Towards Reliable Brain-Computer Interface: Achieving Perfect Accuracy by Sacrificing Time
Name
Jevgeni Savostkin
Abstract
Brain-computer interface (BCI) is a computer system for extracting brain electric
neural signals and using them to control computer applications. For the operation
BCI requires a user to concentrate on some mental tasks. Besides measuring
the signals, BCI converts raw electric signal to digital representation and maps the
data to computer commands. Unfortunately, the probability of predicting the right
command is below 100% and therefore the reliability of these systems is relatively
low.
Low reliability is a huge problem for BCI, since they will not be widely trusted
and used while the prediction accuracy is low. The existing solutions usually try
to improve the prediction accuracy of BCI without focusing too much on the time
what is required for a single user’s concentration attempt. They apply different
prediction models and signal processing techniques in order to raise the accuracy
of prediction. Our solution goes the opposite way – it tries to discover how many
concentration attempts should be done in a row (i.e how long does it take), to
guarantee the prediction accuracy of 99%.
The solution described in the thesis is based on Condorcet’s jury theorem [1].
It states that if we have two options and the chance to pick correct is larger than
50%, then, if we make several attempts in a row, the probability to pick the correct
option by majority vote is rising with the number of attempts. In this work we
apply the main Condorcet’s principle in a BCI perspective. First we develop a
system that can reach the single concentration attempt’s prediction accuracy to
be more than 50% and then we use multiple concentration attempts in a row to
improve the overall accuracy. We expect that given enough attempts we can reach
99% classification accuracy. We compare the empirical results with the theoretical
estimates and discuss them.
The BCI technology is a relatively young field. In order to fully integrate it
into our ordinary life, the contribution from scientists and engineers is required for
converting BCI to a reliable system. The following work contributes to reliability
of BCI systems.
neural signals and using them to control computer applications. For the operation
BCI requires a user to concentrate on some mental tasks. Besides measuring
the signals, BCI converts raw electric signal to digital representation and maps the
data to computer commands. Unfortunately, the probability of predicting the right
command is below 100% and therefore the reliability of these systems is relatively
low.
Low reliability is a huge problem for BCI, since they will not be widely trusted
and used while the prediction accuracy is low. The existing solutions usually try
to improve the prediction accuracy of BCI without focusing too much on the time
what is required for a single user’s concentration attempt. They apply different
prediction models and signal processing techniques in order to raise the accuracy
of prediction. Our solution goes the opposite way – it tries to discover how many
concentration attempts should be done in a row (i.e how long does it take), to
guarantee the prediction accuracy of 99%.
The solution described in the thesis is based on Condorcet’s jury theorem [1].
It states that if we have two options and the chance to pick correct is larger than
50%, then, if we make several attempts in a row, the probability to pick the correct
option by majority vote is rising with the number of attempts. In this work we
apply the main Condorcet’s principle in a BCI perspective. First we develop a
system that can reach the single concentration attempt’s prediction accuracy to
be more than 50% and then we use multiple concentration attempts in a row to
improve the overall accuracy. We expect that given enough attempts we can reach
99% classification accuracy. We compare the empirical results with the theoretical
estimates and discuss them.
The BCI technology is a relatively young field. In order to fully integrate it
into our ordinary life, the contribution from scientists and engineers is required for
converting BCI to a reliable system. The following work contributes to reliability
of BCI systems.
Graduation Thesis language
English
Graduation Thesis type
Master - Software Engineering
Supervisor(s)
Ilya Kuzovkin, Raul Vicente
Defence year
2017