FactorizationMachine: Second order Factorization Machines

In dselivanov/rsparse: Statistical Learning on Sparse Matrices

Second order Factorization Machines

Description

Creates second order Factorization Machines model

Methods

Public methods

• FactorizationMachine$new()

• FactorizationMachine$partial_fit()

• FactorizationMachine$fit()

• FactorizationMachine$predict()

• FactorizationMachine$clone()

---

Method new()

creates Creates second order Factorization Machines model

Usage

FactorizationMachine$new(
  learning_rate_w = 0.2,
  rank = 4,
  lambda_w = 0,
  lambda_v = 0,
  family = c("binomial", "gaussian"),
  intercept = TRUE,
  learning_rate_v = learning_rate_w
)

Arguments

learning_rate_w

learning rate for features intercations

rank

dimension of the latent dimensions which models features interactions

lambda_w

regularization for features interactions

lambda_v

regularization for features

family

one of "binomial", "gaussian"

intercept

logical, indicates whether or not include intecept to the model

learning_rate_v

learning rate for features

---

Method partial_fit()

fits/updates model

Usage

FactorizationMachine$partial_fit(x, y, weights = rep(1, length(y)), ...)

Arguments

x

input sparse matrix. Native format is Matrix::RsparseMatrix. If x is in different format, model will try to convert it to RsparseMatrix with as(x, "RsparseMatrix"). Dimensions should be (n_samples, n_features)

y

vector of targets

weights

numeric vector of length 'n_samples'. Defines how to amplify SGD updates for each sample. May be useful for highly unbalanced problems.

...

not used at the moment

---

Method fit()

shorthand for applying 'partial_fit' 'n_iter' times

Usage

FactorizationMachine$fit(x, y, weights = rep(1, length(y)), n_iter = 1L, ...)

Arguments

x

input sparse matrix. Native format is Matrix::RsparseMatrix. If x is in different format, model will try to convert it to RsparseMatrix with as(x, "RsparseMatrix"). Dimensions should be (n_samples, n_features)

y

vector of targets

weights

numeric vector of length 'n_samples'. Defines how to amplify SGD updates for each sample. May be useful for highly unbalanced problems.

n_iter

number of SGD epochs

...

not used at the moment

---

Method predict()

makes predictions based on fitted model

Usage

FactorizationMachine$predict(x, ...)

Arguments

x

input sparse matrix of shape (n_samples, n_featires)

...

not used at the moment

---

Method clone()

The objects of this class are cloneable with this method.

Usage

FactorizationMachine$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

# Factorization Machines can fit XOR function!
x = rbind(
  c(0, 0),
  c(0, 1),
  c(1, 0),
  c(1, 1)
)
y = c(0, 1, 1, 0)

x = as(x, "RsparseMatrix")
fm = FactorizationMachine$new(learning_rate_w = 10, rank = 2, lambda_w = 0,
  lambda_v = 0, family = 'binomial', intercept = TRUE)
res = fm$fit(x, y, n_iter = 100)
preds = fm$predict(x)
all(preds[c(1, 4)] < 0.01)
all(preds[c(2, 3)] > 0.99)

dselivanov/rsparse documentation built on Feb. 20, 2025, 12:30 p.m.