Predict Spectral Embeddings

Projects test data into the forest embedding space using a pre-trained encoding map.

Usage

# S3 method for class 'encode'
predict(object, rf, x, parallel = TRUE, ...)

Arguments

object

Spectral embedding for the rf learned via eigenmap.

rf

Pre-trained random forest object of class ranger.

x

Data to be embedded.

parallel

Compute in parallel? Must register backend beforehand, e.g. via doParallel.

...

Additional arguments passed to methods.

Value

A matrix of embeddings, with nrow(x) rows and k columns, the latter argument used to learn the eigenmap.

Details

This function uses the weights learned via eigenmap to project new data into the low-dimensional embedding space using the Nyström formula. For details, see Bengio et al. (2004).

References

Bengio, Y., Delalleau, O., Le Roux, N., Paiement, J., Vincent, P., & Ouimet, M. (2004). Learning eigenfunctions links spectral embedding and kernel PCA. Neural Computation, 16(10): 2197-2219.

See also

adversarial_rf

Examples

# Set seed
set.seed(1)

# Split training and test
trn <- sample(1:nrow(iris), 100)
tst <- setdiff(1:nrow(iris), trn)

# Train RF. You can also use RF variants, such as the Adversarial Random
# Forests (ARF).
rf <- ranger::ranger(Species ~ ., data = iris[trn, ], num.trees=50)

# Learn the encodings, which are found using diffusion maps.
emap <- encode(rf, iris[trn, ], k=2)

# Embed test points
emb <- predict(emap, rf, iris[tst, ])