Abstract
In the design of brain-machine interface (BMI) algorithms, the activity of hundreds of chronically recorded neurons is used to reconstruct a variety of kinematic variables. A significant problem introduced with the use of neural ensemble inputs for model building is the explosion in the number of free parameters. Large models not only affect model generalization but also put a computational burden on computing an optimal solution especially when the goal is to implement the BMI in low-power, portable hardware. In this paper, three methods are presented to quantitatively rate the importance of neurons in neural to motor mapping, using single neuron correlation analysis, sensitivity analysis through a vector linear model, and a model-independent cellular directional tuning analysis for comparisons purpose. Although, the rankings are not identical, up to sixty percent of the top 10 ranking cells were in common. This set can then be used to determine a reduced-order model whose performance is similar to that of the ensemble. It is further shown that by pruning the initial ensemble neural input with the ranked importance of cells, a reduced sets of cells (between 40 and 80, depending upon the methods) can be found that exceed the BMI performance levels of the full ensemble.
| Original language | English |
|---|---|
| Pages (from-to) | 943-953 |
| Number of pages | 11 |
| Journal | IEEE Transactions on Biomedical Engineering |
| Volume | 51 |
| Issue number | 6 |
| DOIs | |
| State | Published - Jun 1 2004 |
| Externally published | Yes |
Fingerprint
Keywords
- Brain-machine interface
- Cosine tuning
- Information in neural ensembles
- Sensitivity-based model pruning
ASJC Scopus subject areas
- Biomedical Engineering
Cite this
Ascertaining the importance of neurons to develop better brain-machine interfaces. / Sanchez, Justin C.; Carmena, Jose M.; Lebedev, Mikhail A.; Nicolelis, Miguel A L; Harris, John G.; Principe, Jose C.
In: IEEE Transactions on Biomedical Engineering, Vol. 51, No. 6, 01.06.2004, p. 943-953.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Ascertaining the importance of neurons to develop better brain-machine interfaces
AU - Sanchez, Justin C.
AU - Carmena, Jose M.
AU - Lebedev, Mikhail A.
AU - Nicolelis, Miguel A L
AU - Harris, John G.
AU - Principe, Jose C.
PY - 2004/6/1
Y1 - 2004/6/1
N2 - In the design of brain-machine interface (BMI) algorithms, the activity of hundreds of chronically recorded neurons is used to reconstruct a variety of kinematic variables. A significant problem introduced with the use of neural ensemble inputs for model building is the explosion in the number of free parameters. Large models not only affect model generalization but also put a computational burden on computing an optimal solution especially when the goal is to implement the BMI in low-power, portable hardware. In this paper, three methods are presented to quantitatively rate the importance of neurons in neural to motor mapping, using single neuron correlation analysis, sensitivity analysis through a vector linear model, and a model-independent cellular directional tuning analysis for comparisons purpose. Although, the rankings are not identical, up to sixty percent of the top 10 ranking cells were in common. This set can then be used to determine a reduced-order model whose performance is similar to that of the ensemble. It is further shown that by pruning the initial ensemble neural input with the ranked importance of cells, a reduced sets of cells (between 40 and 80, depending upon the methods) can be found that exceed the BMI performance levels of the full ensemble.
AB - In the design of brain-machine interface (BMI) algorithms, the activity of hundreds of chronically recorded neurons is used to reconstruct a variety of kinematic variables. A significant problem introduced with the use of neural ensemble inputs for model building is the explosion in the number of free parameters. Large models not only affect model generalization but also put a computational burden on computing an optimal solution especially when the goal is to implement the BMI in low-power, portable hardware. In this paper, three methods are presented to quantitatively rate the importance of neurons in neural to motor mapping, using single neuron correlation analysis, sensitivity analysis through a vector linear model, and a model-independent cellular directional tuning analysis for comparisons purpose. Although, the rankings are not identical, up to sixty percent of the top 10 ranking cells were in common. This set can then be used to determine a reduced-order model whose performance is similar to that of the ensemble. It is further shown that by pruning the initial ensemble neural input with the ranked importance of cells, a reduced sets of cells (between 40 and 80, depending upon the methods) can be found that exceed the BMI performance levels of the full ensemble.
KW - Brain-machine interface
KW - Cosine tuning
KW - Information in neural ensembles
KW - Sensitivity-based model pruning
UR - http://www.scopus.com/inward/record.url?scp=2442686774&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=2442686774&partnerID=8YFLogxK
U2 - 10.1109/TBME.2004.827061
DO - 10.1109/TBME.2004.827061
M3 - Article
C2 - 15188862
AN - SCOPUS:2442686774
VL - 51
SP - 943
EP - 953
JO - IEEE Transactions on Biomedical Engineering
JF - IEEE Transactions on Biomedical Engineering
SN - 0018-9294
IS - 6
ER -