Compile a factorization network — factor_net • RcppML

Validates the graph topology, resolves config inheritance, and returns a compiled network ready for fit().

Usage

factor_net(inputs, output, config = factor_config())

Arguments

inputs: A single fn_node (input) or a list of input nodes.
output: The output fn_node (typically a layer).
config: A fn_global_config from factor_config(). Default uses factor_config() defaults.

Value

A factor_net object.

See also

fit, factor_config, nmf_layer, svd_layer, cross_validate_graph

Examples

data(aml)
inp <- factor_input(aml, "aml")
out <- nmf_layer(inp, k = 5)
net <- factor_net(inp, out, config = factor_config(maxit = 10))
net
#> factor_net: 1 layer(s), 1 input(s)
#>   L1: NMF(k=5)
#>   config: maxit=10, tol=0.0001, loss=mse