A non-parametric approach to extending generic binary classifiers for multi-classification

Santhanam, Venkataraman; Morariu, Vlad I.; Harwood, David; Davis, Larry S.

doi:10.1016/j.patcog.2016.04.008

Cited by 21 publications

(12 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although these strategies are effective, they encounter two main obstacles. First, the resulting complicated optimization problem generated by some of these methods leads to time‐consuming training procedure (Galar et al, ; Liu et al, ; Santhanam et al, ) particularly in the case where DL models are exploited as base classifiers. Second, the noisy high‐dimensional feature space of MI tasks hinders the effective use of the OVA and OVO‐based strategies.…”

Section: Introductionmentioning

confidence: 99%

A novel end‐to‐end deep learning scheme for classifying multi‐class motor imagery electroencephalography signals

et al. 2019

View full text Add to dashboard Cite

An important subfield of brain–computer interface is the classification of motor imagery (MI) signals where a presumed action, for example, imagining the hands' motions, is mentally simulated. The brain dynamics of MI is usually measured by electroencephalography (EEG) due to its noninvasiveness. The next generation of brain–computer interface systems can benefit from the generative deep learning (GDL) models by providing end‐to‐end (e2e) machine learning and increasing their accuracy. In this study, to exploit the e2e‐property of deep learning models, a novel GDL methodology is proposed where only minimal objective‐free preprocessing steps are needed. Furthermore, to deal with the complicated multi‐class MI–EEG signals, an innovative multilevel GDL‐based classifying scheme is proposed. The effectiveness of the proposed model and its robustness against noisy MI–EEG signals is evaluated using two different GDL models, that is, deep belief network and stacked sparse autoencoder in e2e manner. Experimental results demonstrate the effectiveness of the proposed methodology with improved accuracy compared with the widely used filter bank common spatial patterns algorithm.

show abstract

Section: Introductionmentioning

confidence: 99%

A novel end‐to‐end deep learning scheme for classifying multi‐class motor imagery electroencephalography signals

et al. 2019

View full text Add to dashboard Cite

show abstract

“…The trip purposes of the respondents' trips were identified by the trained CHMM model trained first, and then compared with the real trip purposes reported by the respondents, so as the accuracy of the proposed method were verified. Since the label variables were added into the questionnaire, the reliability evaluation of the model could be conducted according to the multi-classification problems [35]. Each activity in the trip-chains was regarded as a sample for metrics calculation, so that more calculation samples were used during the evaluation.…”

Section: B Model Validation Resultsmentioning

confidence: 99%

Trip Purpose Identification of Docked Bike-Sharing From IC Card Data Using a Continuous Hidden Markov Model

Cao³

et al. 2020

IEEE Access

View full text Add to dashboard Cite

It is different from the previous supervised learning algorithm based on personal travel questionnaire, the aim of this study is to develop an unsupervised learning methodology to estimate the docked bike-sharing users' trip purposes using IC card data, which trip purposes were unknown from the dataset. The present study is able to extract the trip-chains, which is used to understand the complete individual trip process. A rigorous method is then proposed to interpret the purpose of each leg of the tripchain using a continuous hidden Markov model (CHMM). This method effectively combines the Gaussian mixture model and the hidden Markov model, and realizes the inference based on trip-chains. It is intended to enhance the understanding of docked bike-sharing users' transfer intention, which is different from most trip motivation recognition methods. The Gaussian mixture layer uses the feature space constructed by the spatial and temporal information on trip-chains from the IC card data, as well as the land-use characteristics of the docked bike-sharing docking stations to complete the transfer of the trip-chains to the trip modes. The hidden Markov structure can realize the process from the trip modes to the trip purposes. The IC card data of docked bike-sharing usage in Nanjing, China is used to interpret the specific steps of the proposed model. A questionnaire survey is conducted to obtain the real trip purposes, which is compared with the estimated results from the model to verify the effectiveness of the model.. The results show that the accuracies of single trip recognition and chain trip recognition are 0.770 and 0.756, respectively. Compared with the baseline algorithm, the model also shows good performance. Therefore, the proposed approach can be used to discover and interpret the trip purpose using the IC card data. INDEX TERMS Continuous hidden Markov model, IC card, docked bike-sharing, trip-chain, trip purpose

show abstract

“…However, the ratio of competent to non-competent classifiers becomes 1:1 for data with four classes and monotonically decreases in favor of non-competent classifiers as the number of classes increases. In these imbalanced multiclass settings, a more sophisticated approach using some form of weighted voting should be used instead [60,62].…”

Section: Cassini Simulationsmentioning

confidence: 99%

A random forests quantile classifier for class imbalanced data

O’Brien

Ishwaran

2019

Pattern Recognition

129

View full text Add to dashboard Cite

Extending previous work on quantile classifiers (q-classifiers) we propose the q*-classifier for the class imbalance problem. The classifier assigns a sample to the minority class if the minority class conditional probability exceeds 0 < q* < 1, where q* equals the unconditional probability of observing a minority class sample. The motivation for q*-classification stems from a densitybased approach and leads to the useful property that the q*-classifier maximizes the sum of the true positive and true negative rates. Moreover, because the procedure can be equivalently expressed as a cost-weighted Bayes classifier, it also minimizes weighted risk. Because of this dual optimization, the q*-classifier can achieve near zero risk in imbalance problems, while simultaneously optimizing true positive and true negative rates. We use random forests to apply q*-classification. This new method which we call RFQ is shown to outperform or is competitive with existing techniques with respect to tt-mean performance and variable selection. Extensions to the multiclass imbalanced setting are also considered.

show abstract

A non-parametric approach to extending generic binary classifiers for multi-classification

Cited by 21 publications

References 33 publications

A novel end‐to‐end deep learning scheme for classifying multi‐class motor imagery electroencephalography signals

A novel end‐to‐end deep learning scheme for classifying multi‐class motor imagery electroencephalography signals

Trip Purpose Identification of Docked Bike-Sharing From IC Card Data Using a Continuous Hidden Markov Model

A random forests quantile classifier for class imbalanced data

Contact Info

Product

Resources

About