

The abstract and provided excerpts describe constructing a vector-field KL expansion from the estimated covariance of nominal data, then adapting multilevel orthogonal subspaces from domain geometry to form projection-based hypothesis tests. No equations are exhibited that define the test statistic in terms of itself, rename a fitted parameter as a prediction, or reduce the distribution-free property to a self-citation chain. The central premise (projections yield reliable tests without distributional assumptions) is asserted without reduction to inputs by construction; any justification for pivotality would have to be supplied externally (e.g., via permutation or rank methods) rather than being tautological within the paper's own definitions. This is the normal case of a paper whose internal derivation chain does not collapse.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Ledger constructed from abstract claims only; full text unavailable.

axioms (1)
  • domain assumption Covariance structure of nominal stochastic behavior across a domain suffices to construct optimal multilevel orthogonal functional subspaces for anomaly detection
    Stated as the basis for the detection procedure in the abstract.

pith-pipeline@v0.9.0 · 5754 in / 1182 out tokens · 22339 ms · 2026-05-24T11:56:52.736242+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Karhunen Lo\`eve Expansions of Hilbert Space-Valued Random Elements

    math.FA 2026-04 unverdicted novelty 5.0

    Hilbert space-valued random elements admit Karhunen-Loève expansions precisely when they lie in the appropriate Bochner space and their covariance operator is Hilbert-Schmidt, via a natural isomorphism with computatio...

  2. Stochastic tensor space feature theory with applications to robust machine learning

    stat.ML 2021-10 unverdicted novelty 5.0

    Develops MOS Karhunen-Loeve features from stochastic tensor spaces to generate robust ML features from random fields, reporting high accuracy on Alzheimer's blood plasma data for predicting disease stages.

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · cited by 2 Pith papers · 1 internal anchor

  1. [1]

    Arias-Castro, D

    E. Arias-Castro, D. Donoho, and X. Huo. Near-optimal detection of geometric objects by fast multiscale methods. IEEE Transactions on Information Theory , 51:2402, 2005

    work page 2005

  2. [2]

    Castro, Ervin T´ anczos, and Meng Wang

    Ery Arias-Castro, Rui M. Castro, Ervin T´ anczos, and Meng Wang. Distribution-free detec- tion of structured anomalies: Permutation and rank-based scans. Journal of the American Statistical Association, 113(522):789–801, 2018

    work page 2018

  3. [3]

    A. Aue, S. H¨ ormann, L. Horv´ ath, M. Huˆ skov´ a, and J. G. Steinebach. Sequential testing for the stability of high-frequency portfolio betas,. Econometric Theory, 28:804, 2012

    work page 2012

  4. [4]

    Aue and L

    A. Aue and L. Horv´ ath. Structural breaks in time series. Journal of Time Series Analysis , 34:1, 2013

    work page 2013

  5. [5]

    Identifying human mobility patterns in the rio de janeiro metropolitan area using call detail records

    Matheus HC Barboza, Ricardo de S Alencar, Julio C Chaves, Moacyr AHB Silva, Ro- mulo D Orrico, and Alexandre G Evsukoff. Identifying human mobility patterns in the rio de janeiro metropolitan area using call detail records. Transportation Research Record, page 0361198120977655, 2020

    work page 2020

  6. [6]

    J. E. Castrill´ on-Cand´ as and K. Amaratunga. Fast estimation of continuous karhunen-loeve eigenfunctions using wavelets. IEEE Transactions on Signal Processing , 50(1):78–86, 2002

    work page 2002

  7. [7]

    J. E. Castrill´ on-Cand´ as and Kevin Amaratunga. Spatially adapted multiwavelets and sparse representation of integral equations on general geometries. SIAM Journal on Scientific Com- puting, 24(5):1530–1566, 2003

    work page 2003

  8. [8]

    J. E. Castrill´ on-Cand´ as, M. G. Genton, and R. Yokota. Multi-level restricted maximum likeli- hood covariance estimation and Kriging for large non-gridded spatial datasets. Spatial Statis- tics, 18, Part A:105 – 124, 2016. Spatial Statistics Avignon: Emerging Patterns

    work page 2016

  9. [9]

    Castrill´ on-Cand´ as and Mark Kon

    Julio E. Castrill´ on-Cand´ as and Mark Kon. Anomaly detection: A functional analysis perspec- tive. Journal of Multivariate Analysis , 189:104885, 2022

    work page 2022

  10. [10]

    Stochastic tensor space feature theory with applications to robust machine learning

    Julio Enrique Castrill´ on-Cand´ as, Dingning Liu, and Mark Kon. Stochastic functional analysis with applications to robust machine learning, 2021. 10.48550/ARXIV.2110.01729

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2110.01729 2021

  11. [11]

    Consequences of changing biodiversity

    F Stuart Chapin Iii, Erika S Zavaleta, Valerie T Eviner, Rosamond L Naylor, Peter M Vitousek, Heather L Reynolds, David U Hooper, Sandra Lavorel, Osvaldo E Sala, Sarah E Hobbie, et al. Consequences of changing biodiversity. Nature, 405(6783):234–242, 2000

    work page 2000

  12. [12]

    Delineating urban functional areas with building-level social media data: A dy- namic time warping (dtw) distance based k-medoids method

    Yimin Chen, Xiaoping Liu, Xia Li, Xingjian Liu, Yao Yao, Guohua Hu, Xiaocong Xu, and Fengsong Pei. Delineating urban functional areas with building-level social media data: A dy- namic time warping (dtw) distance based k-medoids method. Landscape and Urban Planning, 160:48–60, 2017

    work page 2017

  13. [13]

    Y. T. D. Cheung, M. J. Spittal, M. K. Williamson, S. J. Tung, and J. Pirkis. Application of scan statistics to detect suicide clusters in australia. PLoS ONE, 8:e54168, 2013

    work page 2013

  14. [14]

    C.-S. J. Chu, M. Stinchcombe, and H. White. Monitoring structural change. Econometrica, 64:1045, 1996

    work page 1996

  15. [15]

    Dasgupta and Y

    S. Dasgupta and Y. Freund. Random projection trees and low dimensional manifolds. In Proceedings of the Fortieth Annual ACM Symposium on Theory of Computing , STOC ’08, pages 537–546, New York, NY, USA, 2008. ACM. 22 STOCHASTIC FUNCTIONAL ANALYSIS AND MULTILEVEL VECTOR FIELD ANOMALY DETECTION

    work page 2008

  16. [16]

    Dette and J G¨ osmann

    H. Dette and J G¨ osmann. A likelihood ratio approach to sequential change point detection for a general class of parameters. Journal of the American Statistical Association , 115(531):1361– 1377, 2020

    work page 2020

  17. [17]

    D’Heedene, K

    S. D’Heedene, K. Amaratunga, and J. Castrill´ on-Cand´ as. Generalized hierarchical bases: a wavelet-ritz-galerkin framework for lagrangian fem. Engineering Computations, 22(1):15–37, Jan 2005

    work page 2005

  18. [18]

    Drusch, U

    M. Drusch, U. Del Bello, S. Carlier, O. Colin, V. Fernandez, F. Gascon, B. Hoersch, C. Isola, P. Laberinti, P. Martimort, A. Meygret, F. Spoto, O. Sy, F. Marchese, and P. Bargellini. Sentinel-2: Esa’s optical high-resolution mission for gmes operational services. Remote Sensing of Environment, 120:25–36, 2012. The Sentinel Missions - New Opportunities for Science

    work page 2012

  19. [19]

    S. Fremdt. Page’s sequential procedure for change-point detection in time series regression. Statistics, 48:1, 2014

    work page 2014

  20. [20]

    Optimal and fast detection of spatial clusters with scan statistics

    Walther G. Optimal and fast detection of spatial clusters with scan statistics. The Annals of Statistics, 38:1010, 2010

    work page 2010

  21. [21]

    Guerriero, P

    M. Guerriero, P. Willett, and J. Glaz. Distributed target detection in sensor networks using scan statistics. IEEE Transactions on Signal Processing , 57:2629, 2009

    work page 2009

  22. [22]

    Gps-based city- wide traffic congestion forecasting using cnn-rnn and c3d hybrid model

    Jingqiu Guo, Yangzexi Liu, Qingyan Yang, Yibing Wang, and Shouen Fang. Gps-based city- wide traffic congestion forecasting using cnn-rnn and c3d hybrid model. Transportmetrica A: Transport Science, pages 1–22, 2020

    work page 2020

  23. [23]

    Harbrecht, M

    H. Harbrecht, M. Peters, and M. Siebenmorgen. Analysis of the domain mapping method for elliptic diffusion problems on random domains. Numerische Mathematik, 134(4):823–856, 2016

    work page 2016

  24. [24]

    Spatiotem- poral patterns of urban human mobility

    Samiul Hasan, Christian M Schneider, Satish V Ukkusuri, and Marta C Gonz´ alez. Spatiotem- poral patterns of urban human mobility. Journal of Statistical Physics , 151(1):304–318, 2013

    work page 2013

  25. [25]

    D. V. Hinkley. Inference about the change-point from cumulative sum tests. Biometrika, 58:509, 1971

    work page 1971

  26. [26]

    Horv´ ath, M

    L. Horv´ ath, M. Huˆ skov´ a, P. Kokoszka, and J. Steinebach. Monitoring changes in linear models. Journal of Statistical Planning and Inference , 126:225, 2004

    work page 2004

  27. [27]

    Huete, K

    A. Huete, K. Didan, T. Miura, E.P. Rodriguez, X. Gao, and L.G. Ferreira. Overview of the radiometric and biophysical performance of the modis vegetation indices. Remote Sensing of Environment, 83(1):195–213, 2002. The Moderate Resolution Imaging Spectroradiometer (MODIS): a new generation of Land Surface Monitoring

    work page 2002

  28. [28]

    Jandhyala, S

    V. Jandhyala, S. Fotopoulos, I. MacNeill, and P. Liu. Inference for single and multiple change- points in time series. Journal of Time Series Analysis , 34:423, 2013

    work page 2013

  29. [29]

    Kirch and S

    C. Kirch and S. Weber. Modified sequential change point procedures based on estimating functions. Electronic Journal of Statistics , 12:1579, 2018

    work page 2018

  30. [30]

    T. L. Lai. Sequential changepoint detection in quality control and dynamical systems. Journal of the Royal Statistical Society, Series B , 57:613, 1995

    work page 1995

  31. [31]

    Diagnosing network-wide traffic anom- alies

    Anukool Lakhina, Mark Crovella, and Christophe Diot. Diagnosing network-wide traffic anom- alies. SIGCOMM Comput. Commun. Rev. , 34(4):219–230, aug 2004

    work page 2004

  32. [32]

    Light and E.W

    W.A. Light and E.W. Cheney. Approximation theory in tensor product spaces. , volume 1169 of Lecture notes in mathematics . Springer, New York, 1985

    work page 1985

  33. [33]

    A spatial scan statistic

    Kulldorff M. A spatial scan statistic. Communications in Statistics: Theory and Methods , 26:1481, 1997

    work page 1997

  34. [34]

    version 9.4 (R2021a)

    MATLAB. version 9.4 (R2021a) . The MathWorks Inc., Natick, Massachusetts, 2021

    work page 2021

  35. [35]

    G. V. Moustakides. Optimal stopping times for detecting changes in distributions. The Annals of Statistics, 14:1379, 1986

    work page 1986

  36. [36]

    D. B. Neil and A. W. Moore. Rapid detection of significant spatial clusters. Proceedings of the Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , page 256, 2004. STOCHASTIC FUNCTIONAL ANALYSIS AND MULTILEVEL VECTOR FIELD ANOMALY DETECTION 23

    work page 2004

  37. [37]

    D. B. Neill. Fast subset scan for spatial pattern detection. Journal of the Royal Statistical Society, 74:337, 2012

    work page 2012

  38. [38]

    E. S. Page. Continuous inspection schemes. Biometrika, 41:100, 1954

    work page 1954

  39. [39]

    K. Pape, D. Wied, and P. Galeano. Monitoring multivariate variance changes. Journal of Empirical Finance, 39:54, 2016

    work page 2016

  40. [40]

    Smartbuddy: defining human behaviors using big data analytics in social internet of things

    Anand Paul, Awais Ahmad, M Mazhar Rathore, and Sohail Jabbar. Smartbuddy: defining human behaviors using big data analytics in social internet of things. IEEE Wireless commu- nications, 23(5):68–74, 2016

    work page 2016

  41. [41]

    Exploring urban spatial features of covid-19 transmission in wuhan based on social media data

    Zhenghong Peng, Ru Wang, Lingbo Liu, and Hao Wu. Exploring urban spatial features of covid-19 transmission in wuhan based on social media data. ISPRS International Journal of Geo-Information, 9(6):402, 2020

    work page 2020

  42. [42]

    Traffic congestion analysis visualisation tool

    Natasha Petrovska and Aleksandar Stevanovic. Traffic congestion analysis visualisation tool. In 2015 IEEE 18th International Conference on Intelligent Transportation Systems , pages 1489–1494. IEEE, 2015

    work page 2015

  43. [43]

    Methods of modern mathematical physics 1: Functional Analysis

    M Reed and B Simon. Methods of modern mathematical physics 1: Functional Analysis . Academic Press, New York, New York, 1972

    work page 1972

  44. [44]

    The role of local land-use on the urban heat island effect of tel aviv as assessed from satellite remote sensing

    Orit Rotem-Mindali, Yaron Michael, David Helman, and Itamar M Lensky. The role of local land-use on the urban heat island effect of tel aviv as assessed from satellite remote sensing. Applied Geography, 56:145–153, 2015

    work page 2015

  45. [45]

    Roy, M.A

    D.P. Roy, M.A. Wulder, T.R. Loveland, Woodcock C.E., R.G. Allen, M.C. Anderson, D. Helder, J.R. Irons, D.M. Johnson, R. Kennedy, T.A. Scambos, C.B. Schaaf, J.R. Schott, Y. Sheng, E.F. Vermote, A.S. Belward, R. Bindschadler, W.B. Cohen, F. Gao, J.D. Hipple, P. Hostert, J. Huntington, C.O. Justice, A. Kilic, V. Kovalskyy, Z.P. Lee, L. Lymburner, J.G. Masek,...

    work page 2014

  46. [46]

    Global biodiversity scenarios for the year 2100

    Osvaldo E Sala, FIII Stuart Chapin, Juan J Armesto, Eric Berlow, Janine Bloomfield, Rodolfo Dirzo, Elisabeth Huber-Sanwald, Laura F Huenneke, Robert B Jackson, Ann Kinzig, et al. Global biodiversity scenarios for the year 2100. science, 287(5459):1770–1774, 2000

    work page 2000

  47. [47]

    Schwab and R

    C. Schwab and R. A. Todor. Karhunen–Lo` eve approximation of random fields by generalized fast multipole methods. Journal of Computational Physics, 217(1):100 – 122, 2006. Uncertainty Quantification in Simulation Science

    work page 2006

  48. [48]

    X. Shao. Self-normalization for time series: A review of recent developments. Journal of the American Statistical Association, 110:1797, 2015

    work page 2015

  49. [49]

    Shao and X

    X. Shao and X. Zhang. Testing for change points in time series. Journal of the American Statistical Association, 105:1228, 2010

    work page 2010

  50. [50]

    Introduction: Human dynamics in perspective

    Shih-Lung Shaw and Daniel Sui. Introduction: Human dynamics in perspective. In Human dynamics research in smart and connected communities , pages 1–11. Springer, 2018

    work page 2018

  51. [51]

    Data on global land-cover change: acquisition, assessment and analysis

    David L Skole. Data on global land-cover change: acquisition, assessment and analysis. Changes in land use and land cover: a global perspective , pages 437–471, 1994

    work page 1994

  52. [52]

    Tausch and J

    J. Tausch and J. White. Multiscale bases for the sparse representation of boundary integral operators on complex geometry. SIAM Journal on Scientific Computing , 24(5):1610–1629, 2003

    work page 2003

  53. [53]

    Wied and P

    D. Wied and P. Galeano. Monitoring correlation change in a sequence of random variables. Journal of Statistical Planning and Inference , 143:186, 2013

    work page 2013

  54. [54]

    Woodcock, Richard Allen, Martha Anderson, Alan Belward, Robert Bindschadler, Warren Cohen, Feng Gao, Samuel N

    Curtis E. Woodcock, Richard Allen, Martha Anderson, Alan Belward, Robert Bindschadler, Warren Cohen, Feng Gao, Samuel N. Goward, Dennis Helder, Eileen Helmer, Rama Ne- mani, Lazaros Oreopoulos, Joh Schott, Prasad S. Thenkabail, Eric F. Vermote, James Vo- gelmann, Michael A. Wulder, and Randolph Wynne. Free access to landsat imagery. Science, 320(5879):101...

    work page 2008

  55. [55]

    Climate science special report: fourth national climate assessment, volume i

    Donald J Wuebbles, David W Fahey, and Kathy A Hibbard. Climate science special report: fourth national climate assessment, volume i. 2017

    work page 2017

  56. [56]

    Predicting 3d human dynamics from video

    Jason Y Zhang, Panna Felsen, Angjoo Kanazawa, and Jitendra Malik. Predicting 3d human dynamics from video. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 7114–7123, 2019

    work page 2019

  57. [57]

    Zhang and L

    T. Zhang and L. Lavitas. Unsupervised self-normalized change-point testing for time series. Journal of the American Statistical Association , 113:637, 2018

    work page 2018

  58. [58]

    Woodcock

    Zhe Zhu and Curtis E. Woodcock. Object-based cloud and cloud shadow detection in landsat imagery. Remote Sensing of Environment , 118:83–94, 2012

    work page 2012