deepknowledgetracing

Tensorflow implementation for the Deep Knowledge Tracing

To use the original DKT model. You could use the following command:

python main.py

---

This project enable the extensability of using the DKT. It provides:

• Specify the hidden layer structure by using -hl, e.g. -hl 200 50 implies using 200 units in the first layer and 50 units in the second layer.
• Specify whether to use gaussian noise on the input data by flating -gn.
• Specify the RNN Cell that to used in the network by -cell {LSTM,GRU,BasicRNN,LayerNormBasicLSTM}
• For more, please look into the below section of program usage.

Program usage

---

usage: main.py [-h] [--num_runs NUM_RUNS] [--num_epochs NUM_EPOCHS]
               [--batch_size BATCH_SIZE]
               [-hl [HIDDEN_LAYER_STRUCTURE [HIDDEN_LAYER_STRUCTURE ...]]]
               [-gn] [-cell {LSTM,GRU,BasicRNN,LayerNormBasicLSTM}]
               [-lr LEARNING_RATE] [--keep_prob KEEP_PROB]
               [--ckpt_dir CKPT_DIR] [--model_name MODEL_NAME]
               [--data_dir DATA_DIR] [--train_file TRAIN_FILE]
               [--test_file TEST_FILE]

optional arguments:
  -h, --help            show this help message and exit
  --num_runs NUM_RUNS   Number of runs to repeat the experiment.
  --num_epochs NUM_EPOCHS
                        Maximum number of epochs to train the network.
  --batch_size BATCH_SIZE
                        The mini-batch size used when training the network.
  -hl [HIDDEN_LAYER_STRUCTURE [HIDDEN_LAYER_STRUCTURE ...]], --hidden_layer_structure [HIDDEN_LAYER_STRUCTURE [HIDDEN_LAYER_
STRUCTURE ...]]
                        The hidden layer structure in the RNN. If there is 2
                        hidden layers with first layer of 200 and second layer
                        of 50. Type in '-hl 200 50'
  -gn, --use_gaussian_noise
                        Flag this to add gaussian noise to the input.
  -cell {LSTM,GRU,BasicRNN,LayerNormBasicLSTM}, --rnn_cell {LSTM,GRU,BasicRNN,LayerNormBasicLSTM}
                        Specify the rnn cell used in the graph.
  -lr LEARNING_RATE, --learning_rate LEARNING_RATE
                        The learning rate when training the model.
  --keep_prob KEEP_PROB
                        Keep probability when training the network.
  --ckpt_dir CKPT_DIR   The base directory that the model parameter going to
                        store.
  --model_name MODEL_NAME
                        The directory that the model parameter going to store.
  --data_dir DATA_DIR   the data directory, default as './data/
  --train_file TRAIN_FILE
                        train data file, default as 'skill_id_train.csv'.
  --test_file TEST_FILE
                        train data file, default as 'skill_id_test.csv'.

Program Run Time

---

The model is trained with GTX1080 with skill\_id\_train.csv, and evaluated over skill\_id\_test.csv.

Model	Time used per epoch (approx.)	Average AUC
O200	90s	0.82144
O184-50	165s	0.82071
G200	92s	0.82506
G-184-50	165s	0.82308
O200_with_layernorm	360s	0.81542