search

zhougr1993 / DeepInterestNetwork

1.62k stars 558 forks source link

Questions about Wide and deep model #45

Closed czisok closed 4 years ago

czisok commented 5 years ago

After reading your paper, I have a few questions about wide & deep model,wish to communicate wiht you. # wide part d_layer_wide_i = tf.concat([tf.gather(u_emb, [0], axis=-1) * tf.gather(i_emb, [0], axis=-1), tf.gather(u_emb, [-1], axis=-1) * tf.gather(i_emb, [-1], axis=-1), tf.gather(u_emb, [hidden_units // 2], axis=-1) * tf.gather(i_emb, [hidden_units // 2], axis=-1)], axis=-1)

  1. Is this place counting cross features?
  2. In the original paper, didn't it say to make cross features of discrete features?
  3. Why select [0],[-1] [hiddens // 2]?
zhougr1993 commented 5 years ago

After reading your paper, I have a few questions about wide & deep model,wish to communicate wiht you. # wide part d_layer_wide_i = tf.concat([tf.gather(u_emb, [0], axis=-1) * tf.gather(i_emb, [0], axis=-1), tf.gather(u_emb, [-1], axis=-1) * tf.gather(i_emb, [-1], axis=-1), tf.gather(u_emb, [hidden_units // 2], axis=-1) * tf.gather(i_emb, [hidden_units // 2], axis=-1)], axis=-1)

  1. Is this place counting cross features?
  2. In the original paper, didn't it say to make cross features of discrete features?
  3. Why select [0],[-1] [hiddens // 2]?
  1. yes 2, 3. In amazon data, there are only two kind of feature category and item. Thus, we select [0],[-1] to mimic the feature crossing in our paper. Actually, wide part makes little difference.