Pith. sign in

REVIEW 3 major objections 6 minor 68 references

Ensemble Machine Learning Model for Inner Speech Recognition: A Subject-Specific Investigation

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Stacking five logistic regression models after per-subject MRMR feature selection classifies four inner-speech words from 128-channel surface EEG with 81.13% accuracy and 81.12% F1, the paper reports.

desk verdict The 81% accuracy claim is likely inflated by test-set leakage in feature selection and K tuning; the paper is a competent but incremental benchmark that needs a redo before its main number can be trusted. read the letter →

arxiv 2412.17824 v1 pith:XC5TMDTL submitted 2024-12-09 eess.SP cs.CL

Shahamat Mustavi Tasin , Muhammad E. H. Chowdhury , Shona Pedersen , Malek Chabbouh , Diala Bushnaq , Raghad Aljindi , Saidul Kabir , Anwarul Hasan This is my paper · ORCID
classification eess.SPcs.CL
keywords InnerSpeechBrain-ComputerInterfaceEEGMachineLearningClassificationEnsembleModelMRMRLogisticRegression
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 inner speech can be decoded from non-invasive scalp EEG when each subject is modelled separately. Using the public Thinking Out Loud dataset, the authors extract 191 time-, frequency-, and time-frequency-domain features from each of 128 channels, reduce the resulting 24,448 features per subject with Minimum Redundancy Maximum Relevance (MRMR), and classify four Spanish inner-speech words with six machine learning classifiers. Their central result is that an ensemble formed by stacking five logistic regression models reaches an overall accuracy of 81.13% and an F1 score of 81.12%, the best result they report for this dataset. If correct, this would show that a computationally light, subject-specific classical machine learning pipeline is a viable route for inner speech brain-computer interfaces.

What carries the argument

The central mechanism is the pairing of subject-specific MRMR feature selection with a stacked ensemble of five logistic regression classifiers. MRMR (Minimum Redundancy Maximum Relevance) scores each feature by the ratio of its relevance to the target word to its redundancy with previously selected features, allowing the authors to cut a 24,448-dimensional feature space down to K=590 features per subject. The paper reports that this reduction, together with the averaging effect of five logistic regression models, raises overall accuracy from 77.89% for a single logistic regression model to 81.13% for the ensemble.

What would settle it

Re-run the subject-specific pipeline with feature selection and the choice of K=590 performed entirely inside each training fold, keeping the same five-logistic-regression ensemble; if the resulting held-out accuracy falls substantially below 81.13% or no longer beats the single logistic regression baseline, the paper's headline claim is not supported.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that per-subject feature selection plus a stacked ensemble of simple linear classifiers solves the four-word inner speech decoding task on this dataset. The authors report that the ensemble of five logistic regression models outperforms every individual classifier they tested, including support vector machines and extra trees, and outperforms the previous deep learning work on the Thinking Out Loud dataset by a large margin. They also report that seven of the ten subjects reach above 80% accuracy, with subject 08 reaching 93.00%, while three subjects (03, 04, and 05) remain at or below about 72%, which they explain by overlapping topographical brain activations for the four words in those subjects. The authors interpret the subject-dependent results as evidence that inner speech production is neurally idiosyncratic and therefore needs subject-specific modelling.

Load-bearing premise

The reported result assumes that choosing which and how many features to keep using the same subject data that later measures accuracy does not leak information from the test trials into the model.

Editorial extensions

If this is right

  • Subject-specific modelling is a more reliable starting point than a one-size-fits-all model when decoding inner speech from EEG, because the same word activates different scalp regions in different subjects.
  • MRMR-style feature selection can compress tens of thousands of EEG features to a few hundred without degrading, and in this paper improving, classification accuracy.
  • On small single-subject EEG datasets, a pipeline of engineered features plus classical classifiers can outperform deep neural networks that need more training data.
  • Combining several logistic regression models into an ensemble is a cheap way to gain around three percentage points of accuracy over the single-model baseline.

Reading between the lines

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

  • A leak-free replication that selects features and the K=590 count inside each training fold is the natural next experiment; the meaningful question is whether the stacked ensemble still beats the single logistic regression and the EEGNet baseline under that protocol.
  • The large subject-to-subject spread (93.00% for subject 08 versus 55.00% for subject 04) suggests that individual neural discriminability, not just the algorithm, determines performance, so future work could try to predict which subjects will benefit most from this pipeline.
  • A cross-lingual or cross-session test with different words would show whether the feature set captures general inner-speech correlates or just word-specific templates in this Spanish cohort.
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

3 major / 6 minor

Summary. The paper proposes a subject-specific machine learning pipeline for inner speech recognition from surface EEG using the publicly available Thinking Out Loud dataset. The authors extract 191 features per channel, compare eight feature selection techniques, evaluate six classical ML and three deep learning models, and propose an ensemble of five logistic regression models, reporting an overall accuracy of 81.13% and F1 of 81.12%. The central claim is that this subject-specific approach with MRMR feature selection surpasses prior work on this dataset.

Significance. If the reported evaluation were valid, the paper would provide a strong empirical demonstration that classical ML with MRMR feature selection can achieve high accuracy in four-class inner speech decoding from surface EEG, which would be a meaningful advance for BCI applications. The study is also notable for its systematic comparison of eight feature selection methods, six classifiers, and three deep learning architectures on a public dataset. However, the validity of the central accuracy claim is compromised by a leakage in the evaluation protocol, as detailed below.

major comments (3)
  1. [§2.6.1 and §3.1] The MRMR feature selection is applied to the full data of each subject before the 10-fold cross-validation split, and the feature count K is selected by varying it from 20 to 610 and choosing K=590 based on the accuracy/F1 curves obtained on the same data. Because the test folds' label information is used to select the features and to tune K, the reported 81.13% accuracy is an optimistically biased estimate of generalization performance. The authors need to perform feature selection and K selection inside each training fold (nested cross-validation) or use a separate held-out validation set, and then report the resulting performance.
  2. [§3.2] The ensemble model selection also leaks test information: the choice to stack five LR models rather than the SVM+ETC+LR ensemble is made by comparing the Table 2 accuracies, which are themselves computed on the same cross-validation folds used for the final evaluation. This additional model-selection step compounds the bias. Furthermore, the manuscript does not specify how the five LR models are combined (e.g., majority voting, averaging, or a meta-classifier), which makes the ensemble result irreproducible.
  3. [Table 2 and §3.2] The deep learning results are at or near chance (1D CNN 23.73%, 2D CNN Resnet18 31.83%) while the classical ML models report 75–81% accuracy for the same four-class problem. The authors attribute this gap to small sample size, but this explanation is not fully convincing given that the same data are used. This large discrepancy reinforces the concern that the ML accuracy is inflated by the feature-selection and model-selection leakage. A properly nested evaluation is needed to determine whether the high accuracy persists when no test-fold information is used.
minor comments (6)
  1. [§2.7] There is a typo: "It it to be noted" should read "It is to be noted".
  2. [§3.1] The feature selection method name is inconsistently written as "ANOV A" in Table 1; it should be "ANOVA".
  3. [§3.2 and Figure 13 caption] There are typos: "consfusion matrix" should be "confusion matrix" and "chnnels" should be "channels".
  4. [Table 2] The numeric values for the 1D and 2D CNN results (23.732, 31.826, 28.004) use an inconsistent decimal separator; they should be written as 23.73, 31.83, and 28.00 for consistency with the rest of the table.
  5. [§3.2] The phrase "the five best logistic regression models" is ambiguous: it is unclear what distinguishes these five LR models (e.g., different feature subsets, different hyperparameters, or different random seeds). The authors should clarify the construction of the ensemble.
  6. [Data Availability Statement] The statement says the processed dataset can be made available upon request, but the original dataset is publicly available; the authors should clarify what exactly would be provided beyond the public data.

Circularity Check

3 steps flagged · score 6.0 of 10

The 81.13% accuracy claim is an in-sample-tuned maximum rather than an independent prediction because MRMR feature ranking, K=590, and the final ensemble are selected using the same CV folds that produce the reported score.

  1. fitted input called prediction [Section 3.1, Figure 10 and hyperparameter tuning of MRMR feature selection; final result in Section 3.2, Table 2]
    "In MRMR algorithm, the number of features is characterized by the parameter “K” and it is varied from 20 to 610 in this analysis. From Figure 10, subject 06 shows an accuracy and F1 score of 87.04% and 86.95% respectively at K = 590, which is the highest among all the feature sets. As a result, 590 features are selected using the MRMR method for each subject, and the performance of different ML and DL models are carried out according to the evaluation methods described in Section 2.7 and 2.8."

    K=590 is chosen because it produced the highest accuracy/F1 among all tested K values, and the same cross-validation estimates are then reported as the model's final performance. The reported 81.13% accuracy is therefore not the prediction of a fixed pipeline; it is the selected maximum of a tuning sweep over K on the very evaluation data used to produce the headline result. The final number is forced by this selection rather than being an independent out-of-sample estimate.

  2. other [Section 2.6.1 and Section 2.8, MRMR feature selection followed by 10-fold CV]
    "In this work, MRMR is applied to the data of each subject separately in the subject-specific investigation. ... For evaluation of ML classifers and 1D CNN model, a 10-fold CV scheme has been followed. In this method, the data is split into ten folds where nine folds are used for model training and the remaining fold is used for validation and this process is repeated ten times."

    MRMR feature selection is applied to the full data of each subject before the data is split into folds, so the relevance and redundancy scores used to rank features incorporate the labels of all trials, including the trials that later appear in validation folds. As a result, the reported cross-validation accuracy is not a clean out-of-sample estimate: the feature ranking has already seen the test folds. The headline accuracy is inflated by this leakage and is partly fitted to the evaluation data.

1 more flagged steps
  1. fitted input called prediction [Section 3.2, selection of the final ensemble from Table 2]
    "It is observed from Table 2 that the LR model has produced 77.89% accuracy and 77.91% F1 score in this dataset, which is the highest among the basic ML models. ... Accordingly, the performance of an ensemble model, comprising of one SVM, one ETC, and one LR model is found to be better than that of the LR model, with a 1.22% and 1.18% increase in accuracy and F1 score respectively."

    The final model is selected after comparing its accuracy on the same cross-validation splits used to report the headline result. The 81.13% figure is the best among the configurations tried, so model selection is itself fitted to the test folds. The reported performance is not a prediction from a pre-specified model; it is the maximum of a post hoc model search conducted on the evaluation data.

full rationale

This is an empirical benchmark paper with no mathematical derivation, so the circularity is evaluative rather than theoretical. The central result (81.13% accuracy and 81.12% F1, Section 3.2) depends on a chain of choices made from the target data: MRMR is applied to each subject's full data before splitting; K is varied from 20 to 610 and K=590 is selected because it gives the highest CV accuracy; and the final ensemble is selected after comparing CV results of several configurations. Each step uses test-fold information that the reported numbers are supposed to estimate, so the headline number effectively reduces to a fitted maximum over feature count and model configuration. This does not make the method invalid, but it invalidates the claim that 81.13% is an unbiased out-of-sample result; proper nested cross-validation or a fully independent tuning set could yield lower performance. No self-citation chain or imported uniqueness theorem is involved.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard ML assumptions plus a problematic methodological assumption: feature selection and hyperparameter tuning on the full dataset before cross-validation. No new physical entities are introduced.

free parameters (2)
  • K (number of MRMR-selected features) = 590
    Chosen by varying K from 20 to 610 and selecting the value with the highest accuracy on the evaluation data (Section 3.1, Figure 10). This is hyperparameter tuning on the test set and biases the reported 81.13%.
  • Number of VMD modes = 6
    The VMD is set to generate six modes and the lowest-frequency IMF is discarded (Section 2.4). This choice is not validated against a ground-truth artifact signal and could affect the cleaned data used for classification.
assumptions (4)
  • domain assumption The Thinking Out Loud dataset provides reliable labels for inner speech trials and the processed data is free of major artifacts.
    The study relies on the public dataset's trial labels and pre-processing by the dataset authors without independent validation.
  • domain assumption Discarding the lowest-frequency VMD mode removes motion artifacts while preserving inner-speech-related EEG activity.
    Section 2.4 states the lowest-frequency IMF is discarded, but no quantitative validation is provided that this removes artifacts without removing signal.
  • domain assumption The 191 features per channel capture discriminative information for inner speech in the action interval.
    The feature set is taken from prior EMG/EEG literature and assumed to transfer to inner speech classification.
  • ad hoc to paper Cross-validation with feature selection applied before folding is unbiased.
    The paper applies MRMR to the whole subject data before CV (Section 2.6.1), which assumes no information leakage. This assumption is violated and is central to the reported accuracy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Ensemble Machine Learning Model for Inner Speech Recognition: A Subject-Specific Investigation." pith.science (2026). https://pith.science/paper/XC5TMDTL

@misc{pith2026241217824,
  author       = {Pith},
  title        = {Pith review of: Ensemble Machine Learning Model for Inner Speech Recognition: A Subject-Specific Investigation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XC5TMDTL}},
  note         = {Machine review of arXiv:2412.17824}
}
read the original abstract

Inner speech recognition has gained enormous interest in recent years due to its applications in rehabilitation, developing assistive technology, and cognitive assessment. However, since language and speech productions are a complex process, for which identifying speech components has remained a challenging task. Different approaches were taken previously to reach this goal, but new approaches remain to be explored. Also, a subject-oriented analysis is necessary to understand the underlying brain dynamics during inner speech production, which can bring novel methods to neurological research. A publicly available dataset, Thinking Out Loud Dataset, has been used to develop a Machine Learning (ML)-based technique to classify inner speech using 128-channel surface EEG signals. The dataset is collected on a Spanish cohort of ten subjects while uttering four words (Arriba, Abajo, Derecha, and Izquierda) by each participant. Statistical methods were employed to detect and remove motion artifacts from the Electroencephalography (EEG) signals. A large number (191 per channel) of time-, frequency- and time-frequency-domain features were extracted. Eight feature selection algorithms are explored, and the best feature selection technique is selected for subsequent evaluations. The performance of six ML algorithms is evaluated, and an ensemble model is proposed. Deep Learning (DL) models are also explored, and the results are compared with the classical ML approach. The proposed ensemble model, by stacking the five best logistic regression models, generated an overall accuracy of 81.13% and an F1 score of 81.12% in the classification of four inner speech words using surface EEG signals. The proposed framework with the proposed ensemble of classical ML models shows promise in the classification of inner speech using surface EEG signals.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 66 canonical work pages

  1. [1]

    up," "down,

    Introduction Language and speech processing has always been an intriguing and complicated topic in neurological research. This is because a number of cerebral structures and networks associated with factors like emotional processing, contextual meaning, and cognitive regulation collectively add to the core language system [1]. A number of models have been...

    work page

  2. [2]

    Thinking Out Loud Dataset

    Materials and Methods For the identification of inner speech from EEG signals, a publicly available dataset [23] has been used. This section provides a detailed discussion of the study materials and methods, which contains dataset 3 description, data processing steps taken by the authors of the dataset, Event -Related Potential (ERP) analysis of the EEG d...

    work page

  3. [3]

    Mutual Information (MI) [44]

    work page

  4. [4]

    Pearson-Coefficient (PC) [45]

    work page

  5. [5]

    Decision Tree-Based Selection [46]

    work page

  6. [6]

    Principal Component Analysis (PCA) [47] 8

    work page

  7. [7]

    The Origin of Speech,

    C. Holden, "The Origin of Speech," Science, vol. 303, pp. 1316-1319, 2004

    work page 2004

  8. [8]

    selecting K best features instead of best K features

    Minimum Redundancy Maximum Relevance (MRMR) [49] Preliminary experiments showed that the MRMR method is the most suitable approach in this study , and therefore, it is discussed briefly in the following section. 2.6.1 MRMR Minimum Redundancy Maximum Relevance (MRMR) is an algorithm for identifying a subset of relevant features while minimizing redundancy....

    work page

Show all 68 references
  1. [9]

    It expresses the accuracy as the ratio of accurate predictions to the total number of predictions

    Accuracy: The accuracy metric measures the proportion of true predictions made by a classifier. It expresses the accuracy as the ratio of accurate predictions to the total number of predictions. As demonstrated by Equation (2), accuracy is determined by dividing the sum of Tru...

  2. [10]

    The F1 score is calculated as the harmonic mean of precision and recall, providing a well-rounded evaluation of a model's performance

    F1 Score: F1 score measures the classification accuracy of a model by considering both the correct positive predictions (precision) and the ability to capture true positive instances (recall). The F1 score is calculated as the harmonic mean of precision and recall, providing a...

  3. [11]

    Arriba” and “Izquierda

    Results and Discussion This study evaluates the efficacy of eight feature selection techniques, with one technique (MRMR) standing out and being selected for all subjects. Analysing the effect of various parameters on particular subjects leads to the selection of optimal hyper...

  4. [12]

    Data were collected from 10 healthy subjects, and four words were uttered by each of them during inner speech tasks

    Conclusion In this study, a publicly available dataset has been used for inner speech recognition from EEG signals. Data were collected from 10 healthy subjects, and four words were uttered by each of them during inner speech tasks. The data have been pre-processed and feature...

  5. [13]

    The Margins of the Language Network in the Brain,

    I. Hertrich, S. Dietrich, and H. Ackermann, "The Margins of the Language Network in the Brain," Frontiers in Communication, vol. 5, 2020-November-06 2020

    2020

  6. [14]

    The Organization of Language and the Brain,

    N. Geschwind, "The Organization of Language and the Brain," Science, vol. 170, pp. 940 -944, 1970

    1970

  7. [15]

    Monkey to human comparative anatomy of the frontal lobe association tracts,

    M. T. de Schotten, F. Dell’Acqua, R. Valabregue, and M. Catani, "Monkey to human comparative anatomy of the frontal lobe association tracts," Cortex, vol. 48, pp. 82-96, 2012

    2012

  8. [16]

    Evolution of the neural language network,

    A. D. Friederici, "Evolution of the neural language network," Psychonomic bulletin & review, vol. 24, pp. 41-47, 2017. 17

    2017

  9. [17]

    Words in the brain's language,

    F. Pulvermüller, "Words in the brain's language," Behavioral and brain sciences, vol. 22, pp. 253- 279, 1999

    1999

  10. [18]

    A review and synthesis of the first 20 years of PET and fMRI studies of heard speech, spoken language and reading,

    C. J. Price, "A review and synthesis of the first 20 years of PET and fMRI studies of heard speech, spoken language and reading," Neuroimage, vol. 62, pp. 816-847, 2012

    2012

  11. [19]

    Neural reuse of action perception circuits for language, concepts and communication,

    F. Pulvermüller, "Neural reuse of action perception circuits for language, concepts and communication," Progress in neurobiology, vol. 160, pp. 1-44, 2018

    2018

  12. [20]

    Body-part-specific Representations of Semantic Noun Categories,

    F. Carota, R. Moseley, and F. Pulvermüller, "Body-part-specific Representations of Semantic Noun Categories," Journal of Cognitive Neuroscience, vol. 24, pp. 1492-1509, 2012

    2012

  13. [21]

    The motor features of action verbs: fMRI evidence using picture naming,

    Y . Zhang, K. Wang, C. Yue, N. Mo, D. Wu, X. Wen, et al., "The motor features of action verbs: fMRI evidence using picture naming," Brain and Language, vol. 179, pp. 22-32, 2018

    2018

  14. [22]

    Interpersonal processes and brain sciences —A new anthropology,

    R. McClelland, "Interpersonal processes and brain sciences —A new anthropology," European Psychiatry, vol. 11, pp. 224s-224s, 1996

    1996

  15. [23]

    Emotion recognition from EEG using higher order crossings,

    P. C. Petrantonakis and L. J. Hadjileontiadis, "Emotion recognition from EEG using higher order crossings," IEEE Transactions on information Technology in Biomedicine, vol. 14, pp. 186 -197, 2009

    2009

  16. [24]

    Multiple kernel learning based on three discriminant features for a P300 speller BCI,

    K. Yoon and K. Kim, "Multiple kernel learning based on three discriminant features for a P300 speller BCI," Neurocomputing, vol. 237, pp. 133-144, 2017

    2017

  17. [25]

    Sonification and textification: Proposing methods for classifying unspoken words from EEG signals,

    E. F. González -Castañeda, A. A. Torres -García, C. A. Reyes -García, and L. Villaseñor -Pineda, "Sonification and textification: Proposing methods for classifying unspoken words from EEG signals," Biomedical Signal Processing and Control, vol. 37, pp. 82-91, 2017

    2017

  18. [26]

    Classifying phonological categories in imagined and articulated speech,

    S. Zhao and F. Rudzicz, "Classifying phonological categories in imagined and articulated speech," in 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2015, pp. 992-996

    2015

  19. [27]

    Covert speech vs. motor imagery: a comparative study of class separability in identical environments,

    A. Jahangiri, J. M. Chau, D. R. Achanccaray, and F. Sepulveda, "Covert speech vs. motor imagery: a comparative study of class separability in identical environments," in 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) , ...

    2018

  20. [28]

    Jahangiri and F

    A. Jahangiri and F. Sepulveda, "The relative contribution of high -gamma linguistic processing stages of word production, and motor imagery of articulation in class separability of covert speech tasks in EEG data," Journal of medical systems, vol. 43, pp. 1-9, 2019

    2019

  21. [29]

    Optimizing layers improves CNN generalization and transfer learning for imagined speech decoding from EEG,

    C. Cooney, R. Folli, and D. Coyle, "Optimizing layers improves CNN generalization and transfer learning for imagined speech decoding from EEG," in 2019 IEEE international conference on systems, man and cybernetics (SMC), 2019, pp. 1311-1316

    2019

  22. [30]

    Classification of V owels from Imagined Speech with Convolutional Neural Networks,

    M.-O. Tamm, Y . Muhammad, and N. Muhammad, "Classification of V owels from Imagined Speech with Convolutional Neural Networks," Computers, vol. 9, p. 46, 2020

    2020

  23. [31]

    Learning patterns in imaginary vowels for an intelligent brain computer interface (BCI) design,

    P. Ghane and G. Hossain, "Learning patterns in imaginary vowels for an intelligent brain computer interface (BCI) design," arXiv preprint arXiv:2010.12066, 2020

    2010 arXiv

  24. [32]

    Multiclass covert speech classification using extreme learning machine,

    D. Pawar and S. Dhage, "Multiclass covert speech classification using extreme learning machine," Biomedical Engineering Letters, vol. 10, pp. 217-226, 2020

    2020

  25. [33]

    Inferring imagined speech using EEG signals: a new approach using Riemannian manifold features,

    C. H. Nguyen, G. K. Karavas, and P. Artemiadis, "Inferring imagined speech using EEG signals: a new approach using Riemannian manifold features," Journal of neural engineering, vol. 15, p. 016002, 2017

    2017

  26. [34]

    Thinking out loud, an open-access EEG-based BCI dataset for inner speech recognition,

    N. Nieto, V . Peterson, H. L. Rufiner, J. E. Kamienkowski, and R. Spies, "Thinking out loud, an open-access EEG-based BCI dataset for inner speech recognition," Scientific Data, vol. 9, p. 52, 2022

    2022

  27. [35]

    Nidal and A

    K. Nidal and A. S. Malik, EEG/ERP analysis: methods and applications: Crc Press, 2014. 18

    2014

  28. [36]

    A. N. Akansu and R. A. Haddad, Multiresolution signal decomposition: transforms, subbands, and wavelets: Academic press, 2001

    2001

  29. [37]

    Motion artifacts correction from single -channel EEG and fNIRS signals using novel wavelet packet decomposition in combination with canonical correlation analysis,

    M. S. Hossain, M. E. Chowdhury, M. B. I. Reaz, S. H. M. Ali, A. A. A. Bakar, S. Kiranyaz, et al., "Motion artifacts correction from single -channel EEG and fNIRS signals using novel wavelet packet decomposition in combination with canonical correlation analysis," Sensors, vol....

    2022

  30. [38]

    The empirical mode decomposition and the Hilbert spectrum for nonlinear and non -stationary time series analysis,

    N. E. Huang, Z. Shen, S. R. Long, M. C. Wu, H. H. Shih, Q. Zheng , et al., "The empirical mode decomposition and the Hilbert spectrum for nonlinear and non -stationary time series analysis," Proceedings of the Royal Society of London. Series A: mathematical, physical and engin...

    1998

  31. [39]

    Ensemble empirical mode decomposition: a noise-assisted data analysis method,

    Z. Wu and N. E. Huang, "Ensemble empirical mode decomposition: a noise-assisted data analysis method," Advances in adaptive data analysis, vol. 1, pp. 1-41, 2009

    2009

  32. [40]

    Singular-spectrum analysis: A toolkit for short, noisy chaotic signals,

    R. Vautard, P. Yiou, and M. Ghil, "Singular-spectrum analysis: A toolkit for short, noisy chaotic signals," Physica D: Nonlinear Phenomena, vol. 58, pp. 95-126, 1992

    1992

  33. [41]

    Variational mode decomposition,

    K. Dragomiretskiy and D. Zosso, "Variational mode decomposition," IEEE transactions on signal processing, vol. 62, pp. 531-544, 2013

    2013

  34. [42]

    Novel approaches for the removal of motion artifact from EEG recordings,

    P. Gajbhiye, R. K. Tripathy, A. Bhattacharyya, and R. B. Pachori, "Novel approaches for the removal of motion artifact from EEG recordings," IEEE Sensors Journal, vol. 19, pp. 10600 - 10608, 2019

    2019

  35. [43]

    Motion artifacts correction from EEG and fNIRS signals using novel multiresolution analysis,

    M. S. Hossain, M. B. I. Reaz, M. E. Chowdhury, S. H. Ali, A. A. A. Bakar, S. Kiranyaz , et al. , "Motion artifacts correction from EEG and fNIRS signals using novel multiresolution analysis," IEEE Access, vol. 10, pp. 29760-29777, 2022

    2022

  36. [44]

    A novel non -invasive estimation of respiration rate from motion corrupted photoplethysmograph signal using machine learning model,

    M. N. I. Shuzan, M. H. Chowdhury, M. S. Hossain, M. E. Chowdhury, M. B. I. Reaz, M. M. Uddin, et al. , "A novel non -invasive estimation of respiration rate from motion corrupted photoplethysmograph signal using machine learning model," IEEE Access, vol. 9, pp. 96775 - 96790, 2021

    2021

  37. [45]

    Study of stability of time -domain features for electromyographic pattern recognition,

    D. Tkach, H. Huang, and T. A. Kuiken, "Study of stability of time -domain features for electromyographic pattern recognition," Journal of neuroengineering and rehabilitation, vol. 7, pp. 1-13, 2010

    2010

  38. [46]

    Comparison of different time and frequency domain feature extraction methods on elbow gesture’s EMG,

    C. Altın and O. Er, "Comparison of different time and frequency domain feature extraction methods on elbow gesture’s EMG," European journal of interdisciplinary studies, vol. 2, pp. 35- 44, 2016

    2016

  39. [47]

    Frequency domain characteristics of ground reaction forces during walking of young and elderly females,

    N. Stergiou, G. Giakas, J. E. Byrne, and V . Pomeroy, "Frequency domain characteristics of ground reaction forces during walking of young and elderly females," Clinical Biomechanics, vol. 17, pp. 615-617, 2002

    2002

  40. [48]

    Blind source separation of single-sensor recordings: Application to ground reaction force signals,

    R. El Halabi, "Blind source separation of single-sensor recordings: Application to ground reaction force signals," Université de Lyon, 2018

    2018

  41. [49]

    Machine learning -based classification of healthy and impaired gaits using 3D -GRF signals,

    M. N. I. Shuzan, M. E. Chowdhury, M. B. I. Reaz, A. Khandakar, F. F. Abir, M. A. A. Faisal, et al., "Machine learning -based classification of healthy and impaired gaits using 3D -GRF signals," Biomedical Signal Processing and Control, vol. 81, p. 104448, 2023

    2023

  42. [50]

    DCT domain feature extraction scheme based on motor unit action potential of EMG signal for neuromuscular disease classification,

    A. B. M. S. U. Doulah, S. A. Fattah, W. P. Zhu, and M. O. Ahmad, "DCT domain feature extraction scheme based on motor unit action potential of EMG signal for neuromuscular disease classification," Healthcare technology letters, vol. 1, pp. 26-31, 2014

    2014

  43. [51]

    A mother wavelet selection study for vertical ground reaction force signals,

    R. Soubra, M. O. Diab, and B. Moslem, "A mother wavelet selection study for vertical ground reaction force signals," in 2017 2nd International Conference on Bio -engineering for Smart Technologies (BioSMART), 2017, pp. 1-5

    2017

  44. [52]

    Overfitting and undercomputing in machine learning,

    T. Dietterich, "Overfitting and undercomputing in machine learning," ACM computing surveys (CSUR), vol. 27, pp. 326-327, 1995. 19

    1995

  45. [53]

    Modular neural -SVM scheme for speech emotion recognition using ANOV A feature selection method,

    M. Sheikhan, M. Bejani, and D. Gharavian, "Modular neural -SVM scheme for speech emotion recognition using ANOV A feature selection method," Neural Computing and Applications, vol. 23, pp. 215-227, 2013

    2013

  46. [54]

    Performance assessment of artificial neural network using chi -square and backward elimination feature selection methods for landslide susceptibility analysis,

    B. T. Pham, D. Van Dao, T. D. Acharya, T. Van Phong, R. Costache, H. Van Le, et al., "Performance assessment of artificial neural network using chi -square and backward elimination feature selection methods for landslide susceptibility analysis," Environmental Earth Sciences, ...

    2021

  47. [55]

    Mutual information -based feature selection for multilabel classification,

    G. Doquire and M. Verleysen, "Mutual information -based feature selection for multilabel classification," Neurocomputing, vol. 122, pp. 148-155, 2013

    2013

  48. [56]

    Daily activity feature selection in smart homes based on pearson correlation coefficient,

    Y . Liu, Y . Mu, K. Chen, Y . Li, and J. Guo, "Daily activity feature selection in smart homes based on pearson correlation coefficient," Neural Processing Letters, vol. 51, pp. 1771-1787, 2020

    2020

  49. [57]

    Predicting breast cancer from risk factors using SVM and extra -trees-based feature selection method,

    G. Alfian, M. Syafrudin, I. Fahrurrozi, N. L. Fitriyani, F. T. D. Atmaji, T. Widodo, et al., "Predicting breast cancer from risk factors using SVM and extra -trees-based feature selection method," Computers, vol. 11, p. 136, 2022

    2022

  50. [58]

    Principal component analysis,

    H. Abdi and L. J. Williams, "Principal component analysis," Wiley interdisciplinary reviews: computational statistics, vol. 2, pp. 433-459, 2010

    2010

  51. [59]

    ReliefF for multi -label feature selection,

    N. Spolaôr, E. A. Cherman, M. C. Monard, and H. D. Lee, "ReliefF for multi -label feature selection," in 2013 Brazilian Conference on Intelligent Systems, 2013, pp. 6-11

    2013

  52. [60]

    Maximum relevance and minimum redundancy feature selection methods for a marketing machine learning platform,

    Z. Zhao, R. Anand, and M. Wang, "Maximum relevance and minimum redundancy feature selection methods for a marketing machine learning platform," in 2019 IEEE international conference on data science and advanced analytics (DSAA), 2019, pp. 442-452

    2019

  53. [61]

    The best two independent measurements are not the two best,

    T. M. Cover, "The best two independent measurements are not the two best," IEEE Transactions on Systems, Man, and Cybernetics, pp. 116-117, 1974

    1974

  54. [62]

    Mazzanti

    S. Mazzanti. (2021, 10/6/2023). “MRMR” Explained Exactly How You Wished Someone Explained to You

    2021

  55. [63]

    Can Ensemble of Classifiers Provide Better Recognition Results in Packaging Activity?,

    A. Nazmus Sakib, P. Basak, S. Doha Uddin, S. Mustavi Tasin, and M. A. R. Ahad, "Can Ensemble of Classifiers Provide Better Recognition Results in Packaging Activity?," in Sensor-and Video- Based Activity and Behavior Computing: Proceedings of 3rd International Conference on Ac...

    2021

  56. [64]

    Heart Failure Emergency Readmission Prediction Using Stacking Machine Learning Model,

    M. S. Rahman, H. R. Rahman, J. Prithula, M. E. Chowdhury, M. U. Ahmed, J. Kumar, et al., "Heart Failure Emergency Readmission Prediction Using Stacking Machine Learning Model," Diagnostics, vol. 13, p. 1948, 2023

    1948

  57. [65]

    Machine learning with ensemble stacking model for automated sleep staging using dual -channel EEG signal,

    S. K. Satapathy, A. K. Bhoi, D. Loganathan, B. Khandelwal, and P. Barsocchi, "Machine learning with ensemble stacking model for automated sleep staging using dual -channel EEG signal," Biomedical Signal Processing and Control, vol. 69, p. 102898, 2021

    2021

  58. [66]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770 - 778

    2016

  59. [67]

    Very deep convolutional networks for large -scale image recognition,

    K. Simonyan and A. Zisserman, "Very deep convolutional networks for large -scale image recognition," arXiv preprint arXiv:1409.1556, 2014

    2014 arXiv

  60. [68]

    Inner speech classification using eeg signals: A deep learning approach,

    B. van den Berg, S. van Donkelaar, and M. Alimardani, "Inner speech classification using eeg signals: A deep learning approach," in 2021 IEEE 2nd International Conference on Human - Machine Systems (ICHMS), 2021, pp. 1-4

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.