Detection and Statistical Modeling of Birth-Death Anomaly

Farhang Yeganegi; Nasser M Nasrabadi; Salman Mohamadi

arxiv: 1906.11788 · v1 · pith:A7IHOUCXnew · submitted 2019-06-27 · 📡 eess.SP

Detection and Statistical Modeling of Birth-Death Anomaly

Salman Mohamadi , Farhang Yeganegi , Nasser M Nasrabadi This is my paper

Pith reviewed 2026-05-25 14:35 UTC · model grok-4.3

classification 📡 eess.SP

keywords anomaly detectionARIMA modelARCH/GARCHheteroskedasticitystatistical signal processingtime seriesvariance modeling

0 comments

The pith

An ARIMA signal plus ARCH/GARCH anomaly variance model extracts heteroskedastic anomalies from nonstationary time series.

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

The paper sets out a statistical framework for detecting anomalies by modeling the observed signal as the sum of a main component and an anomalous noise term. The main component is taken to follow an ARIMA process to accommodate nonstationarity common in natural signals. The anomaly term is modeled as additive noise whose variance obeys an ARMA process and shows heteroskedasticity, which is captured by fitting an ARCH or GARCH model to the residuals after the ARIMA step. This decomposition lets the anomaly pattern be isolated and characterized without assuming the overall series is stationary. The approach is intended for applied signal-processing tasks where variance changes are a signature of the anomaly itself.

Core claim

Given an additive model in which the signal of interest follows an ARIMA process and the anomaly appears as noise whose variance follows an ARMA process while possessing heteroskedastic properties, ARCH or GARCH modeling extracts the anomaly pattern from the residuals.

What carries the argument

The additive decomposition of an ARIMA-modeled signal and an ARCH/GARCH-modeled heteroskedastic anomaly whose variance follows an ARMA process.

If this is right

Anomaly patterns become extractable from the residuals once an ARIMA model is fitted to the main signal.
The framework accommodates nonstationarity in the signal of interest through the ARIMA assumption.
ARCH/GARCH captures the changing variance that defines the heteroskedastic anomaly.
The resulting model supplies a general statistical procedure for anomaly detection in signal-processing applications.

Where Pith is reading between the lines

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

The same decomposition could be tested on real datasets containing documented birth-death anomalies to measure extraction accuracy.
Alternative variance models besides ARCH/GARCH might be substituted to check whether the heteroskedasticity assumption is essential.
The additive structure suggests the method could extend to other time-series domains where variance shifts mark rare events.

Load-bearing premise

The observed series can be expressed as an ARIMA signal plus additive noise whose variance follows an ARMA process with heteroskedastic behavior.

What would settle it

In a controlled series where the anomaly is known and injected, fitting an ARIMA model and then applying ARCH/GARCH to the residuals fails to isolate or characterize the injected anomaly.

read the original abstract

Generally, anomaly detection has a great importance particularly in applied statistical signal processing. Here we provide a general framework in order to detect anomaly through the statistical modeling. In this paper, it is assumed that a signal is corrupted by noise whose variance follows an ARMA model. The assumption on the signal is further compromised to encompass the inherent nonstationarity associated with natural phenomenon, hence, the signal of interest is assumed to follow an ARIMA model and the noise to denote an anomaly, however, unknown. Anomaly is assumed to possess heteroskedastic properties, therefore, ARCH/GARCH modeling could extract the anomaly pattern given an additive model for signal of interest and anomaly.

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.

Desk Editor's Note private letter to a colleague

This paper states some modeling assumptions for anomaly detection but supplies no equations, algorithm, or evidence to make the idea work.

read the letter

The one thing to know is that this paper describes a modeling setup for anomaly detection but gives no method to actually do the detection or any evidence that it works. It assumes the observed signal is the sum of an ARIMA process for the signal of interest and an anomaly term whose variance follows an ARMA model and shows heteroskedasticity. The suggestion is that ARCH or GARCH models can pull out the anomaly. That's the whole contribution. This is not new. Time series anomaly detection using ARIMA and GARCH variants is common in the literature. The paper applies existing tools without adding new results or addressing limitations of those tools. The paper does identify that natural signals can be nonstationary and that anomalies might have time-varying variance, which is a fair point for some applications in signal processing. The soft spots are substantial. There are no equations showing how the models are combined or estimated jointly. No algorithm for separating the components. No discussion of when the anomaly can be identified from the signal. No validation on synthetic or real data. The claim that ARCH/GARCH extracts the pattern is asserted without justification. Because of this, the approach reduces to fitting standard models and hoping the residuals or parameters reveal the anomaly, but without benchmarks it's circular. This paper is for readers who want a high-level sketch of using these models in anomaly detection. It won't give value to someone looking for a working method or theoretical advance. It does not show clear engagement with how to make the idea operational. I would not send this to peer review. It needs the missing technical content before it can be evaluated seriously.

Referee Report

3 major / 1 minor

Summary. The paper proposes a general framework for anomaly detection by modeling an observed signal as the sum of a signal of interest following an ARIMA process (to capture nonstationarity) and an unknown anomaly whose variance follows an ARMA process and exhibits heteroskedasticity, suggesting that ARCH/GARCH models can extract the anomaly pattern.

Significance. The proposed statistical modeling approach, if accompanied by derivations, identifiability results, and validation, could contribute to anomaly detection in nonstationary signals within statistical signal processing. The manuscript, however, contains no such supporting material.

major comments (3)

[Abstract] Abstract: the claim that 'ARCH/GARCH modeling could extract the anomaly pattern' is asserted without any equations, likelihood, estimation procedure, or separation method showing how the heteroskedastic anomaly component is isolated from the composite ARIMA signal plus anomaly.
[Abstract] Abstract: no identifiability conditions, parameter estimation algorithm, or conditions under which the ARMA variance model for the anomaly can be distinguished from the ARIMA signal are provided, leaving the additive model y = s + a unoperationalized.
[Abstract] Abstract: the manuscript states modeling assumptions but supplies neither data, simulations, error analysis, nor any empirical demonstration that the heteroskedastic properties allow anomaly extraction via ARCH/GARCH.

minor comments (1)

The title references 'Birth-Death Anomaly' but the abstract provides no definition or connection to this specific anomaly type.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments. Our manuscript outlines a general statistical modeling framework for anomaly detection in nonstationary signals using ARIMA for the signal of interest and ARCH/GARCH for the heteroskedastic anomaly in an additive model. We agree that the current version is high-level and lacks detailed mathematical and empirical support. We address each major comment below, noting that the paper is positioned as a conceptual contribution rather than a complete methodological paper.

read point-by-point responses

Referee: [Abstract] Abstract: the claim that 'ARCH/GARCH modeling could extract the anomaly pattern' is asserted without any equations, likelihood, estimation procedure, or separation method showing how the heteroskedastic anomaly component is isolated from the composite ARIMA signal plus anomaly.

Authors: The framework is based on the standard additive decomposition y = s + a, where s follows ARIMA(p,d,q) to model the nonstationary signal, and a is the anomaly with variance following an ARMA process, which is the definition of ARCH/GARCH models. The extraction would involve fitting an ARIMA to the observed y and then applying GARCH to the residuals to capture the anomaly variance. However, we concede that no explicit likelihood or separation algorithm is derived in the manuscript. revision: no
Referee: [Abstract] Abstract: no identifiability conditions, parameter estimation algorithm, or conditions under which the ARMA variance model for the anomaly can be distinguished from the ARIMA signal are provided, leaving the additive model y = s + a unoperationalized.

Authors: The distinction is conceptual: the ARIMA models the conditional mean, while the ARCH/GARCH models the conditional variance of the anomaly component. Identifiability would require assumptions such as the anomaly being zero-mean and the processes being independent. We agree that formal conditions and an estimation algorithm (e.g., maximum likelihood for the combined model) are not provided. revision: no
Referee: [Abstract] Abstract: the manuscript states modeling assumptions but supplies neither data, simulations, error analysis, nor any empirical demonstration that the heteroskedastic properties allow anomaly extraction via ARCH/GARCH.

Authors: As a brief proposal, the manuscript focuses on the modeling assumptions without including simulations or data analysis. Such empirical validation would be valuable but is not included in this version of the work. revision: no

Circularity Check

0 steps flagged

No significant circularity; framework rests on modeling assumptions without a closed derivation chain.

full rationale

The paper states modeling assumptions (signal follows ARIMA, anomaly variance follows ARMA with heteroskedasticity addressed via ARCH/GARCH in an additive model) but provides no derivation, equations, or first-principles result that reduces to its inputs by construction. No predictions are claimed that equate to fitted parameters, no self-citations are load-bearing, and no uniqueness theorems or ansatzes are smuggled in. The approach is presented as a general framework based on these assumptions rather than a self-referential extraction procedure. This is a standard non-circular modeling proposal.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The framework rests entirely on domain assumptions about signal and noise structure with no independent evidence or derivations provided in the abstract; several free parameters are implied by the model fitting but not enumerated.

free parameters (3)

ARIMA model orders and coefficients
Parameters of the ARIMA model for the signal of interest are fitted as part of the framework.
ARMA variance parameters
Parameters governing the noise variance model.
ARCH/GARCH parameters
Parameters to capture heteroskedasticity of the anomaly.

axioms (3)

domain assumption Signal follows an ARIMA model
Stated directly in the abstract for handling nonstationarity.
domain assumption Noise variance follows an ARMA model
Assumed for the corrupting noise in the abstract.
domain assumption Anomaly possesses heteroskedastic properties
Invoked to justify use of ARCH/GARCH modeling.

pith-pipeline@v0.9.0 · 5638 in / 1274 out tokens · 34884 ms · 2026-05-25T14:35:53.537410+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

St = Xt + at Zt ... Xt ... ARIMA ... Zt ... GARCH ... ARCH/GARCH modeling could extract the anomaly pattern
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

ARIMA-GARCH model ... birth-death anomaly

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.

Reference graph

Works this paper leans on

23 extracted references · 23 canonical work pages

[1]

Anomaly Anomaly or outlier detection [1, 2], is a technique that can d etect the presence of abnormal signals

INTRODUCTION 1.1. Anomaly Anomaly or outlier detection [1, 2], is a technique that can d etect the presence of abnormal signals. Anomaly is a pattern in the data that does not conform to the normal state or behavior. Import ant ap- plications include the detection of radar and sonar system i ntruder, covert telecommunication, cyber intrusion, biomedical a...

work page
[2]

In processes th at show anomaly pattern especially natural processes, for ins tance, in epilepsy, even patients using drug to control the seizure, i t occurs time to time [8]

PROBLEM FORMULATION Many stochastic processes, in particular natural processe s can be modeled as a birth-death processes, such processes are cons id- ered as a continuous-time Markov chains [7]. In processes th at show anomaly pattern especially natural processes, for ins tance, in epilepsy, even patients using drug to control the seizure, i t occurs tim...

work page 2000
[3]

First we model a synthesi zed signal possessing one anomaly pattern which is gradually pr ovoked

DETECTION OF SIMULATED AND NATURAL ANOMALIES For anomalous natural and synthesized signals, this modeli ng is tested and results seem plausible. First we model a synthesi zed signal possessing one anomaly pattern which is gradually pr ovoked. As it is seen in Figure 3, signal is anomalous, and the model we ll detects and traces the anomaly pattern. In fac...

work page 2000
[4]

The time varying mean is assumed t o follow an ARIMA and the anomalous signal as a GARCH model

CONCLUSION In this paper, detection of an anomalous signal is declared i n pres- ence of a time varying mean. The time varying mean is assumed t o follow an ARIMA and the anomalous signal as a GARCH model. The observed signal may possess anomalous pattern for some t ime. This time varying pattern is captured adaptively using an ad aptive CFAR. The propose...

work page
[5]

Outlier detection and t ime series modeling,

Bovas Abraham and Alice Chuang, “Outlier detection and t ime series modeling,” Technometrics, vol. 31, no. 2, pp. 241–248, 1989

work page 1989
[6]

Anomaly detection: A survey,

V arun Chandola, Arindam Banerjee, and Vipin Kumar, “Anomaly detection: A survey,” ACM computing surveys (CSUR), vol. 41, no. 3, pp. 15, 2009

work page 2009
[7]

The forward discount anomaly and the ris k premium: A survey of recent evidence,

Charles Engel, “The forward discount anomaly and the ris k premium: A survey of recent evidence,” Journal of empirical ﬁnance, vol. 3, no. 2, pp. 123–192, 1996

work page 1996
[8]

Combining negative selection and classiﬁcation techniqu es for anomaly detection,

Fabio Gonzalez, Dipankar Dasgupta, and Robert Kozma, “Combining negative selection and classiﬁcation techniqu es for anomaly detection,” in Evolutionary Computation, 2002. CEC’02. Proceedings of the 2002 Congress on . IEEE, 2002, vol. 1, pp. 705–710

work page 2002
[9]

Anomaly-based network intrusion detection: Techniques, systems and challenges,

Pedro Garcia-Teodoro, Jesus Diaz-V erdejo, Gabriel Mac i´ a- Fern´ andez, and Enrique V´ azquez, “Anomaly-based network intrusion detection: Techniques, systems and challenges, ” computers & security, vol. 28, no. 1-2, pp. 18–28, 2009

work page 2009
[10]

A classiﬁca tion framework for anomaly detection,

Ingo Steinwart, Don Hush, and Clint Scovel, “A classiﬁca tion framework for anomaly detection,” Journal of Machine Learn- ing Research, vol. 6, no. Feb, pp. 211–232, 2005

work page 2005
[11]

Transition prob- abilities for general birth–death processes with applicat ions in ecology, genetics, and evolution,

Forrest W Crawford and Marc A Suchard, “Transition prob- abilities for general birth–death processes with applicat ions in ecology, genetics, and evolution,” Journal of mathematical bi- ology, vol. 65, no. 3, pp. 553–580, 2012

work page 2012
[12]

A wavelet-chaos methodology for analysis of eegs and eeg subbands to detect seizure and epilepsy,

Hojjat Adeli, Samanwoy Ghosh-Dastidar, and Nahid Dadme hr, “A wavelet-chaos methodology for analysis of eegs and eeg subbands to detect seizure and epilepsy,” IEEE Transactions on Biomedical Engineering, vol. 54, no. 2, pp. 205–211, 2007

work page 2007
[13]

Conditional anomaly detection,

Xiuyao Song, Mingxi Wu, Christopher Jermaine, and Sanja y Ranka, “Conditional anomaly detection,” IEEE Transactions on Knowledge and Data Engineering , vol. 19, no. 5, pp. 631– 645, 2007

work page 2007
[14]

Anomaly detection using forecasting methods arima and hwds,

Eduardo HM Pena, Marcos VO de Assis, and Mario Lemes Proenc ¸a, “Anomaly detection using forecasting methods arima and hwds,” in Chilean Computer Science Society (SCCC), 2013 32nd International Conference of the . IEEE, 2013, pp. 63–66

work page 2013
[15]

Anomaly detection on social media using arima models,

Tim Isbister, “Anomaly detection on social media using arima models,” 2015

work page 2015
[16]

George EP Box, Gwilym M Jenkins, Gregory C Reinsel, and Greta M Ljung, Time series analysis: forecasting and control , John Wiley & Sons, 2015

work page 2015
[17]

Autoregressive conditional heterosc edasticity with estimates of the variance of united kingdom inﬂation,

Robert F Engle, “Autoregressive conditional heterosc edasticity with estimates of the variance of united kingdom inﬂation,” Econometrica: Journal of the Econometric Society , pp. 987– 1007, 1982

work page 1982
[18]

Generalized autoregressive conditi onal het- eroskedasticity,

Tim Bollerslev, “Generalized autoregressive conditi onal het- eroskedasticity,” Journal of econometrics , vol. 31, no. 3, pp. 307–327, 1986

work page 1986
[19]

An adaptive detection algorithm,

Edward J Kelly, “An adaptive detection algorithm,” IEEE transactions on aerospace and electronic systems , , no. 2, pp. 115–127, 1986

work page 1986
[20]

Constant-false-alarm-rate process ors for lo- cally nonstationary clutter,

Ramon Nitzberg, “Constant-false-alarm-rate process ors for lo- cally nonstationary clutter,” IEEE Transactions on Aerospace and Electronic Systems, , no. 3, pp. 399–405, 1973

work page 1973
[21]

A cfar processor for the detection of unknown random signals in nonstationary correlated noise,

QT Zhang, “A cfar processor for the detection of unknown random signals in nonstationary correlated noise,” Signal pro- cessing, vol. 48, no. 1, pp. 17–26, 1996

work page 1996
[22]

Targ et detection in sea-clutter using stationary wavelet transfo rms,

Vichet Duk, Luke Rosenberg, and Brian Wai-Him Ng, “Targ et detection in sea-clutter using stationary wavelet transfo rms,” IEEE Transactions on Aerospace and Electronic Systems , vol. 53, no. 3, pp. 1136–1146, 2017

work page 2017
[23]

Arima-garch modeling for epileptic seizure prediction,

Salman Mohamadi, Hamidreza Amindavar, and SM Ali Tayaranian Hosseini, “Arima-garch modeling for epileptic seizure prediction,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 994–998

work page 2017

[1] [1]

Anomaly Anomaly or outlier detection [1, 2], is a technique that can d etect the presence of abnormal signals

INTRODUCTION 1.1. Anomaly Anomaly or outlier detection [1, 2], is a technique that can d etect the presence of abnormal signals. Anomaly is a pattern in the data that does not conform to the normal state or behavior. Import ant ap- plications include the detection of radar and sonar system i ntruder, covert telecommunication, cyber intrusion, biomedical a...

work page

[2] [2]

In processes th at show anomaly pattern especially natural processes, for ins tance, in epilepsy, even patients using drug to control the seizure, i t occurs time to time [8]

PROBLEM FORMULATION Many stochastic processes, in particular natural processe s can be modeled as a birth-death processes, such processes are cons id- ered as a continuous-time Markov chains [7]. In processes th at show anomaly pattern especially natural processes, for ins tance, in epilepsy, even patients using drug to control the seizure, i t occurs tim...

work page 2000

[3] [3]

First we model a synthesi zed signal possessing one anomaly pattern which is gradually pr ovoked

DETECTION OF SIMULATED AND NATURAL ANOMALIES For anomalous natural and synthesized signals, this modeli ng is tested and results seem plausible. First we model a synthesi zed signal possessing one anomaly pattern which is gradually pr ovoked. As it is seen in Figure 3, signal is anomalous, and the model we ll detects and traces the anomaly pattern. In fac...

work page 2000

[4] [4]

The time varying mean is assumed t o follow an ARIMA and the anomalous signal as a GARCH model

CONCLUSION In this paper, detection of an anomalous signal is declared i n pres- ence of a time varying mean. The time varying mean is assumed t o follow an ARIMA and the anomalous signal as a GARCH model. The observed signal may possess anomalous pattern for some t ime. This time varying pattern is captured adaptively using an ad aptive CFAR. The propose...

work page

[5] [5]

Outlier detection and t ime series modeling,

Bovas Abraham and Alice Chuang, “Outlier detection and t ime series modeling,” Technometrics, vol. 31, no. 2, pp. 241–248, 1989

work page 1989

[6] [6]

Anomaly detection: A survey,

V arun Chandola, Arindam Banerjee, and Vipin Kumar, “Anomaly detection: A survey,” ACM computing surveys (CSUR), vol. 41, no. 3, pp. 15, 2009

work page 2009

[7] [7]

The forward discount anomaly and the ris k premium: A survey of recent evidence,

Charles Engel, “The forward discount anomaly and the ris k premium: A survey of recent evidence,” Journal of empirical ﬁnance, vol. 3, no. 2, pp. 123–192, 1996

work page 1996

[8] [8]

Combining negative selection and classiﬁcation techniqu es for anomaly detection,

Fabio Gonzalez, Dipankar Dasgupta, and Robert Kozma, “Combining negative selection and classiﬁcation techniqu es for anomaly detection,” in Evolutionary Computation, 2002. CEC’02. Proceedings of the 2002 Congress on . IEEE, 2002, vol. 1, pp. 705–710

work page 2002

[9] [9]

Anomaly-based network intrusion detection: Techniques, systems and challenges,

Pedro Garcia-Teodoro, Jesus Diaz-V erdejo, Gabriel Mac i´ a- Fern´ andez, and Enrique V´ azquez, “Anomaly-based network intrusion detection: Techniques, systems and challenges, ” computers & security, vol. 28, no. 1-2, pp. 18–28, 2009

work page 2009

[10] [10]

A classiﬁca tion framework for anomaly detection,

Ingo Steinwart, Don Hush, and Clint Scovel, “A classiﬁca tion framework for anomaly detection,” Journal of Machine Learn- ing Research, vol. 6, no. Feb, pp. 211–232, 2005

work page 2005

[11] [11]

Transition prob- abilities for general birth–death processes with applicat ions in ecology, genetics, and evolution,

Forrest W Crawford and Marc A Suchard, “Transition prob- abilities for general birth–death processes with applicat ions in ecology, genetics, and evolution,” Journal of mathematical bi- ology, vol. 65, no. 3, pp. 553–580, 2012

work page 2012

[12] [12]

A wavelet-chaos methodology for analysis of eegs and eeg subbands to detect seizure and epilepsy,

Hojjat Adeli, Samanwoy Ghosh-Dastidar, and Nahid Dadme hr, “A wavelet-chaos methodology for analysis of eegs and eeg subbands to detect seizure and epilepsy,” IEEE Transactions on Biomedical Engineering, vol. 54, no. 2, pp. 205–211, 2007

work page 2007

[13] [13]

Conditional anomaly detection,

Xiuyao Song, Mingxi Wu, Christopher Jermaine, and Sanja y Ranka, “Conditional anomaly detection,” IEEE Transactions on Knowledge and Data Engineering , vol. 19, no. 5, pp. 631– 645, 2007

work page 2007

[14] [14]

Anomaly detection using forecasting methods arima and hwds,

Eduardo HM Pena, Marcos VO de Assis, and Mario Lemes Proenc ¸a, “Anomaly detection using forecasting methods arima and hwds,” in Chilean Computer Science Society (SCCC), 2013 32nd International Conference of the . IEEE, 2013, pp. 63–66

work page 2013

[15] [15]

Anomaly detection on social media using arima models,

Tim Isbister, “Anomaly detection on social media using arima models,” 2015

work page 2015

[16] [16]

George EP Box, Gwilym M Jenkins, Gregory C Reinsel, and Greta M Ljung, Time series analysis: forecasting and control , John Wiley & Sons, 2015

work page 2015

[17] [17]

Autoregressive conditional heterosc edasticity with estimates of the variance of united kingdom inﬂation,

Robert F Engle, “Autoregressive conditional heterosc edasticity with estimates of the variance of united kingdom inﬂation,” Econometrica: Journal of the Econometric Society , pp. 987– 1007, 1982

work page 1982

[18] [18]

Generalized autoregressive conditi onal het- eroskedasticity,

Tim Bollerslev, “Generalized autoregressive conditi onal het- eroskedasticity,” Journal of econometrics , vol. 31, no. 3, pp. 307–327, 1986

work page 1986

[19] [19]

An adaptive detection algorithm,

Edward J Kelly, “An adaptive detection algorithm,” IEEE transactions on aerospace and electronic systems , , no. 2, pp. 115–127, 1986

work page 1986

[20] [20]

Constant-false-alarm-rate process ors for lo- cally nonstationary clutter,

Ramon Nitzberg, “Constant-false-alarm-rate process ors for lo- cally nonstationary clutter,” IEEE Transactions on Aerospace and Electronic Systems, , no. 3, pp. 399–405, 1973

work page 1973

[21] [21]

A cfar processor for the detection of unknown random signals in nonstationary correlated noise,

QT Zhang, “A cfar processor for the detection of unknown random signals in nonstationary correlated noise,” Signal pro- cessing, vol. 48, no. 1, pp. 17–26, 1996

work page 1996

[22] [22]

Targ et detection in sea-clutter using stationary wavelet transfo rms,

Vichet Duk, Luke Rosenberg, and Brian Wai-Him Ng, “Targ et detection in sea-clutter using stationary wavelet transfo rms,” IEEE Transactions on Aerospace and Electronic Systems , vol. 53, no. 3, pp. 1136–1146, 2017

work page 2017

[23] [23]

Arima-garch modeling for epileptic seizure prediction,

Salman Mohamadi, Hamidreza Amindavar, and SM Ali Tayaranian Hosseini, “Arima-garch modeling for epileptic seizure prediction,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 994–998

work page 2017