"""
Module description:
"""
__version__ = '0.3.1'
__author__ = 'Vito Walter Anelli, Claudio Pomo, Daniele Malitesta, Felice Antonio Merra'
__email__ = 'vitowalter.anelli@poliba.it, claudio.pomo@poliba.it, daniele.malitesta@poliba.it, felice.merra@poliba.it'
from tqdm import tqdm
import tensorflow as tf
import numpy as np
from elliot.recommender.visual_recommenders.VBPR import pairwise_pipeline_sampler_vbpr as ppsv
from elliot.recommender import BaseRecommenderModel
from elliot.recommender.base_recommender_model import init_charger
from elliot.recommender.recommender_utils_mixin import RecMixin
from elliot.recommender.visual_recommenders.VBPR.VBPR_model import VBPRModel
[docs]class VBPR(RecMixin, BaseRecommenderModel):
r"""
VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback
For further details, please refer to the `paper <http://www.aaai.org/ocs/index.php/AAAI/AAAI16/paper/view/11914>`_
Args:
lr: Learning rate
epochs: Number of epochs
factors: Number of latent factors
factors_d: Dimension of visual factors
batch_size: Batch size
batch_eval: Batch for evaluation
l_w: Regularization coefficient
l_b: Regularization coefficient of bias
l_e: Regularization coefficient of projection matrix
To include the recommendation model, add it to the config file adopting the following pattern:
.. code:: yaml
models:
VBPR:
meta:
save_recs: True
lr: 0.0005
epochs: 50
factors: 100
factors_d: 20
batch_size: 128
batch_eval: 128
l_w: 0.000025
l_b: 0
l_e: 0.002
"""
@init_charger
def __init__(self, data, config, params, *args, **kwargs):
self._params_list = [
("_batch_eval", "batch_eval", "be", 512, int, None),
("_factors", "factors", "factors", 100, None, None),
("_factors_d", "factors_d", "factors_d", 20, None, None),
("_learning_rate", "lr", "lr", 0.0005, None, None),
("_l_w", "l_w", "l_w", 0.000025, None, None),
("_l_b", "l_b", "l_b", 0, None, None),
("_l_e", "l_e", "l_e", 0.002, None, None),
("_loader", "loader", "load", "VisualAttributes", None, None),
]
self.autoset_params()
if self._batch_size < 1:
self._batch_size = self._data.transactions
self._ratings = self._data.train_dict
self._side = getattr(self._data.side_information, self._loader, None)
item_indices = [self._side.item_mapping[self._data.private_items[item]] for item in range(self._num_items)]
self._sampler = ppsv.Sampler(self._data.i_train_dict,
item_indices,
self._side.visual_feature_folder_path,
self._epochs)
self._next_batch = self._sampler.pipeline(self._data.transactions, self._batch_size)
self._model = VBPRModel(self._factors,
self._factors_d,
self._learning_rate,
self._l_w,
self._l_b,
self._l_e,
self._side.visual_features_shape,
self._num_users,
self._num_items,
self._seed)
# only for evaluation purposes
self._next_eval_batch = self._sampler.pipeline_eval(self._batch_eval)
@property
def name(self):
return "VBPR" \
+ f"_{self.get_base_params_shortcut()}" \
+ f"_{self.get_params_shortcut()}"
[docs] def train(self):
if self._restore:
return self.restore_weights()
loss = 0
steps = 0
it = 0
with tqdm(total=int(self._data.transactions // self._batch_size), disable=not self._verbose) as t:
for batch in self._next_batch:
steps += 1
loss += self._model.train_step(batch)
t.set_postfix({'loss': f'{loss.numpy() / steps:.5f}'})
t.update()
if steps == self._data.transactions // self._batch_size:
t.reset()
self.evaluate(it, loss.numpy() / steps)
it += 1
steps = 0
loss = 0
[docs] def get_recommendations(self, k: int = 100):
predictions_top_k_test = {}
predictions_top_k_val = {}
for index, offset in enumerate(range(0, self._num_users, self._batch_eval)):
offset_stop = min(offset + self._batch_eval, self._num_users)
predictions = np.empty((offset_stop - offset, self._num_items))
for batch in self._next_eval_batch:
item_rel, item_abs, feat = batch
p = self._model.predict_item_batch(offset, offset_stop,
item_rel[0], item_rel[-1],
tf.Variable(feat))
predictions[:(offset_stop - offset), item_rel] = p
recs_val, recs_test = self.process_protocol(k, predictions, offset, offset_stop)
predictions_top_k_val.update(recs_val)
predictions_top_k_test.update(recs_test)
return predictions_top_k_val, predictions_top_k_test