parmigene (version 1.1.0)

mrnet: Maximum Relevance Minimum Redundancy

Description

A function that infers the interaction network using the MRNET algorithm.

Usage

mrnet(mi)

Arguments

mi

matrix of the mutual information.

Value

A square weighted adjacency matrix of the inferred network.

Details

The MRNET approach starts by selecting the variable \(X_i\) having the highest mutual information with the target Y.

Then, it repeatedly enlarges the set of selected variables \(S\) by taking the \(X_k\) that maximizes

$$I(X_k;Y) - mean(I(X_k;X_i))$$

for all \(X_i\) already in S.

The procedure stops when the score becomes negative.

By default, the function uses all the available cores. You can set the actual number of threads used to N by exporting the environment variable OMP_NUM_THREADS=N.

References

H. Peng, F.long and C.Ding. Feature selection based on mutual information: Criteria of max-dependency, max relevance and min redundancy. IEEE transaction on Pattern Analysis and Machine Intelligence, 2005.

See Also

aracne.a

aracne.m

clr

Examples

# NOT RUN {
mat <- matrix(rnorm(1000), nrow=10)
mi  <- knnmi.all(mat)
grn <- mrnet(mi)
# }