Pith. sign in

REVIEW 4 major objections 4 minor 35 references

Bayesian Spatial Field Reconstruction with Unknown Distortions in Sensor Networks

T0 review · 4 major / 4 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 Unknown affine sensor distortions can be inferred jointly with the spatial field, giving a closed-form best linear unbiased estimator and empirical-Bayes reconstructions nearly as accurate as knowing the distortions exactly.

desk verdict Solid extension of the authors' earlier mixture-GP work, with one real covariance typo and an evaluation that stays inside the model's own prior; worth a serious referee, not acceptance as-is. read the letter →

arxiv 1908.05835 v4 pith:ZFUXK35S submitted 2019-08-16 eess.SP

Qikun Xiang , Ido Nevat , Gareth W. Peters This is my paper · ORCID
classification eess.SP
keywords sensornetworksGaussianprocessspatialfieldreconstructionS-BLUEempiricalBayescross-entropymethoditeratedconditionalmodescalibration
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

This paper aims to establish that a fusion center can reconstruct a spatial random field even when it does not know which sensors are reliable, provided every sensor's distortion is an unknown affine gain-and-offset map whose parameters follow a known mixture prior that includes an atom at 'no distortion.' The paper derives the posterior of the distortion parameters and the posterior predictive distribution of the field, then builds two families of estimators: S-BLUE, a closed-form linear estimator using only prior moments, and empirical Bayes estimators that first estimate each sensor's distortion parameters by MAP optimization (cross-entropy or iterated conditional modes) and then plug them into the predictive. It also provides distributed versions that fuse local reconstructions by selecting the cluster with the smallest Bayes risk or predictive variance. If correct, this would let environmental monitoring networks tolerate miscalibrated, compromised, or misused sensors without excluding them or requiring ground-truth calibration, and the experiments support this by showing near-oracle accuracy when distortions are generated from the assumed mixture.

What carries the argument

The engine of the paper is the affine distortion mixture: each sensor's observation is $y_{n,m}=a_n(f(x_n)+\epsilon_{n,m})+b_n$ with $\epsilon_{n,m}\sim N(0,\varsigma^2)$, and $(a_n,b_n)$ follows a finite mixture prior $\pi(\psi_n)=q_n^{(0)}\delta_{(1,0)}+\sum_{k=1}^K q_n^{(k)}\pi_k(\psi_n)$ that includes an atom at no distortion. Theorems 1 and 2 show that, conditional on $\psi$, the sufficient statistics $(g_n,s_n)=(\sum_m y_{n,m},\sum_m y_{n,m}^2)$ leave the field and distortion parameters in a joint Gaussian form, so the posterior of $\psi$ and the predictive of $f_*$ are explicit up to the normalizing constant. Theorem 3 converts those moment expressions into S-BLUE, whose weights require only the prior moments $\mathbb{E}[a]$, $\mathbb{E}[b]$, $\mathbb{E}[aa^\top]$, $\mathbb{E}[bb^\top]$, $\mathbb{E}[ab^\top]$; this is what makes the online estimate a single $O(N)$ linear combination after an offline $O(N^3)$ inversion. The two empirical-Bayes optimizers, CEM and ICM, are the tools that replace the intractable posterior average by a point estimate $\hat\psi$.

What would settle it

Simulate a sensor network in which a known fraction of sensors have nonlinear or time-varying distortions (for example, gain that drifts linearly over time), generated independently of the assumed mixture prior, then run S-BLUE and the empirical-Bayes estimators on these data; if the reconstruction error degrades substantially relative to the oracle while the same methods perform well under affine distortions, the model's robustness claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that, in the mixture-of-Gaussian-process-expert model with affine gain-offset distortions, everything needed for spatial reconstruction is computable in closed form except the normalizing constant of the posterior over distortions. Theorem 1 gives the log-posterior of the distortion parameters $\psi=(\psi_n)_{n=1:N}$; Theorem 2 gives the posterior predictive $p(f_*|y)$ as a Gaussian-mixture integral whose integrand $p(f_*|y,\psi)$ is Gaussian. Theorem 3 then shows the Spatial Best Linear Unbiased Estimator (S-BLUE), restricted to estimators linear in the per-sensor sample means $\bar{g}=M^{-1}g$, has the closed form $\hat h(y)=\mu_* + \mathrm{Cov}[\bar{g},f_*]^\top \mathrm{Cov}[\bar{g}]^{-1}(\bar{g}-\mathbb{E}[\bar{g}])$, with all moments evaluated from the mixture prior. For nonlinear estimation, the paper computes a MAP estimate of $\psi$ by cross-entropy or iterated conditional modes and plugs it into the Gaussian predictive, calling this empirical Bayes. The paper's experiments, including a real temperature dataset with synthetic distortions, show these estimators track the oracle that knows the distortions and beat the naive estimator that ignores them.

Load-bearing premise

The load-bearing premise is that every sensor's distortion is exactly a simple 'scale then add' map, $u\mapsto a_n u+b_n$, with $(a_n,b_n)$ drawn from a known finite mixture that includes an atom at no distortion; nonlinear, drifting, or otherwise unanticipated faults are not covered by the model.

Editorial extensions

If this is right

  • A fusion center can reconstruct the field at arbitrary locations without knowing which sensors are faulty, provided the affine-mixture prior is correct; per-sensor distortion parameters are estimated implicitly or explicitly.
  • S-BLUE needs no online optimization: after an offline $O(N^3)$ preparation, the online update is $O(N)$, making real-time field reconstruction feasible for large networks.
  • The empirical-Bayes estimators recover most of the oracle performance when observations per sensor and SNR are sufficiently high, but with scarce data they can misclassify all sensors as distorting, and at very high SNR both optimizers degrade as the paper's Remark 4 notes.
  • Distributed versions improve scalability and, in high-SNR settings, can beat the centralized empirical-Bayes estimators because lower-dimensional optimization is easier.
  • The method detects distorting sensors with low false-positive and false-negative rates when SNR and sample size are adequate, and reconstruction quality degrades only mildly as the proportion of distorting sensors grows, unlike the naive baseline.

Reading between the lines

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

  • A natural by-product of the mixture prior is a per-sensor fault score: the posterior probability that $Z_n=0$ (no distortion) is exactly the quantity behind the paper's false-positive and false-negative rates, and could be surfaced to a network operator as a reliability flag without extra computation.
  • Because S-BLUE uses only prior moments, it cannot adapt to the actual distortions; a cheap hybrid policy would run S-BLUE online as a safe default and switch to an empirical-Bayes estimate once enough observations per sensor accumulate.
  • The paper's Remark 4, that very high SNR makes both optimizers degrade, suggests the posterior mode becomes ill-posed when the likelihood is extremely concentrated; a testable fix is to replace the point estimate $\hat\psi$ with a Laplace approximation or a few posterior samples.
  • The same affine-distortion plus Gaussian-field template applies to other calibration-inverse problems, such as reconstructing pollutant or radio-frequency fields from crowdsourced sensors, where the gain-offset model is already common.
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

4 major / 4 minor

Summary. The paper develops a Bayesian framework for reconstructing a Gaussian-process spatial field from sensor readings that are subject to unknown affine distortions (gain and offset), modeled by a finite mixture prior with an atom at no distortion. It derives the posterior over distortion parameters (Theorem 1), the posterior predictive distribution (Theorem 2), a closed-form Spatial Best Linear Unbiased Estimator (Theorem 3), and empirical-Bayes estimators whose MAP distortion estimates are obtained by Cross-Entropy (CEM) and Iterated Conditional Modes (ICM), together with distributed variants. Experiments on synthetic fields and on US EPA temperature data with simulated distortions compare S-BLUE, CEM, and ICM against oracle and naive baselines. The paper claims that the proposed methods reconstruct the field with materially lower MSE than the naive baseline and close to the oracle in the tested settings.

Significance. If corrected, the paper is a useful practical contribution: it provides a closed-form linear estimator with a clean offline/online complexity split, tractable empirical-Bayes alternatives, and distributed versions, and the derivations of Theorems 1 and 2 appear sound. The computational complexity analysis is careful and the experimental comparison is reasonably extensive. However, the central theorem contains an algebraic error, the CEM objective and the ICM pseudocode have inconsistencies, and the experimental validation only exercises the exact generative distortion model. These issues are fixable, and the main claims are plausible conditional on those fixes; the paper's contribution is an estimation scheme for a specific and well-motivated distortion model rather than a fundamentally new statistical methodology.

major comments (4)
  1. [§IV, Theorem 3, Eq. (19); Appendix A-C, Eq. (43)] Equation (19) writes the cross-covariance contribution to Cov[g_bar] as diag(mu)(E[ab^T] + E[ab^T]^T). Expanding Cov(A mu + b) gives diag(mu) E[ab^T] + E[ab^T]^T diag(mu), not diag(mu)(E[ab^T] + E[ab^T]^T). The two expressions coincide only when diag(mu) commutes with E[ab^T], which holds for a constant prior mean; all experiments in the paper use a constant prior mean, so the numerical results are not directly affected, but the theorem as stated is incorrect and must be corrected before publication.
  2. [§V-A, CEM objective after Eq. (22)] The CEM section defines the objective as J(psi) = log p(psi|y) + log pi(psi). Since log p(psi|y) in Theorem 1 already contains the term +log pi(psi), this definition double-counts the prior and is not the MAP objective of Eq. (22). As written, Algorithm 2 would maximize log p(y|psi) + 2 log pi(psi) up to constants. The authors should either define J(psi) = log p(y|psi) + log pi(psi) or explicitly state that the expression denoted p(psi|y) in the CEM section is the unnormalized likelihood contribution only.
  3. [§V-B, Algorithm 4 (ICM)] The quantities nu_n and zeta_n in Eqs. (25)-(26) depend on psi_{(-n)} through g_tilde_{(-n)}, but the pseudocode computes them once in Lines 2-3 before the repeat loop and never recomputes them after each coordinate update. Without recomputation, the algorithm is not the iterated conditional-mode algorithm described in the text, and the claim that it converges to a conditional mode is unsupported. The authors should move the computation of nu_n and zeta_n inside the repeat loop and update the complexity analysis accordingly, or explain why no update is needed.
  4. [§VII and §VIII, experimental validation] All simulated distortions, in both the synthetic experiments and the real-data experiment, are generated from the exact affine mixture prior specified in A4/P4, and the real-data "ground truth" is itself a posterior sample from the fitted GP. The paper therefore demonstrates self-consistency under the model but not robustness to distortion families outside the assumed class, such as saturation, deadband, clipping, time-varying gain, or component densities different from the prior. I recommend adding misspecification experiments of this kind, or explicitly limiting the robustness claims to the affine mixture family.
minor comments (4)
  1. [Algorithm 3, E-step] The displayed formula for Q(theta; theta_t) is missing the logarithm: it should read Q(theta; theta_t) = sum_s sum_k p[s]_k log p(psi[s] | z[s]=k; theta), as the surrounding text correctly states.
  2. [Abstract and Conclusion] There are typographical errors: "W e" in the abstract and "preformed" in the conclusion should be corrected.
  3. [Algorithm 2] The termination condition for the CEM algorithm is not specified; a concrete stopping rule, such as a threshold on the relative change of J or on the change of the sampling distribution parameters, would make the algorithm reproducible.
  4. [§VIII-C] The replication protocol for the real-data experiment is not fully clear: it should state explicitly whether the distortion parameters are re-drawn for each of the 20 days or fixed across days, and how many noise realizations are averaged for each reported MSE.

Circularity Check

0 steps flagged · score 0.0 of 10

Derivation chain is self-contained; the in-model evaluation is a validation limitation, not circularity.

full rationale

Walking the claimed derivation chain: Theorem 1 marginalizes the GP latent values in a linear-Gaussian model, Theorem 2 applies standard conditional Gaussian identities and integrates over the parameter posterior, and Theorem 3 minimizes the explicit Bayes risk over the linear family, with all required moments evaluated directly from the stated mixture prior in Eqs. (17)-(19) and (40)-(45). The empirical Bayes estimators in Section V explicitly define psi-hat as the maximizer of log p(psi|y) in Eq. (22) and then plug it into the conditional predictive distribution; no fitted parameter is renamed as a new prediction. The distributed fusion rules are convex combinations chosen by the stated risk criteria, not by assuming the conclusion. Self-citations such as [1], [24], [25], and [28] are used as background for Gaussian-process sensor-network modeling; the distortion mixture model itself is fully specified in the paper (A4, P3-P4) rather than imported as an unverified black box. The semi-synthetic real-data experiment does generate ground truth from a posterior sample of the fitted GP and draws simulated distortions from the assumed prior, so the evaluation is in-model; this is an external-validity limitation, but it is not a circular step in the derivation chain.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim depends on known GP hyperparameters, a fixed affine distortion form, and a fully specified mixture prior whose parameters are chosen by the authors and also used to generate the simulated data. None of these is derived or tested for misspecification. The invented entities field is empty because the model introduces no new physical objects; the spike-and-slab distortion prior is a modeling device rather than an entity.

free parameters (7)
  • GP hyperparameters (signal mean, signal std, noise std, length-scale) = signal mean 30.5034 C, signal std 4.6587 C, noise std 1.6340 C, length-scale 174.3699 km (Table III)
    Estimated by maximum marginal likelihood on 20 days of EPA temperature data, then treated as fixed known inputs to all estimators; synthetic experiments set them directly.
  • Distortion prior probabilities q_k^(n) = q1=0.5 in synthetic exp. 1; q1=q2=q3=1/6 in exp. 2 and real data
    Chosen by the authors and used both to generate simulated distortions and as algorithm inputs, making the validation in-model.
  • Distortion prior component parameters (nu_k, Xi_k) = Examples: a~logN(0.25,0.12), b~N(6,3^2); a~logN(-0.4,0.05^2), b~N(0,0.2^2), etc.
    Chosen by hand; the algorithms assume these are known, and the generative model in the experiments uses the same values.
  • Number of distortion categories K = K=1 in synthetic exp. 1; K=3 in exp. 2 and real data
    Model order selected by the authors; the generative process in simulations uses the same K.
  • CEM optimization hyperparameters (S, rho, termination) = Not fully specified; S and rho left generic, termination 'until condition is triggered'
    Affect whether the global MAP is found; no convergence analysis is provided.
  • ICM initialization and restart count = Random initialization; 'multiple start initialization strategy' without a number
    ICM only finds a local mode; the paper does not specify how many restarts are needed.
  • Number of clusters in distributed experiment = 8
    Chosen via hierarchical clustering for the real-data experiment; affects distributed fusion behavior.
assumptions (6)
  • domain assumption The spatial field f is a Gaussian process with known mean mu and covariance C (P1).
    All posterior and predictive formulas condition on mu and C; if they are misspecified, the reconstructions are miscalibrated.
  • domain assumption Observation noise epsilon is i.i.d. Gaussian with known variance sigma^2 (P2).
    The closed-form likelihood (Theorem 1) and the predictive distribution (Theorem 2) depend on Gaussianity and on sigma^2 being known.
  • domain assumption Sensor distortion is an affine gain-offset transformation T(u)=a u + b (A4).
    The entire model, including the distortion priors and the S-BLUE moments, is built on this affine form; nonlinear or state-dependent distortions are outside the model.
  • domain assumption The distortion parameters psi_n are independent across sensors and follow a known finite mixture prior with an atom at no distortion (P3-P5).
    S-BLUE uses only prior moments of (a_n,b_n), and empirical Bayes uses the mixture components; if the true marginal distribution differs, estimates are biased.
  • ad hoc to paper The MAP optimization (CEM or ICM) finds a usable mode of the posterior.
    Remark 4 reports that both CEM and ICM deteriorate sharply above SNR 20dB because the posterior becomes highly irregular; there is no guarantee of global convergence.
  • domain assumption GP hyperparameters estimated from 20 days of temperature data are valid for the test days used in the experiment.
    Section VIII-B estimates hyperparameters from days 181-200 then applies them to each of those days; seasonal and interannual variation would break this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian Spatial Field Reconstruction with Unknown Distortions in Sensor Networks." pith.science (2026). https://pith.science/paper/ZFUXK35S

@misc{pith2026190805835,
  author       = {Pith},
  title        = {Pith review of: Bayesian Spatial Field Reconstruction with Unknown Distortions in Sensor Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZFUXK35S}},
  note         = {Machine review of arXiv:1908.05835}
}
read the original abstract

Spatial regression of random fields based on potentially biased sensing information is proposed in this paper. One major concern in such applications is that since it is not known a-priori what the accuracy of the collected data from each sensor is, the performance can be negatively affected if the collected information is not fused appropriately. For example, the data collector may measure the phenomenon inappropriately, or alternatively, the sensors could be out of calibration, thus introducing random gain and bias to the measurement process. Such readings would be systematically distorted, leading to incorrect estimation of the spatial field. To combat this detrimental effect, we develop a robust version of the spatial field model based on a mixture of Gaussian process experts. We then develop two different approaches for Bayesian spatial field reconstruction: the first algorithm is the Spatial Best Linear Unbiased Estimator (S-BLUE), in which one considers the quadratic loss function and restricts the estimator to the linear family of transformations; the second algorithm is based on empirical Bayes, which utilises a two-stage estimation procedure to produce accurate predictive inference in the presence of "misbehaving" sensors. In addition, we develop the distributed version of these two approaches to drastically improve the computational efficiency in large-scale settings. We present extensive simulation results using both synthetic datasets and semi-synthetic datasets with real temperature measurements and simulated distortions to draw useful conclusions regarding the performance of each of the algorithms.

Figures

Figures reproduced from arXiv: 1908.05835 by the authors.

Figure 1
Figure 1. Directed acyclic graph (DAG) of the model using the pl [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Contour plots of simulated spatial fields used in the t [PITH_FULL_IMAGE:figures/full_fig_p025_2.png] view at source ↗
Figure 3
Figure 3. Synthetic experiment 1 – relative MSE (log-scale) wi [PITH_FULL_IMAGE:figures/full_fig_p026_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Synthetic experiment 1 – FPR and FNR of CEM and ICM with [PITH_FULL_IMAGE:figures/full_fig_p026_4.png]
Figure 5
Figure 5. Figure 5: Synthetic experiment 2 – relative MSE against varyin [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]
Figure 6
Figure 6. Figure 6: Synthetic experiment 2 – FPR, FNR of CEM and ICM agains [PITH_FULL_IMAGE:figures/full_fig_p029_6.png]
Figure 7
Figure 7. Figure 7: Synthetic experiment 2 – relative MSE against varyin [PITH_FULL_IMAGE:figures/full_fig_p029_7.png]
Figure 8
Figure 8. Figure 8: Synthetic experiment 2 – relative MSE against varyin [PITH_FULL_IMAGE:figures/full_fig_p030_8.png]
Figure 9
Figure 9. Figure 9: Spatial locations of temperature monitoring sites. [PITH_FULL_IMAGE:figures/full_fig_p031_9.png]
Figure 10
Figure 10. Figure 10: Histogram and normal Q-Q plot of the temperature mea [PITH_FULL_IMAGE:figures/full_fig_p031_10.png]
Figure 11
Figure 11. Figure 11: Experiment with real data – relative MSE against var [PITH_FULL_IMAGE:figures/full_fig_p036_11.png]
Figure 12
Figure 12. Figure 12: Experiment with real data – FPR, FNR of CEM and ICM. [PITH_FULL_IMAGE:figures/full_fig_p036_12.png]
Figure 13
Figure 13. Figure 13: Experiment with real data – Reconstructed spatial fi [PITH_FULL_IMAGE:figures/full_fig_p037_13.png]
Figure 14
Figure 14. Figure 14: Experiment with real data – Reconstructed spatial fi [PITH_FULL_IMAGE:figures/full_fig_p038_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 35 canonical work pages

  1. [1]

    Estima tion of Spatially Correlated Random Fields in Heterogeneous Wireless Sensor Networks,

    I. Nevat, G. W. Peters, F. Septier, and T. Matsui, “Estima tion of Spatially Correlated Random Fields in Heterogeneous Wireless Sensor Networks,” IEEE T ransactions on Signal Processing , vol. 63, no. 10, pp. 2597– 2609, 2015. 2, 3, 4

    work page 2015

  2. [2]

    Optimal multi-type s ensor placements in gaussian spatial fields for environmental monitoring,

    C. Sun, Y . Yu, V . O. Li, and J. C. Lam, “Optimal multi-type s ensor placements in gaussian spatial fields for environmental monitoring,” in 2018 IEEE International Smart Cities Conference (ISC2) . IEEE, 2018, pp. 1–8. 2, 3, 4 39

    work page 2018

  3. [3]

    Short-ter m solar power forecasting based on weighted gaussian process regression,

    H. Sheng, J. Xiao, Y . Cheng, Q. Ni, and S. Wang, “Short-ter m solar power forecasting based on weighted gaussian process regression,” IEEE T ransactions on Industrial Electronics , vol. 65, no. 1, pp. 300–308, 2018. 2, 4

    work page 2018

  4. [4]

    Sohraby , D

    K. Sohraby , D. Minoli, and T. Znati, Wireless sensor networks: technology, protocols, and appl ications. John Wiley & Sons, 2007. 2

    work page 2007

  5. [5]

    De centralized hypothesis testing in wireless sensor networks in the presence of misbehaving nodes,

    E. Soltanmohammadi, M. Orooji, and M. Naraghi-Pour, “De centralized hypothesis testing in wireless sensor networks in the presence of misbehaving nodes,” IEEE T ransactions on Information Forensics and Security , vol. 8, no. 1, pp. 205–215, 2013. 2

    work page 2013

  6. [6]

    Calibration of mul ti-target tracking algorithms using non-cooperative targets,

    B. Ristic, D. E. Clark, and N. Gordon, “Calibration of mul ti-target tracking algorithms using non-cooperative targets,” IEEE Journal of Selected T opics in Signal Processing , vol. 7, no. 3, pp. 390–398, 2013. 2

    work page 2013

  7. [7]

    Draft roadmap for next generation air monit oring,

    T. Watkins, “Draft roadmap for next generation air monit oring,” Environmental Protection Agency , 2013. 2

    work page 2013

  8. [8]

    Model-based ren dezvous calibration of mobile sensor networks for monitoring air quality ,

    A. Arfire, A. Marjovi, and A. Martinoli, “Model-based ren dezvous calibration of mobile sensor networks for monitoring air quality ,” in 2015 IEEE SENSORS . IEEE, 2015, pp. 1–4. 2

    work page 2015

Show all 35 references
  1. [9]

    Random access sen sor networks: Field reconstruction from incomplete data,

    F. Fazel, M. Fazel, and M. Stojanovic, “Random access sen sor networks: Field reconstruction from incomplete data,” in 2012 Information Theory and Applications Workshop . IEEE, 2012, pp. 300–305. 2

    2012

  2. [10]

    Quality of information in mobile crowdsensing: Survey and research challenges,

    F. Restuccia, N. Ghosh, S. Bhattacharjee, S. K. Das, and T. Melodia, “Quality of information in mobile crowdsensing: Survey and research challenges,” ACM T ransactions on Sensor Networks (TOSN) , vol. 13, no. 4, p. 34, 2017. 2

    2017

  3. [11]

    Using multi-parameters for calibr ation of low-cost sensors in urban environment,

    X. Fang and I. Bate, “Using multi-parameters for calibr ation of low-cost sensors in urban environment,” in Proceedings of the 2017 International Conference on Embedd ed Wireless Systems and Networks . Junction Publishing, 2017, pp. 1–11. 2

    2017

  4. [12]

    Possible artifacts of data biases in the recen t global surface warming hiatus,

    T. R. Karl, A. Arguez, B. Huang, J. H. Lawrimore, J. R. McM ahon, M. J. Menne, T. C. Peterson, R. S. V ose, and H.-M. Zhang, “Possible artifacts of data biases in the recen t global surface warming hiatus,” Science, vol. 348, no. 6242, pp. 1469–1472, 2015. 2

    2015

  5. [13]

    Convex o ptimization approaches for blind sensor calibration using sparsity ,

    C ¸ . Bilen, G. Puy , R. Gribonval, and L. Daudet, “Convex o ptimization approaches for blind sensor calibration using sparsity ,” IEEE T ransactions on Signal Processing , vol. 62, no. 18, pp. 4847–4856, 2014. 2

    2014

  6. [14]

    On re ndezvous in mobile sensing networks,

    O. Saukh, D. Hasenfratz, C. Walser, and L. Thiele, “On re ndezvous in mobile sensing networks,” in Real-World Wireless Sensor Networks . Springer, 2014, pp. 29–42. 2, 6

    2014

  7. [15]

    Info rmed nonnegative matrix factorization methods for mobile sensor network calibration,

    C. Dorffer, M. Puigt, G. Delmaire, and G. Roussel, “Info rmed nonnegative matrix factorization methods for mobile sensor network calibration,” IEEE T ransactions on Signal and Information Processing ove r Networks , vol. 4, no. 4, pp. 667–682, 2018. 2, 6

    2018

  8. [16]

    Sensor calibration fo r off-the-grid spectral estimation,

    Y . C. Eldar, W. Liao, and S. Tang, “Sensor calibration fo r off-the-grid spectral estimation,” Applied and Computational Harmonic Analysis , 2018. 2

    2018

  9. [17]

    A non-convex approach to joi nt sensor calibration and spectrum estimation,

    M. Cho, W. Liao, and Y . Chi, “A non-convex approach to joi nt sensor calibration and spectrum estimation,” in 2018 IEEE Statistical Signal Processing Workshop (SSP) . IEEE, 2018, pp. 398–402. 2

    2018

  10. [18]

    A secure opti mum distributed detection scheme in under-attack wireless sensor networks,

    E. Nurellari, D. McLernon, and M. Ghogho, “A secure opti mum distributed detection scheme in under-attack wireless sensor networks,” IEEE T ransactions on Signal and Information Processing ove r Networks , vol. 4, no. 2, pp. 325–337, 2018. 2

    2018

  11. [19]

    Collaborative spectrum sensing in the presence of byzantine attacks in cognitive radio networks,

    A. S. Rawat, P . Anand, H. Chen, and P . K. Varshney , “Collaborative spectrum sensing in the presence of byzantine attacks in cognitive radio networks,” IEEE T ransactions on Signal Processing , vol. 59, no. 2, pp. 774–786, 2010. 3

    2010

  12. [20]

    Distributed ev ent detection under byzantine attack in wireless sensor 40 networks,

    P . Zhang, J. Y . Koh, S. Lin, and I. Nevat, “Distributed ev ent detection under byzantine attack in wireless sensor 40 networks,” in 2014 IEEE Ninth International Conference on Intelligent Se nsors, Sensor Networks and Information Processing (ISSNIP) . IEEE, 2014, pp. 1–6. 3

    2014

  13. [21]

    Crowdsourcing urban survei llance: The development of homeland security markets for environmental sensor networks,

    T. Monahan and J. T. Mokos, “Crowdsourcing urban survei llance: The development of homeland security markets for environmental sensor networks,” Geoforum, vol. 49, pp. 279–288, 2013. 3

    2013

  14. [22]

    Crowd-based learning of spatial fields for the internet of things: From harvesting of data to inference,

    E. Arias-de Reyna, P . Closas, D. Dardari, and P . M. Djuri c, “Crowd-based learning of spatial fields for the internet of things: From harvesting of data to inference,” IEEE Signal Processing Magazine , vol. 35, no. 5, pp. 130–139, 2018. 3

    2018

  15. [23]

    Spatial field reconstruction and sensor selection in heterogeneous sensor networks with stochastic energy ha rvesting,

    P . Zhang, I. Nevat, G. W. Peters, F. Septier, and M. A. Osb orne, “Spatial field reconstruction and sensor selection in heterogeneous sensor networks with stochastic energy ha rvesting,” IEEE T ransactions on Signal Processing , vol. 66, no. 9, pp. 2245–2257, 2018. 3

    2018

  16. [24]

    How to Utilize Sen sor Network Data to Efficiently Perform Model Calibration and Spatial Field Reconstruction,

    G. W. Peters, I. Nevat, and T. Matsui, “How to Utilize Sen sor Network Data to Efficiently Perform Model Calibration and Spatial Field Reconstruction,” in Modern Methodology and Applications in Spatial-T emporal Modeling. Springer, 2015, pp. 25–62. 3

    2015

  17. [25]

    Random Field Reconstruction With Quantization in Wireless Sensor Networks,

    I. Nevat, G. W. Peters, and I. B. Collings, “Random Field Reconstruction With Quantization in Wireless Sensor Networks,” IEEE T ransactions on Signal Processing , vol. 61, pp. 6020–6033, 2013. 3, 4

    2013

  18. [26]

    Sampling and reconst ruction of spatial fields using mobile sensors,

    J. Unnikrishnan and M. V etterli, “Sampling and reconst ruction of spatial fields using mobile sensors,” IEEE T ransactions on Signal Processing , vol. 61, no. 9, pp. 2328–2340, 2013. 3

    2013

  19. [27]

    Estimating spatial averages of envi ronmental parameters based on mobile crowdsensing,

    I. Koukoutsidis, “Estimating spatial averages of envi ronmental parameters based on mobile crowdsensing,” ACM T ransactions on Sensor Networks (TOSN) , vol. 14, no. 1, pp. 1–26, 2017. 3

    2017

  20. [28]

    A trust-base d mixture of gaussian processes model for reliable regression in participatory sensing

    Q. Xiang, J. Zhang, I. Nevat, and P . Zhang, “A trust-base d mixture of gaussian processes model for reliable regression in participatory sensing.” in IJCAI, 2017, pp. 3866–3872. 3, 4

    2017

  21. [29]

    Location-aw are cooperative spectrum sensing via Gaussian Processes,

    I. Nevat, G. W. Peters, and I. B. Collings, “Location-aw are cooperative spectrum sensing via Gaussian Processes,” in Communications Theory Workshop (AusCTW), 2012 Australian . IEEE, 2012, pp. 19–24. 4

    2012

  22. [30]

    C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine Learning (Adaptive Computa tion and Machine Learning). The MIT Press, 2005. 5, 11, 32

    2005

  23. [31]

    Calibree: A self-calibration system for mobile sensor networks,

    E. Miluzzo, N. D. Lane, A. T. Campbell, and R. Olfati-Sab er, “Calibree: A self-calibration system for mobile sensor networks,” in International Conference on Distributed Computing in Sens or Systems . Springer, 2008, pp. 314–331. 6

    2008

  24. [32]

    The cross-entropy method for combinat orial and continuous optimization,

    R. Rubinstein, “The cross-entropy method for combinat orial and continuous optimization,” Methodology and computing in applied probability , vol. 1, no. 2, pp. 127–190, 1999. 14

    1999

  25. [33]

    A gentle tutorial of the em algorithm and its application t o parameter estimation for gaussian mixture and hidden markov models,

    J. A. Bilmes et al., “A gentle tutorial of the em algorithm and its application t o parameter estimation for gaussian mixture and hidden markov models,” 1998. 16

    1998

  26. [34]

    On the statistical analysis of dirty picture s,

    J. Besag, “On the statistical analysis of dirty picture s,” Journal of the Royal Statistical Society: Series B (Methodo log- ical), vol. 48, no. 3, pp. 259–279, 1986. 17

    1986

  27. [35]

    Friedman, T

    J. Friedman, T. Hastie, and R. Tibshirani, The elements of statistical learning . Springer series in statistics New York, 2001, vol. 1, no. 10. 33 41 APPENDIX A PROOF OF THEOREMS A. Proof of Theorem 1 Proof. For n = 1, . . . , N, the log model likelihood given f := (f (xn))n=1...

    2001

Pith tools

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