Shared factorization across multiple inputs (multi-modal)

Creates a node representing shared-H factorization of multiple input matrices. The resulting H is shared across all inputs; each input gets its own W.

Usage

factor_shared(...)

Arguments

...

Two or more fn_node objects (inputs or layers).

Value

An fn_node of type "shared".

Details

Execution concatenates inputs row-wise and runs a single NMF: rbind(X1, X2, ...) = rbind(W1, W2, ...) * diag(d) * H.

See also

factor_concat, factor_add, factor_net

Examples

data(aml)
# Split into two views
inp1 <- factor_input(aml[1:400, ], name = "view1")
inp2 <- factor_input(aml[401:824, ], name = "view2")
shared <- factor_shared(inp1, inp2)