MathDQN

The research paper MathDQN: Solving Arithmetic Word Problems via Deep Reinforcement Learning has been accepted at the Thirty-Second AAAI Conference on Artificial Intelligence (AAAI-18).

---

Requirements (Original):

• Tensorflow 1.2.1
• Python 2.7

Run

To run the experiment with default parameters:

    python DQN.py 0(fold number)

Now, we only provide code for dataset AI2.(code for other two datasets will be integrated in this code soon)

Dataset

AI2 (Hosseini et. al., 2014)

• Learning to Solve Arithmetic Word Problems with Verb Categorization
• code (Java)