Pith. sign in

REVIEW 3 major objections 3 minor 12 references

Some examples of application for predicting of compressive sensing method

T0 review · 3 major / 3 minor · reviewed 2026-05-24 · grok-4.3

Pith's one-line read SALSA extrapolation outperforms linear extrapolation across electrical, temperature and stock data while retaining more original shape than causal smoothing.

desk verdict SALSA is applied to forecasting on three series but the work supplies no experimental protocol, parameters, or statistics, leaving the superiority claims unevaluable. read the letter →

arxiv 1907.11508 v1 pith:Z3SWQS2E submitted 2019-07-26 stat.ME

classification stat.ME
keywords SALSAalgorithmcompressivesensingforecastingextrapolationtimeserieselectricalsignalsstockpricestemperaturedata
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

The paper tests the SALSA algorithm as a forecasting method on simulated electrical signals, Australian Bureau of Meteorology temperature records, and Australian stock exchange prices. It compares performance against linear extrapolation and causal smoothing extrapolation using L2 distance and qualitative shape retention. SALSA beats linear extrapolation in every case tested. Against causal smoothing, SALSA usually keeps more of the original function's shape and statistics, though it produces higher L2 error on the complex systems. The work concludes that SALSA is superior for the electrical signals it was designed to handle and gives a wider range of possible outcomes, while causal smoothing offers a more conservative forecast for complex systems.

What carries the argument

SALSA algorithm used for extrapolation on time series, evaluated by L2 Euclidean distance and qualitative shape retention against linear and causal smoothing baselines.

What would settle it

A new time series outside the three tested datasets where SALSA extrapolation produces higher L2 error than linear extrapolation or loses more shape features than causal smoothing.

Watch

Extended reading notes

Core claim

Application of the SALSA algorithm to forecasting shows it is a better method than linear extrapolation in all tested cases and usually retains more shape and statistical elements of the original function than causal smoothing extrapolation, leading to the conclusion that causal smoothing can provide a more conservative forecast for complex systems while SALSA more accurately predicts the range of possible events and is the superior method for electrical signals.

Load-bearing premise

The three chosen datasets and two error measures are sufficient to support general statements about performance on electrical signals versus complex systems.

Editorial extensions

If this is right

  • SALSA extrapolation is the superior forecasting method for electrical signals.
  • SALSA retains more shape and statistical elements of the original function than causal smoothing in most cases.
  • Causal smoothing extrapolation can provide a more conservative forecast for complex systems.
  • SALSA more accurately predicts the range of possible events in the tested complex systems.

Reading between the lines

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

  • SALSA could be applied to other physical time series such as sensor streams or biological signals to test broader utility.
  • A hybrid approach that switches between SALSA and causal smoothing based on signal complexity might balance range prediction with conservatism.
  • The method's performance on longer forecast horizons or with added noise could be measured to check robustness beyond the reported experiments.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The manuscript applies the SALSA compressive sensing algorithm as a forecasting method to three time series (a simulated electrical signal, Australian Bureau of Meteorology temperature records, and Australian Stock Exchange stock prices) and compares the results to linear extrapolation and causal smoothing extrapolation. It claims that SALSA outperforms linear extrapolation in all cases while usually retaining more shape and statistical elements than causal smoothing, although it exhibits higher L2 Euclidean distance than causal smoothing on the complex-system examples; the conclusion is that causal smoothing yields more conservative forecasts for complex systems whereas SALSA is superior for electrical signals and range prediction.

Significance. If the experimental comparisons were fully documented and reproducible, the work could provide concrete illustrations of compressive-sensing techniques applied to forecasting outside their original reconstruction setting. The distinction drawn between electrical-signal and complex-system behavior, together with the emphasis on shape retention versus L2 error, would be of interest to the statistical signal-processing community. At present the absence of protocols, parameters, and quantitative shape metrics prevents any assessment of whether these distinctions are robust.

major comments (3)
  1. [Abstract] Abstract and experimental description: comparative performance claims (SALSA superior to linear extrapolation in all cases; higher L2 than causal smoothing on complex systems) are asserted without any statement of the experimental protocol, forecast horizon, SALSA parameter values, implementation of the two baseline methods, dataset lengths, or sampling details. This renders the central empirical claims unverifiable.
  2. [Discussion] Results and discussion: the paper invokes an unquantified notion of 'retaining more shape and statistical elements' to qualify the L2 comparison, yet supplies neither a formal definition nor a numerical proxy for this notion, nor any error bars or hypothesis tests on the reported distances. The generalization from the three chosen series to statements about 'electrical signals' versus 'complex systems' therefore rests on an unreported and untested qualitative judgment.
  3. [Methods] Methods: no description is given of how the simulated electrical signal was generated, how the extrapolation windows were chosen, or how causal smoothing was implemented, all of which are load-bearing for reproducing or evaluating the reported superiority claims.
minor comments (3)
  1. [Abstract] Abstract: 'casual smoothing' is presumably a typographical error for 'causal smoothing'.
  2. [Abstract] Abstract: 'imperially stated' should read 'empirically stated'.
  3. [Title] Title: the phrasing 'predicting of compressive sensing method' is grammatically awkward and should be revised for clarity.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments. We agree that the manuscript requires additional detail on experimental protocols and methods to support the empirical claims and will revise accordingly.

read point-by-point responses
  1. Referee: [Abstract] Abstract and experimental description: comparative performance claims (SALSA superior to linear extrapolation in all cases; higher L2 than causal smoothing on complex systems) are asserted without any statement of the experimental protocol, forecast horizon, SALSA parameter values, implementation of the two baseline methods, dataset lengths, or sampling details. This renders the central empirical claims unverifiable.

    Authors: We acknowledge that these details are absent from the submitted manuscript. In the revision we will insert a dedicated experimental protocol subsection specifying the forecast horizons, SALSA regularization and other parameter values, the precise implementations of linear extrapolation and causal smoothing, the lengths of each time series, and the sampling rates or selection procedures. This will render the performance comparisons verifiable. revision: yes

  2. Referee: [Discussion] Results and discussion: the paper invokes an unquantified notion of 'retaining more shape and statistical elements' to qualify the L2 comparison, yet supplies neither a formal definition nor a numerical proxy for this notion, nor any error bars or hypothesis tests on the reported distances. The generalization from the three chosen series to statements about 'electrical signals' versus 'complex systems' therefore rests on an unreported and untested qualitative judgment.

    Authors: The shape-retention claim is currently qualitative and based on visual inspection of the figures. We agree this is insufficient. In revision we will either introduce a quantitative proxy (e.g., preservation of autocorrelation structure or spectral content) with accompanying numerical values, or restrict the discussion to the three concrete examples without broad generalization to signal classes. We will also add error bars or repeated-trial statistics on the L2 distances where computationally feasible. revision: yes

  3. Referee: [Methods] Methods: no description is given of how the simulated electrical signal was generated, how the extrapolation windows were chosen, or how causal smoothing was implemented, all of which are load-bearing for reproducing or evaluating the reported superiority claims.

    Authors: We will expand the Methods section to describe the generation of the simulated electrical signal (including any underlying model or parameters), the criteria and specific values used to select extrapolation windows, and the exact implementation of causal smoothing (kernel, bandwidth, or algorithmic details). These additions will permit full reproduction of the experiments. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: purely empirical comparisons

full rationale

The manuscript applies the SALSA algorithm to three specific time series (simulated electrical signal, BOM temperature, ASX stocks) and reports L2 distances plus qualitative shape retention versus linear extrapolation and causal smoothing. No derivation, ansatz, uniqueness theorem, or parameter-fitting step is presented that could reduce to its own inputs; the claims rest on direct numerical comparison of the chosen methods on the chosen data. This is a standard empirical study whose conclusions can be checked against the reported tables without reference to any self-referential construction.

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

The abstract describes an empirical application study and introduces no new mathematical axioms, free parameters, or postulated entities.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Some examples of application for predicting of compressive sensing method." pith.science (2026). https://pith.science/paper/Z3SWQS2E

@misc{pith2026190711508,
  author       = {Pith},
  title        = {Pith review of: Some examples of application for predicting of compressive sensing method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Z3SWQS2E}},
  note         = {Machine review of arXiv:1907.11508}
}
read the original abstract

This paper considers application of the SALSA algorithm as a method of forecasting and applies it to simulated electrical signal, temperature recording from the Australian Bureau of Meteorology and stock prices from the Australian stock exchange. It compares it to basic linear extrapolation and casual smoothing extrapolation, in all cases SALSA extrapolation proves to be a better method of forecasting than linear extrapolation. However, it cannot be imperially stated that it is superior to Causal smoothing extrapolation in complex systems as it has a higher L2 euclidean in these experiments. while usually retaining more shape and statistical elements of the original function than Causal smoothing extrapolation. Leading to the conclusion the Causal Smoothing extrapolation can provide a more conservative forecast for complex systems while the SALSA algorithm more accurately predicts the range of possible events as well as being the superior forecasting method for electrical signals, the physical process it is designed to forecast.

Figures

Figures reproduced from arXiv: 1907.11508 by the authors.

Figure 1
Figure 1. Monte Carlo simulation, forecast of 10 point [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Salsa and Causal extrapolation forecasts for 0.2 seconds of an electrical signal repeated [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Salsa and Causal extrapolation forecasts for 1 seconds of an electrical signal repeated [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Salsa and Causal and linear extrapolation forecasts for 2 days temperature data over [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Salsa and Causal and linear extrapolation forecasts for 7 days temperature data over [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Salsa and Causal and linear extrapolation forecasts for 2 days ASX stock data over the [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Salsa and Causal and linear extrapolation forecasts for 5 days ASX stock data over the [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

12 extracted references · 12 canonical work pages

  1. [1]

    Dokuchaev, ”On causal extrapolation of sequences with applications to forecasting, Applied Mathematics and Computation, vol

    N. Dokuchaev, ”On causal extrapolation of sequences with applications to forecasting, Applied Mathematics and Computation, vol. 328, pp. 276-286, 2018

  2. [2]

    Rowe, ”Applications of band-limited extrapolation to forecasting of weather and financial time series,” arXiv, N/A, 2019

    N. Rowe, ”Applications of band-limited extrapolation to forecasting of weather and financial time series,” arXiv, N/A, 2019

  3. [3]

    M. V. Afonso, J. M. Bioucas-Dias and M. A. T. Figueriredo, ”Fast Image Recovery Using Variable Splitting and Constrained Optimization,” IEEE Transcation on Image Processing, vol. 19, no. 9, pp. 2345-2356, 2010

  4. [4]

    [Accessed 11 10 2018]

    Australian Government Bureau of Meterorology , ”Australian Govern- ment Bureau of Meterorology,” http://www.bom.gov.au/jsp/ncc/cdio/weatherData /av?p nccObsCode=122&p display type=dailyDataFile&p startYear=&p c=&p stn num=009225. [Accessed 11 10 2018]

  5. [5]

    http://www.bom.gov.au/clim data/cdio/metadata/pdf/siteinfo/IDCJMD0040.009225.SiteInfo.pdf

    Australian Government Bureau of Meterology, ”Basic Climatological Station Metadata,” 26 7 2018. http://www.bom.gov.au/clim data/cdio/metadata/pdf/siteinfo/IDCJMD0040.009225.SiteInfo.pdf. [Accessed 31 10 2018]

  6. [6]

    https://www.marketindex.com.au/asx200

    Market Index., ”S&P/ASX 200,” 10 10 2018. https://www.marketindex.com.au/asx200. [Accessed 10 10 2018]

  7. [7]

    https://www2.commsec.com.au/Private/MarketPrices/TradeHistory /TradeHis- 16 tory.aspx?stockCode=CBA

    Commonwealth Securities Limited, ”Trade History CBA,” 26 10 2018. https://www2.commsec.com.au/Private/MarketPrices/TradeHistory /TradeHis- 16 tory.aspx?stockCode=CBA. [Accessed 2018 10 26]

  8. [8]

    [Accessed 2 11 2018]

    Australian Government Department of Enviorment and energy, ”Portfolio Budget Statements 2017-18,” http://www.environment.gov.au/about-us/publications/budget/portfolio- budget- statements-2017-18. [Accessed 2 11 2018]

Show all 12 references
  1. [9]

    Bingham, ”Szegs theorem and its probabilistic descendants,” Probab

    N. Bingham, ”Szegs theorem and its probabilistic descendants,” Probab. Surveys 9 , vol. 9, pp. 287-324, 2012

  2. [10]

    Dokuchaev, ”Filtration and extrapolation in trajectory analysis based on spectrum methods,” Working paper, 2019

    N. Dokuchaev, ”Filtration and extrapolation in trajectory analysis based on spectrum methods,” Working paper, 2019

  3. [11]

    Selesnick, ”Introduction to Sparsity in Signal Processing,” Ivan Selesnick, 2012

    I. Selesnick, ”Introduction to Sparsity in Signal Processing,” Ivan Selesnick, 2012

  4. [12]

    Selesnick, ”L1-Norm Penalized Least Squares with Salsa,” Connexions, 2014

    I. Selesnick, ”L1-Norm Penalized Least Squares with Salsa,” Connexions, 2014. 17 Appendix A.1 Monte Carlo Extrapolation %Start by creating Monte-Carlo random numbers 1-D %Z(t)=A(t)Z(t-1)+o*u(t) %Input carlosim(N,o,z0,v) for N for 2N+1 points, z0 initial points, v %dimensions, ...

Pith tools

Reviewed May 24, 2026 · model on record in the stance chip above.