Tongue-Movement Communication and Control Concept for Hands-Free Human–Machine Interfaces

Vaidyanathan, Ravi; Chung, Beomsu; Gupta, Lalit; Kook, Hyunseok; Kota, Srinivas; West, J.

doi:10.1109/tsmca.2007.897919

Cited by 53 publications

(59 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It was demonstrated that the ear-pressure signals are distinct for each tongue movement and the signals can be classified accurately. Consequently, tongue movements can be mapped, via ear pressure signals, to generate control signals for HMIs without inserting any device in the oral cavity [4]. We have also demonstrated that spoken words can be recognized from the ear pressure signals [3].…”

Section: Single Channel Classification Examplementioning

confidence: 81%

“…Each movement was repeated at least 100 times so that 100 tongue movements could be randomly selected to represent each tongue movement class. The signals were filtered and segmented using the techniques developed in [4]. The durations of the TMEP segments were approximately 200 msec, that is 400 samples.…”

Section: Single Channel Classification Examplementioning

confidence: 99%

“…In the previous work reported in [4], exactly the same TMEP signals were classified using a decision fusion classifier, a matched filter, a Gaussian classifier using AR model parameters, and a nonlinear alignment classifier. It was shown that, in general, the best results were obtained using the decision fusion classifier.…”

Section: Tmep Signal Classificationmentioning

confidence: 99%

“…It is not unusual for the resulting dimension to be high especially for signals that are voluntarily generated by humans for human-machine-interface (HMI) control and communication applications. Examples of such signals include speech [1]- [3], ear pressure signals [3], [4], electromyographic signals [5], [6], electroencephalogram (EEG) and event-related potential (ERP) brainwaveforms [6]- [9], and gestures [10], [11]. Due to practical issues related to the data acquisition methods, lack of concentration, discomfort, and fatigue, it may not always be possible to collect enough reliable signals to exceed the dimension of the signal space.…”

Section: Introductionmentioning

confidence: 99%

“…These particular signal classification problems are selected because they are typical of signal classification problems in which the curse of dimensionality is known to occur frequently. Furthermore, the signals are also typical of signals that can be used to control HMIs [3], [4], [11].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Feature Ranking Strategy to Facilitate Multivariate Signal Classification

Gupta

Kota

Murali

et al. 2010

IEEE Trans. Syst., Man, Cybern. C

View full text Add to dashboard Cite

A strategy is introduced to rank and select principal component transform (PCT) and discrete transform (DCT) transform coefficient features to overcome the curse of dimensionality frequently encountered in implementing multivariate signal classifiers due to small sample sizes. The criteria considered for ranking include the magnitude, variance, inter-class separation, and classification accuracies of the individual features. The feature ranking and selection strategy is applied to overcome the dimensionality problem which often plagues the implementation and evaluation of practical Gaussian signal classifiers.The applications of the resulting PCT-and DCT-Gaussian signal classification strategies are demonstrated by classifying single-channel tongue-movement ear-pressure signals and multi-channel event-related potentials. Through these experiments, it is shown that the dimension of the feature space can be decreased quite significantly by means of the feature ranking and selection strategy. The ranking strategy not only facilitates overcoming the dimensionality curse for multivariate classifier implementation but also provides a means to further select, out of a rank-ordered set, a smaller set of features that give the best classification accuracies. Results show that the PCT-and DCT-Gaussian classifiers yield higher classification accuracies than those reported in previous classification studies on the same signal sets. Amongst the combinations of the two transforms and four feature selection criteria, the PCT-Gaussian classifiers using the maximum magnitude and maximum variance selection criteria gave the best classification accuracies across the two sets of classification experiments. Most noteworthy is the fact that the multivariate Gaussian signal classifiers developed in this paper can be implemented without having to collect a prohibitively large number of training signals simply to satisfy the dimensionality conditions. Consequently, the classification strategies can be beneficial for designing personalized human-machine-interface (HMI) signal classifiers for individuals from whom only a limited number of training signals can reliably be collected due to severe disabilities.

show abstract

Section: Single Channel Classification Examplementioning

confidence: 81%

Section: Single Channel Classification Examplementioning

confidence: 99%

Section: Tmep Signal Classificationmentioning

confidence: 99%