gfun_multimodal/gfun/vgfs/vanillaFun.py

73 lines
2.4 KiB
Python

from gfun.vgfs.viewGen import ViewGen
from gfun.vgfs.learners.svms import NaivePolylingualClassifier
from gfun.vgfs.commons import _normalize
class VanillaFunGen(ViewGen):
"""
View Generator (x): original funnelling architecture proposed by Moreo, Esuli and
Sebastiani in DOI: https://doi.org/10.1145/3326065
"""
def __init__(self, base_learner, n_jobs=-1):
"""
Init Posterior Probabilities embedder (i.e., VanillaFunGen)
:param base_learner: naive monolingual learners to be deployed as first-tier
learners. Should be able to return posterior probabilities.
:param base_learner:
:param n_jobs: integer, number of concurrent workers
"""
print("- init VanillaFun View Generating Function")
self.learners = base_learner
self.n_jobs = n_jobs
self.doc_projector = NaivePolylingualClassifier(
base_learner=self.learners,
n_jobs=self.n_jobs,
)
self.vectorizer = None
self.load_trained = False
def fit(self, lX, lY):
if self.load_trained:
return self.load_trained()
print("- fitting VanillaFun View Generating Function")
lX = self.vectorizer.transform(lX)
self.doc_projector.fit(lX, lY)
return self
def transform(self, lX):
"""
(1) Vectorize documents;
(2) Project them according to the learners SVMs;
(3) Apply L2 normalization to the projection and returns it.
:param lX: dict {lang: indexed documents}
:return: document projection to the common latent space.
"""
lX = self.vectorizer.transform(lX)
lZ = self.doc_projector.predict_proba(lX)
lZ = _normalize(lZ, l2=True)
return lZ
def fit_transform(self, lX, lY):
return self.fit(lX, lY).transform(lX)
def save_vgf(self, model_id):
import pickle
from os.path import join
from os import makedirs
vgf_name = "vanillaFunGen"
_basedir = join("models", "vgfs", "posterior")
makedirs(_basedir, exist_ok=True)
_path = join(_basedir, f"{vgf_name}_{model_id}.pkl")
with open(_path, "wb") as f:
pickle.dump(self, f)
return self
def __str__(self):
_str = f"[VanillaFunGen (-p)]\n- base learner: {self.learners}\n- n_jobs: {self.n_jobs}\n"
# - parameters: {self.first_tier_parameters}
return _str