|
trigger_out = self._build_trigger_net() |
|
sim_out = self._build_similarity_net() |
|
logits = tf.multiply(sim_out, trigger_out) |
代码中实现的是:
trigger_out = tf.exp(<dnn output>)
sim_out = tf.exp(<dnn output>)
logits = tf.multiplay(sim_out, trigger_out)
相当于是:
$$ PATH_{uji} = e^{t_{uj}} e^{s_{ji}} $$
论文中的合并公式(公式(10))是:
$$
PATH_{uji} = MEG(t_{uj}, s_{ji}) = \ln(1 + e^{t_{uj}} e^{s_{ji}})
$$
EasyRec/easy_rec/python/model/pdn.py
Lines 43 to 45 in d965fb7
代码中实现的是:
相当于是:
论文中的合并公式(公式(10))是: