Biomedical Engineering Reference
In-Depth Information
TaBlE 4.3:
Comparison of testing performance for PLS and NLMS
NlMS
PlS
Testing correlation coefficient
for a reaching task
0.75 ± 0.20
0.77 ± 0.18
with a random initial vector
w
(0).
T
(0 <
T
< 1) is a balancing parameter to remove the oscillating be-
havior near convergence. The convergence rate is affected by
T
that produces a trade-off between the
convergence speed and the accuracy. Note that the fastest convergence can be obtained with
T
= 1.
The consecutive projection vectors are also learned by the deflation method to form in each column
of a projection matrix
w
. After projection onto the subspace by
w
, we embed the input signal at
each channel with an
L
-tap delay line and design the Wiener filter to estimate the HP. Figure
4.3
illustrates the overall diagram of the subspace Wiener filter. For this architecture, the testing perfor-
mance of the model also improved with a decrease in CC variance as shown in Table
4.3
.
4.2 ChaNNEl SElECTIoN
Because our ultimate goal for BMIs is to design the most accurate reconstructions of hand kine-
matics from cortical activity using adaptive signal processing techniques, it seems natural to equate
here neural importance (selection of important neurons) to model fitting quality. Moreover, these
measures should be compared with the available neurophysiological knowledge, with the hope that
we can understand better the data, and enhance our methodologies and, ultimately, the performance
of BMIs. Therefore, the importance of neurons will be ascertained using three techniques:
•
•
•
Sensitivity analysis through trained linear and nonlinear models
L
1
-norm penalty pruning
Real-time input selection
Given a set of data, we would like to evaluate how well these three methodologies are able to
find important neurons for building BMI models. Second, we would like to use this information to
tackle the model generalization issues encountered in BMIs. The goals of the study are formulated
in the following questions:
•
Can our methods automatically indicate important neurons for the prediction of the kine-
matic variables in the tasks studied?
