Factorization Machine

Factorization Machine in tensorflow.
Compatible with python2 and python3.
Tested on:

• tensorflow v1.4.1 & v1.8.0
• CPU & GPU
• py2.7 & py3.6

Model Structure

Example

python train_example.py
python restore_example.py
tensorboard --logdir=tensorboard_log/

Functions

• FM Core
  train_step(sess, input_x, input_y, lr)
  eval_loss(sess, input_x, input_y)
  eval_metrics(sess, input_x, input_y)
  get_embedding(sess, input_x)
  saver(): saver for model only
  ckpt_saver(): saver for all the variables(including opt etc.)

• FM Classifier
  predict_proba(sess, input_x)
  eval_auc(sess, input_x, input_y)

• FM Regressor
  predict(sess, input_x)

Reference:

• Factorization Machine