def __init__(self, data, config, params, *args, **kwargs):
self._random = np.random
self._params_list = [
("_learning_rate", "lr", "lr", 0.0001, None, None),
("_user_mlp", "user_mlp", "umlp", "(64,32)",
lambda x: list(make_tuple(str(x))),
lambda x: self._batch_remove(str(x), " []").replace(",", "-")),
("_item_mlp", "item_mlp", "imlp", "(64,32)",
lambda x: list(make_tuple(str(x))),
lambda x: self._batch_remove(str(x), " []").replace(",", "-")),
("_neg_ratio", "neg_ratio", "negratio", 5, None, None),
("_reg", "reg", "reg", 0.001, None, None),
("_similarity", "similarity", "sim", "cosine", None, None)
]
self.autoset_params()
self._max_ratings = np.max(self._data.sp_i_train_ratings)
self._transactions_per_epoch = self._data.transactions + self._neg_ratio * self._data.transactions
if self._batch_size < 1:
self._batch_size = self._data.transactions + self._neg_ratio * self._data.transactions
self._sampler = pws.Sampler(self._data.i_train_dict,
self._data.sp_i_train_ratings,
self._neg_ratio)
self._ratings = self._data.train_dict
self._sp_i_train = self._data.sp_i_train
self._i_items_set = list(range(self._num_items))
self._model = DeepMatrixFactorizationModel(
self._num_users, self._num_items, self._user_mlp, self._item_mlp,
self._reg, self._similarity, self._max_ratings,
self._data.sp_i_train_ratings, self._learning_rate)
def __init__(self, data, config, params, *args, **kwargs):
"""
Create a NAIS instance.
(see https://arxiv.org/pdf/1809.07053.pdf for details about the algorithm design choices).
"""
self._random = np.random
self._params_list = [
("_factors", "factors", "factors", 100, None, None),
("_algorithm", "algorithm", "algorithm", "concat", None, None),
("_weight_size", "weight_size", "weight_size", 32, None, None),
("_lr", "lr", "lr", 0.001, None, None),
("_l_w", "l_w", "l_w", 0.001, None, None),
("_l_b", "l_b", "l_b", 0.001, None, None),
("_alpha", "alpha", "alpha", 0.5, lambda x: min(max(0, x), 1),
None), ("_beta", "beta", "beta", 0.5, None, None),
("_neg_ratio", "neg_ratio", "neg_ratio", 0.5, None, None)
]
self.autoset_params()
if self._batch_size < 1:
self._batch_size = self._data.transactions
self._ratings = self._data.train_dict
self._sampler = pws.Sampler(self._data.i_train_dict,
self._data.sp_i_train_ratings,
self._neg_ratio)
self._model = NAIS_model(self._data, self._algorithm,
self._weight_size, self._factors, self._lr,
self._l_w, self._l_b, self._alpha, self._beta,
self._num_users, self._num_items)
def __init__(self, data, config, params, *args, **kwargs):
"""
Create a FISM instance.
(see http://glaros.dtc.umn.edu/gkhome/node/1068 for details about the algorithm design choices).
Args:
data: data loader object
params: model parameters {_factors: embedding size,
[l_w, l_b]: regularization,
lr: learning rate}
"""
self._random = np.random
self._params_list = [
("_factors", "factors", "factors", 100, None, None),
("_lr", "lr", "lr", 0.001, None, None),
("_l_w", "l_w", "l_w", 0.001, None, None),
("_l_b", "l_b", "l_b", 0.001, None, None),
("_alpha", "alpha", "alpha", 0.5, lambda x: min(max(0, x), 1),
None),
("_neg_ratio", "neg_ratio", "neg_ratio", 0.5, None, None),
]
self.autoset_params()
if self._batch_size < 1:
self._batch_size = self._data.transactions
self._ratings = self._data.train_dict
self._sampler = pws.Sampler(self._data.i_train_dict,
self._data.sp_i_train_ratings,
self._neg_ratio)
self._model = FISM_model(self._data, self._factors, self._lr,
self._l_w, self._l_b, self._alpha,
self._num_users, self._num_items)