Toward Improving fNIRS Classification: A Study on Activation Functions in Deep Neural Architectures

Behtom Adeli; John McLinden; Ming Shao; Pankaj Pandey; Yalda Shahriari

REVIEW 3 major objections 4 minor 20 references

Toward Improving fNIRS Classification: A Study on Activation Functions in Deep Neural Architectures

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read On fNIRS classification, symmetric activations like Tanh and Abs(x) outperform ReLU in most deep architectures, with gains depending on the network.

desk verdict The MAF symmetry experiment fails its own control, so the paper's central claim about symmetric activations is unsupported despite a useful benchmark. read the letter →

arxiv 2507.11436 v1 pith:ZYG2IUMN submitted 2025-07-15 cs.LG cs.AI

Behtom Adeli , John McLinden , Pankaj Pandey , Ming Shao , Yalda Shahriari This is my paper · ORCID

classification cs.LGcs.AI

keywords functionalnear-infraredspectroscopy(fNIRS)fNIRSclassificationactivationfunctionssymmetricTanhAbsolutefunctionconvolutionalneuralnetworkauditoryoddball

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

Functional near-infrared spectroscopy (fNIRS) records brain hemodynamic responses that are noisy, low in signal-to-noise ratio, and highly variable, yet deep-learning classifiers for fNIRS usually default to ReLU activations without evidence that this suits the signal. This paper compares seven activation functions across four convolutional architectures — fNIRSNet, AbsoluteNet, MDNN, and ShallowConvNet — on a single auditory oddball dataset, holding preprocessing, optimizer, batch size, and training schedule fixed. Its central finding is that symmetric activation functions, especially $\mathrm{Tanh}$ and the absolute-value function $\mathrm{Abs}(x)$, reach higher test accuracy than $\mathrm{ReLU}$ in most of the networks tested. A controlled variant called the Modified Absolute Function interpolates between $\mathrm{ReLU}$ and $\mathrm{Abs}$ and shows that keeping the negative side of neuron pre-activations is what drives the gain. If this holds, fNIRS practitioners can improve classification accuracy by choosing activations matched to the bidirectional nature of hemodynamic signals instead of defaulting to $\mathrm{ReLU}$.

What carries the argument

The central device is activation-function symmetry, meaning evenness, $f(x)=f(-x)$, implemented as the Modified Absolute Function $\mathrm{MAF}(x)=x$ for $x\ge 0$ and $\mathrm{MAF}(x)=\alpha x$ for $x<0$. This one-parameter family contains $\mathrm{ReLU}$ ($\alpha=0$) and $\mathrm{Abs}$ ($\alpha=-1$) as special cases, so sweeping $\alpha$ under a fixed architecture isolates what preserving or discarding the negative pre-activations contributes, independently of network depth or training setup. The four networks supply the test beds: two dual-branch spatio-temporal CNNs (fNIRSNet, AbsoluteNet), a deeper single-branch CNN (MDNN), and a shallow temporal-spatial baseline (ShallowConvNet).

What would settle it

Retrain the same four networks on a second fNIRS dataset, such as motor imagery or cognitive workload, and sweep $\alpha$ in the Modified Absolute Function exactly as this paper does; if $\alpha=0$ ($\mathrm{ReLU}$) matches or beats $\alpha=-1$ ($\mathrm{Abs}$) there, the claim that symmetric activations are inherently better for fNIRS fails to generalize. Alternatively, on the current dataset, remove the sign structure of the inputs (for instance by taking per-channel absolute values before the network); if $\mathrm{Tanh}$ and $\mathrm{Abs}$ then lose their edge over $\mathrm{ReLU}$, the polarity-preservation mechanism is confirmed.

Watch

Extended reading notes

Core claim

The paper's central claim is that activation-function choice materially changes fNIRS classification accuracy, and that symmetric functions are the better default in this domain. Across fNIRSNet, AbsoluteNet, and MDNN, $\mathrm{Tanh}$ and/or $\mathrm{Abs}$ deliver the highest test accuracy — up to $86.33\%$ for $\mathrm{Tanh}$ on AbsoluteNet versus $75.76\%$ for $\mathrm{ReLU}$, and $78.92\%$ for $\mathrm{Abs}$ on MDNN versus $73.96\%$ for $\mathrm{ReLU}$ — while $\mathrm{ReLU}$ is outperformed in most cases and the Square function is consistently the weakest standalone activation. The MAF experiment, which scales negative inputs by $\alpha \in \{-2,-1,0,2\}$, isolates symmetry as the mechanism: $\alpha=-1$ (Abs) and $\alpha=-2$ beat $\alpha=0$ (ReLU) on both networks tested, and $\alpha=2$ degrades performance, in one case to chance level. The authors conclude that fNIRS signals carry meaningful information in both positive and negative polarities around the mean, and that symmetric activations preserve that information.

Load-bearing premise

The load-bearing premise is that fNIRS signals carry class-relevant information in both positive and negative pre-activation values, so a symmetric activation that keeps both polarities produces the accuracy gains; if that polarity information is absent or specific to this auditory task, the advantage over $\mathrm{ReLU}$ will not generalize.

Editorial extensions

If this is right

Choosing $\mathrm{Tanh}$ or $\mathrm{Abs}$ over $\mathrm{ReLU}$ raises test accuracy by several points in the deeper fNIRS-specific networks: AbsoluteNet goes from $75.76\%$ with $\mathrm{ReLU}$ to $86.33\%$ with $\mathrm{Tanh}$, and MDNN from $73.96\%$ to $78.92\%$ with $\mathrm{Abs}$.
The benefit of symmetric activations shrinks in the shallow baseline, so the gain appears to scale with network depth and complexity rather than being a universal constant.
Discarding negative pre-activations, as $\mathrm{ReLU}$ does, limits representational power for fNIRS signals that vary in both directions around a baseline.
The Square function, despite appearing in some domain-specific designs as part of a mixture, performs poorly as a standalone activation and is not recommended alone.
A heterogeneous strategy — different activation functions in different layers or network branches — is the natural next step the paper identifies but does not test.

Reading between the lines

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

Beyond the paper's data, the polarity explanation predicts that any preprocessing that destroys the sign structure of the hemodynamic response — for example, aggressive detrending or per-channel absolute-value scaling — should erase the $\mathrm{Tanh}$/$\mathrm{Abs}$ advantage over $\mathrm{ReLU}$; this is directly testable on the same dataset.
If symmetry is the active ingredient, the result may transfer to other zero-centered, bidirectional neurophysiological signals such as EEG or MEG, but the single-dataset design leaves that open.
A learnable MAF, with $\alpha$ trained per layer or per channel, would let a network decide how much negative information to preserve and could outperform both fixed $\mathrm{ReLU}$ and fixed $\mathrm{Abs}$; the paper's $\alpha$ sweep supplies the baseline for such a variant.
Because only nine subjects contributed the data, the observed gaps may partly reflect subject-specific response shapes; an across-subject replication with more participants is the most direct way to confirm that symmetry itself — not subject idiosyncrasy — drives the result.

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.

Desk Editor's Note

read the letter

First, the good news: this is the first systematic comparison of activation functions for fNIRS classification across multiple architectures, and the main benchmark in Tables III–VI is run carefully—same preprocessing, same optimizer and learning rate, same 5-fold CV, softmax output. The finding that Tanh matches or beats ReLU on all four networks is plausible and worth knowing. The authors also list their limitations honestly (single auditory dataset, homogeneous activation replacement) and don't inflate clinical claims.

The problem is the MAF experiment, which is the paper's key evidence for the symmetry hypothesis. By Eq. (8), MAF(α=-1) is exactly Abs(x) and MAF(α=0) is exactly ReLU(x). Yet Table VII reports AbsoluteNet test accuracies of 85.70±2.75 for α=-1 versus 77.73±4.73 for Absolute in Table IV, and 55.94±8.44 for α=0 versus 75.76±2.92 for ReLU. MDNN shows the same pattern. These gaps are far larger than the reported stds, and the paper says the pipeline is identical, so the controlled comparison is not controlled. Either the implementation changed or the seeds/optimization were not held fixed. That means the MAF numbers cannot isolate symmetry. Table VII also has a clear typo (MDNN α=2: 64.95±64.95), which suggests the table wasn't checked. Without the MAF result, the symmetry claim rests on the main tables, where the evidence is more mixed: Absolute does not beat ReLU in ShallowConvNet, and in fNIRSNet the difference is within noise.

Other soft spots: no statistical significance tests, so many pairwise differences sit inside one std; the dataset is small (9 subjects, 918 trials per class); no code is released. The conclusion that Absolute "consistently ranked among top-performing" is overbroad for the same reason.

Net: the benchmark part is a useful data point, but the paper's central argument—symmetry helps—is not supported by the evidence as presented. I would send this to a serious referee, because the question is relevant and the main tables are publishable after fixing the MAF analysis, adding significance tests, and softening the claims. As a reviewer, I'd ask for a major revision. I would not cite it in its current form.

Referee Report

3 major / 4 minor

Summary. The paper presents an empirical benchmark of seven activation functions (ReLU, ELU, Swish, Sigmoid, Tanh, Square, Absolute) in four deep architectures (fNIRSNet, AbsoluteNet, MDNN, ShallowConvNet) for binary classification of fNIRS data recorded during an auditory oddball task. All models are trained on the same dataset (9 subjects, 918 trials per class) with a unified preprocessing and training protocol, and results are reported as mean ± standard deviation over 5-fold cross-validation. The paper claims that symmetric activation functions such as Tanh and Abs(x) can outperform ReLU in several architectures, and introduces a Modified Absolute Function (MAF) with varying negative-side scaling to isolate the effect of symmetry. The conclusions emphasize that activation-function choice should be aligned with the bidirectional nature of fNIRS signals.

Significance. If the findings are reliable, the paper offers a practically useful guideline for fNIRS deep learning: symmetric activations like Tanh and Absolute may be better default choices than ReLU in domain-specific architectures. The study is commendable for using a standardized pipeline, consistent hyperparameters, and a domain-specific activation family (MAF) to probe symmetry. The paper also includes an explicit limitations section acknowledging the single-dataset scope. However, the evidence base is currently thin: a single small dataset, no statistical significance testing, and an internally inconsistent MAF comparison that undermines the paper's most direct controlled test. Because the central claim is empirical and benchmark-oriented, the value of the paper depends on the soundness of the comparison, which is not yet established.

major comments (3)

[Section IV, Tables IV, V, VII] The MAF experiment is internally inconsistent. By Eq. (8), MAF(x)=x for x≥0 and MAF(x)=αx for x<0, so α=-1 is exactly Abs(x) and α=0 is exactly ReLU(x). Yet the reported test accuracies for these identical functions differ substantially between the main tables and Table VII. For AbsoluteNet, Table VII reports α=-1 as 85.70±2.75 while Table IV reports Abs as 77.73±4.73, and α=0 as 55.94±8.44 while ReLU is 75.76±2.92. For MDNN, α=-1 is 76.53±2.84 versus Abs 78.92±2.56, and α=0 is 72.17±5.57 versus ReLU 73.96±4.13. Since the functions are mathematically identical, the differences imply uncontrolled variation in the experimental pipeline (seeds, initialization, early stopping, or preprocessing) between the two sets of runs. As presented, the MAF analysis does not isolate symmetry; it reintroduces the very confounds it is designed to control. This is load-bearing because the conclusion that the MAF analysis 'confirmed that symmetrical activations like Absolute outperform ReLU' rests on this comparison. The authors must either reconcile the results with identical conditions or remove the MAF-based symmetry claim.
[Table VII] The MDNN α=2 row reports a test accuracy of 64.95 with a standard deviation of 64.95. A standard deviation equal to the mean is implausible for an accuracy measure and strongly suggests a typographical error (e.g., a misplaced decimal or a mis-copied value). This error, combined with the inconsistencies described above, reduces confidence in the reliability of Table VII and should be corrected before the paper can be considered publishable.
[Section III.B and Section IV] The study uses a single dataset of 9 subjects and 5-fold cross-validation, but performs no statistical significance tests to support the ranking of activation functions. Many of the reported differences are within one standard deviation; for example, MDNN Tanh (74.94±3.27) versus ReLU (73.96±4.13) is a difference of less than 1 percentage point, and in ShallowConvNet ReLU (66.56±2.74) actually exceeds Absolute (64.98±4.25). Even larger apparent gaps, such as AbsoluteNet Tanh (86.33±2.10) versus ReLU (75.76±2.92), would be more convincing with a paired test over the five folds or across subjects. Without such tests, the central claim that symmetric activations 'can outperform' ReLU is not statistically established, and the generalization beyond this dataset is speculative. The authors should add appropriate significance tests or explicitly temper the conclusions to the measured folds.

minor comments (4)

[Abstract] The sentence 'nonlinearity, low signal-to-noise ratio (SNR), and signal variability poses significant challenges' has a subject-verb agreement error: 'pose' should be used with the plural subject. Please correct the verb.
[Section IV, first paragraph] The text refers to 'tables III–IV' when results actually span Tables III through VII. This cross-reference is incorrect and should be updated.
[Section IV, MAF paragraph] The paragraph states 'Table IV presents a controlled evaluation of the Modified Absolute Function,' but the MAF results are in Table VII, not Table IV. This is a cross-reference error that should be fixed alongside the data-consistency issues above.
[Table II] The header cell 'Simoid' is a typo for 'Sigmoid.' In the same table, 'Square' is listed as non-smooth, but the square function is smooth; the table appears to conflate smoothness of the function with properties of its gradient at zero. This should be clarified.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; empirical benchmark with self-cited architectures but independent evaluation.

full rationale

This is an empirical benchmark study with no formal derivation chain: the central claim that symmetric activations such as Tanh and Abs can outperform ReLU is supported by measured test accuracies reported in Tables III–VII. The self-citations [16], [17] supply the fNIRSNet and AbsoluteNet architectures, but the activation-function comparison is executed and evaluated in this paper itself rather than imported from those references. No parameter is fitted so as to force the reported ordering, and the MAF alpha sweep is pre-specified via Eq. (8). A consistency concern exists—MAF(alpha=-1) is definitionally Abs and MAF(alpha=0) is definitionally ReLU, yet Table VII reports different accuracies than Tables IV–V (e.g., 85.70 vs 77.73 for AbsoluteNet)—but that is a reproducibility or pipeline-inconsistency issue, not circularity, because the reported numbers are observed outcomes rather than re-statements of the input definitions. No uniqueness theorem or ansatz is imported through citation to force the symmetry conclusion, and the paper explicitly acknowledges its single-dataset limitation. Therefore no significant circularity is found; the minor self-citation is not load-bearing, justifying a score of 1.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

The central claim rests on standard deep-learning assumptions and on the signal-property assumption about bidirectional fNIRS information. No free parameters are fitted; the MAF alpha values are hand-picked experimental conditions rather than fitted constants. The paper introduces no new entities.

assumptions (2)

domain assumption The cited network architectures (fNIRSNet, AbsoluteNet, MDNN, ShallowConvNet) are correctly reimplemented per their original papers.
The evaluation treats the architectures as given and does not re-verify their internal design, so an implementation error would directly affect all comparisons.
domain assumption fNIRS hemodynamic signals carry class-relevant information in both positive and negative deviations from baseline, so retaining both polarities via symmetric activations is beneficial.
This assumption underlies the interpretation of the MAF results and is stated in Section IV and V, but the paper never measures the polarity distribution of pre-activation values independently.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Toward Improving fNIRS Classification: A Study on Activation Functions in Deep Neural Architectures." pith.science (2026). https://pith.science/paper/ZYG2IUMN

@misc{pith2026250711436,
  author       = {Pith},
  title        = {Pith review of: Toward Improving fNIRS Classification: A Study on Activation Functions in Deep Neural Architectures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZYG2IUMN}},
  note         = {Machine review of arXiv:2507.11436}
}

read the original abstract

Activation functions are critical to the performance of deep neural networks, particularly in domains such as functional near-infrared spectroscopy (fNIRS), where nonlinearity, low signal-to-noise ratio (SNR), and signal variability poses significant challenges to model accuracy. However, the impact of activation functions on deep learning (DL) performance in the fNIRS domain remains underexplored and lacks systematic investigation in the current literature. This study evaluates a range of conventional and field-specific activation functions for fNIRS classification tasks using multiple deep learning architectures, including the domain-specific fNIRSNet, AbsoluteNet, MDNN, and shallowConvNet (as the baseline), all tested on a single dataset recorded during an auditory task. To ensure fair a comparison, all networks were trained and tested using standardized preprocessing and consistent training parameters. The results show that symmetrical activation functions such as Tanh and the Absolute value function Abs(x) can outperform commonly used functions like the Rectified Linear Unit (ReLU), depending on the architecture. Additionally, a focused analysis of the role of symmetry was conducted using a Modified Absolute Function (MAF), with results further supporting the effectiveness of symmetrical activation functions on performance gains. These findings underscore the importance of selecting proper activation functions that align with the signal characteristics of fNIRS data.

Figures

Figures reproduced from arXiv: 2507.11436 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 17 canonical work pages

[1]

Y . Liu, J. Zhang, C. Gao, J. Qu, and L. Ji, ”Natural-logari thm-rectiﬁed activation function in convolutional neural networks,” in Proc. IEEE 5th Int. Conf. Comput. Commun. (ICCC), 2019, pp. 2000–2008

work page 2019
[2]

Eastmond, A

C. Eastmond, A. Subedi, S. De, and X. Intes, ”Deep learnin g in fNIRS: a review,” Neurophotonics, vol. 9, no. 4, p. 041411, 2022

work page 2022
[3]

R. H. K. Emanuel, P . D. Docherty, H. Lunt, et al., ”The effe ct of acti- vation functions on accuracy, convergence speed, and miscl assiﬁcation conﬁdence in CNN text classiﬁcation: a comprehensive explo ration”, J. Supercomput., vol. 80, pp. 292–312, 2024. doi: 10.1007/s 11227- 023-05441-7

doi:10.1007/s 2024
[4]

S. R. Dubey, S. K. Singh, and B. B. Chaudhuri, ”Activa- tion functions in deep learning: A comprehensive survey and benchmark,” Neurocomputing, vol. 503, pp. 92–108, 2022. do i: 10.1016/j.neucom.2022.06.111

doi:10.1016/j.neucom.2022.06.111 2022
[5]

Rawat and C

U. Rawat and C. S. Rai, ”Advancing brain tumor detection w ith neurofusion: an innovative CNN-LSTM model featuring a nove l activation function,” Service Oriented Computing and Applications , pp. 1–19, 2024

work page 2024
[6]

Y . Wang, Y . Li, Y . Song, and X. Rong, ”The inﬂuence of the ac tivation function in a convolution neural network model of facial exp ression recognition,” Applied Sciences , vol. 10, no. 5, p. 1897, 2020

work page 2020
[7]

D. B. Mehta, P . A. Barot, and S. G. Langhnoja, ”Effect of di fferent activation functions on EEG signal classiﬁcation based on n eural networks”, in Proc. 4th Int. Conf. Comput. Methodol. Commun . (ICCMC), Erode, India, 2020, pp. 132–135

work page 2020
[8]

Tanveer, M

M. Tanveer, M. J. Khan, M. J. Qureshi, N. Naseer, and K.-S. Hong, ”Enhanced drowsiness detection using deep learning: an fNI RS study,” IEEE Access, vol. 7, pp. 137920–137929, 2019

work page 2019

Show all 20 references

[9]

Mirbagheri, A

M. Mirbagheri, A. Jodeiri, N. Hakimi, V . Zakeri, and S. K. Setare- hdan, ”Accurate stress assessment based on functional near infrared spectroscopy using deep learning approach,” in Proc. 26th N at. 4th Int. Iranian Conf. Biomed. Eng. (ICBME), Tehran, Iran, 2019, pp. 4–10. 10...

2019
[10]

R., Abtahi, M., & Mankodiya, K

Dargazany, A. R., Abtahi, M., & Mankodiya, K. An end-to- end (deep) neural network applied to raw EEG, fNIRs and body motion data for data fusion and BCI classiﬁcation task without any pre-/ post- processing. arXiv preprint, 2019, arXiv:1907.09523

2019 arXiv
[11]

Saadati, J

M. Saadati, J. Nelson, and H. Ayaz, ”Mental workload cla ssiﬁcation from spatial representation of fNIRS recordings using conv olutional neural networks,” in Proc. IEEE 29th Int. Workshop Mach. Lea rn. Signal Process. (MLSP), 2019, pp. 1–6

2019
[12]

Y ang, R

D. Y ang, R. Huang, S. H. Y oo, M. J. Shin, J. A. Y oon, Y . I. Sh in, and K. S. Hong, ”Detection of mild cognitive impairment usin g convolutional neural network: temporal-feature maps of fu nctional near-infrared spectroscopy,” Front. Aging Neurosci., vol . 12, p. 141, 2020

2020
[13]

Ghonchi, M

H. Ghonchi, M. Fateh, V . Abolghasemi, S. Ferdowsi, and M . Rezvani, ”Deep recurrent–convolutional neural network for classiﬁ cation of simultaneous EEG–fNIRS signals”, IET Signal Process., vol . 14, no. 3, pp. 142–153, 2020

2020
[14]

T. Ma, S. Wang, Y . Xia, X. Zhu, J. Evans, Y . Sun, and S. He, ” CNN- based classiﬁcation of fNIRS signals in motor imagery BCI sy stem,” J. Neural Eng., vol. 18, no. 5, 2021. doi: 10.1088/1741-2552 /abf187

2021 doi
[15]

Q. He, L. Feng, G. Jiang, and P . Xie, ”Multimodal multita sk neural network for motor imagery classiﬁcation with EEG and fNIRS s ig- nals,” IEEE Sens. J., vol. 21, no. 2, pp. 20695–20706, 2022

2022
[16]

Pandey, J

P . Pandey, J. McLinden, N. Rahimi, C. Kumar, M. Shao, K. M . Spencer, S. Ostadabbas, and Y . Shahriari, ”fNIRSNet: A mult i- view spatio-temporal convolutional neural network fusion for func- tional near-infrared spectroscopy-based auditory event c lassiﬁcation,” Eng. Appl. A...

2024
[17]

Adeli, J

B. Adeli, J. McLinden, P . Pandey, M. Shao, and Y . Shahria ri, ”AbsoluteNet: A deep learning neural network to classify ce rebral hemodynamic responses of auditory processing”, IEEE Inter national Workshop on Machine Learning for Signal Processing (MLSP), Istan- bul, Turkey, ...

2025
[18]

R. T. Schirrmeister, J. T. Springenberg, L. D. J. Fieder er, M. Glasstet- ter, K. Eggensperger, M. Tangermann, and T. Ball, ”Deep learning with convolutional neural networks for EEG decoding and visuali zation,” Hum. Brain Mapp., vol. 38, no. 11, pp. 5391–5420, 2017

2017
[19]

Elfwing, E

S. Elfwing, E. Uchibe, and K. Doya, ”Sigmoid-weighted l inear units for neural network function approximation in reinforcemen t learning,” Neural Networks , vol. 107, pp. 3–11, 2018

2018
[20]

McLinden, S

J. McLinden, S. B. Borgheai, S. I. Hosni, C. Kumar, N. Rah imi, M. Shao, et al., ”Individual-speciﬁc characterization of e vent- related hemodynamic responses during an auditory task: an e x- ploratory study,” Behav. Brain Res., vol. 436, p. 114074, 20 23. doi: 10.1016/j.bbr.2...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Y . Liu, J. Zhang, C. Gao, J. Qu, and L. Ji, ”Natural-logari thm-rectiﬁed activation function in convolutional neural networks,” in Proc. IEEE 5th Int. Conf. Comput. Commun. (ICCC), 2019, pp. 2000–2008

work page 2019

[2] [2]

Eastmond, A

C. Eastmond, A. Subedi, S. De, and X. Intes, ”Deep learnin g in fNIRS: a review,” Neurophotonics, vol. 9, no. 4, p. 041411, 2022

work page 2022

[3] [3]

R. H. K. Emanuel, P . D. Docherty, H. Lunt, et al., ”The effe ct of acti- vation functions on accuracy, convergence speed, and miscl assiﬁcation conﬁdence in CNN text classiﬁcation: a comprehensive explo ration”, J. Supercomput., vol. 80, pp. 292–312, 2024. doi: 10.1007/s 11227- 023-05441-7

doi:10.1007/s 2024

[4] [4]

S. R. Dubey, S. K. Singh, and B. B. Chaudhuri, ”Activa- tion functions in deep learning: A comprehensive survey and benchmark,” Neurocomputing, vol. 503, pp. 92–108, 2022. do i: 10.1016/j.neucom.2022.06.111

doi:10.1016/j.neucom.2022.06.111 2022

[5] [5]

Rawat and C

U. Rawat and C. S. Rai, ”Advancing brain tumor detection w ith neurofusion: an innovative CNN-LSTM model featuring a nove l activation function,” Service Oriented Computing and Applications , pp. 1–19, 2024

work page 2024

[6] [6]

Y . Wang, Y . Li, Y . Song, and X. Rong, ”The inﬂuence of the ac tivation function in a convolution neural network model of facial exp ression recognition,” Applied Sciences , vol. 10, no. 5, p. 1897, 2020

work page 2020

[7] [7]

D. B. Mehta, P . A. Barot, and S. G. Langhnoja, ”Effect of di fferent activation functions on EEG signal classiﬁcation based on n eural networks”, in Proc. 4th Int. Conf. Comput. Methodol. Commun . (ICCMC), Erode, India, 2020, pp. 132–135

work page 2020

[8] [8]

Tanveer, M

M. Tanveer, M. J. Khan, M. J. Qureshi, N. Naseer, and K.-S. Hong, ”Enhanced drowsiness detection using deep learning: an fNI RS study,” IEEE Access, vol. 7, pp. 137920–137929, 2019

work page 2019