"""Implementation of FM."""
from ..bases import ModelMeta, TfBase
from ..feature.multi_sparse import true_sparse_field_size
from ..layers import embedding_lookup, tf_dense
from ..tfops import dropout_config, reg_config, tf
from ..tfops.features import compute_dense_feats, compute_sparse_feats
from ..utils.misc import count_params
from ..utils.validate import (
check_dense_values,
check_multi_sparse,
check_sparse_indices,
dense_field_size,
sparse_feat_size,
sparse_field_size,
)
[docs]class FM(TfBase, metaclass=ModelMeta):
"""*Factorization Machines* algorithm.
Note this implementation is actually a mixture of FM and NFM, since it uses one
dense layer in the final output
Parameters
----------
task : {'rating', 'ranking'}
Recommendation task. See :ref:`Task`.
data_info : :class:`~libreco.data.DataInfo` object
Object that contains useful information for training and inference.
loss_type : {'cross_entropy', 'focal'}, default: 'cross_entropy'
Loss for model training.
embed_size: int, default: 16
Vector size of embeddings.
n_epochs: int, default: 10
Number of epochs for training.
lr : float, default 0.001
Learning rate for training.
lr_decay : bool, default: False
Whether to use learning rate decay.
epsilon : float, default: 1e-5
A small constant added to the denominator to improve numerical stability in
Adam optimizer.
According to the `official comment <https://github.com/tensorflow/tensorflow/blob/v1.15.0/tensorflow/python/training/adam.py#L64>`_,
default value of `1e-8` for `epsilon` is generally not good, so here we choose `1e-5`.
Users can try tuning this hyperparameter if the training is unstable.
reg : float or None, default: None
Regularization parameter, must be non-negative or None.
batch_size : int, default: 256
Batch size for training.
sampler : {'random', 'unconsumed', 'popular'}, default: 'random'
Negative sampling strategy.
- ``'random'`` means random sampling.
- ``'unconsumed'`` samples items that the target user did not consume before.
- ``'popular'`` has a higher probability to sample popular items as negative samples.
.. versionadded:: 1.1.0
num_neg : int, default: 1
Number of negative samples for each positive sample, only used in `ranking` task.
use_bn : bool, default: True
Whether to use batch normalization.
dropout_rate : float or None, default: None
Probability of an element to be zeroed. If it is None, dropout is not used.
multi_sparse_combiner : {'normal', 'mean', 'sum', 'sqrtn'}, default: 'sqrtn'
Options for combining `multi_sparse` features.
seed : int, default: 42
Random seed.
lower_upper_bound : tuple or None, default: None
Lower and upper score bound for `rating` task.
tf_sess_config : dict or None, default: None
Optional TensorFlow session config, see `ConfigProto options
<https://github.com/tensorflow/tensorflow/blob/v2.10.0/tensorflow/core/protobuf/config.proto#L431>`_.
References
----------
[1] *Steffen Rendle* `Factorization Machines
<https://www.csie.ntu.edu.tw/~b97053/paper/Rendle2010FM.pdf>`_.
[2] *Xiangnan He et al.* `Neural Factorization Machines for Sparse Predictive Analytics
<https://arxiv.org/pdf/1708.05027.pdf>`_.
"""
user_variables = ("embedding/user_linear_var", "embedding/user_embeds_var")
item_variables = ("embedding/item_linear_var", "embedding/item_embeds_var")
sparse_variables = ("embedding/sparse_linear_var", "embedding/sparse_embeds_var")
dense_variables = ("embedding/dense_linear_var", "embedding/dense_embeds_var")
def __init__(
self,
task,
data_info,
loss_type="cross_entropy",
embed_size=16,
n_epochs=20,
lr=0.001,
lr_decay=False,
epsilon=1e-5,
reg=None,
batch_size=256,
sampler="random",
num_neg=1,
use_bn=True,
dropout_rate=None,
multi_sparse_combiner="sqrtn",
seed=42,
lower_upper_bound=None,
tf_sess_config=None,
):
super().__init__(task, data_info, lower_upper_bound, tf_sess_config)
self.all_args = locals()
self.loss_type = loss_type
self.embed_size = embed_size
self.n_epochs = n_epochs
self.lr = lr
self.lr_decay = lr_decay
self.epsilon = epsilon
self.reg = reg_config(reg)
self.batch_size = batch_size
self.sampler = sampler
self.num_neg = num_neg
self.use_bn = use_bn
self.dropout_rate = dropout_config(dropout_rate)
self.seed = seed
self.sparse = check_sparse_indices(data_info)
self.dense = check_dense_values(data_info)
if self.sparse:
self.sparse_feature_size = sparse_feat_size(data_info)
self.sparse_field_size = sparse_field_size(data_info)
self.multi_sparse_combiner = check_multi_sparse(
data_info, multi_sparse_combiner
)
self.true_sparse_field_size = true_sparse_field_size(
data_info, self.sparse_field_size, self.multi_sparse_combiner
)
if self.dense:
self.dense_field_size = dense_field_size(data_info)
def build_model(self):
tf.set_random_seed(self.seed)
self.labels = tf.placeholder(tf.float32, shape=[None])
self.is_training = tf.placeholder_with_default(False, shape=[])
self.linear_embed, self.pairwise_embed = [], []
self._build_user_item()
if self.sparse:
self._build_sparse()
if self.dense:
self._build_dense()
linear_embed = tf.concat(self.linear_embed, axis=1)
pairwise_embed = tf.concat(self.pairwise_embed, axis=1)
# B * 1
linear_term = tf_dense(units=1, activation=None)(linear_embed)
# B * K
pairwise_term = 0.5 * tf.subtract(
tf.square(tf.reduce_sum(pairwise_embed, axis=1)),
tf.reduce_sum(tf.square(pairwise_embed), axis=1),
)
# For original FM, just add K dim together:
# pairwise_term = 0.5 * tf.reduce_sum(pairwise_term, axis=1)
if self.use_bn:
pairwise_term = tf.layers.batch_normalization(
pairwise_term, training=self.is_training
)
pairwise_term = tf_dense(units=1, activation=tf.nn.elu)(pairwise_term)
self.output = tf.squeeze(tf.add(linear_term, pairwise_term))
self.serving_topk = self.build_topk(self.output)
count_params()
def _build_user_item(self):
self.user_indices = tf.placeholder(tf.int32, shape=[None])
self.item_indices = tf.placeholder(tf.int32, shape=[None])
linear_user_embeds = embedding_lookup(
indices=self.user_indices,
var_name="user_linear_var",
var_shape=(self.n_users + 1, 1),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
linear_item_embeds = embedding_lookup(
indices=self.item_indices,
var_name="item_linear_var",
var_shape=(self.n_items + 1, 1),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
user_embeds = embedding_lookup(
indices=self.user_indices,
var_name="user_embeds_var",
var_shape=(self.n_users + 1, self.embed_size),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
item_embeds = embedding_lookup(
indices=self.item_indices,
var_name="item_embeds_var",
var_shape=(self.n_items + 1, self.embed_size),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
self.linear_embed.extend([linear_user_embeds, linear_item_embeds])
self.pairwise_embed.extend(
[user_embeds[:, tf.newaxis, :], item_embeds[:, tf.newaxis, :]]
)
def _build_sparse(self):
self.sparse_indices = tf.placeholder(
tf.int32, shape=[None, self.sparse_field_size]
)
linear_sparse_embed = compute_sparse_feats(
self.data_info,
self.multi_sparse_combiner,
self.sparse_indices,
var_name="sparse_linear_var",
var_shape=[self.sparse_feature_size],
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
pairwise_sparse_embed = compute_sparse_feats(
self.data_info,
self.multi_sparse_combiner,
self.sparse_indices,
var_name="sparse_embeds_var",
var_shape=(self.sparse_feature_size, self.embed_size),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
self.linear_embed.append(linear_sparse_embed)
self.pairwise_embed.append(pairwise_sparse_embed)
def _build_dense(self):
self.dense_values = tf.placeholder(
tf.float32, shape=[None, self.dense_field_size]
)
linear_dense_embed = compute_dense_feats(
self.dense_values,
var_name="dense_linear_var",
var_shape=[self.dense_field_size],
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
pairwise_dense_embed = compute_dense_feats(
self.dense_values,
var_name="dense_embeds_var",
var_shape=(self.dense_field_size, self.embed_size),
initializer=tf.glorot_uniform_initializer(),
regularizer=self.reg,
)
self.linear_embed.append(linear_dense_embed)
self.pairwise_embed.append(pairwise_dense_embed)