Neural Programmer: Inducing Latent Programs with Gradient Descent

TL;DR

Neural Programmer introduces a differentiable neural network that learns to induce complex programs with basic arithmetic and logic operations, enabling better reasoning in tasks like table comprehension.

Contribution

The paper presents Neural Programmer, a novel end-to-end differentiable model that learns to induce compositional programs using weak supervision and gradient descent.

Findings

Neural Programmer achieves near-perfect accuracy on complex table comprehension tasks.

Adding gradient noise significantly improves training stability.

Traditional models perform poorly compared to Neural Programmer on the dataset.

Abstract

Deep neural networks have achieved impressive supervised classification performance in many tasks including image recognition, speech recognition, and sequence to sequence learning. However, this success has not been translated to applications like question answering that may involve complex arithmetic and logic reasoning. A major limitation of these models is in their inability to learn even simple arithmetic and logic operations. For example, it has been shown that neural networks fail to learn to add two binary numbers reliably. In this work, we propose Neural Programmer, an end-to-end differentiable neural network augmented with a small set of basic arithmetic and logic operations. Neural Programmer can call these augmented operations over several steps, thereby inducing compositional programs that are more complex than the built-in operations. The model learns from a weak supervision signal which is the result of execution of the correct program, hence it does not require expensive annotation of the correct program itself. The decisions of what operations to call, and what data segments to apply to are inferred by Neural Programmer. Such decisions, during training, are done in a differentiable fashion so that the entire network can be trained jointly by gradient descent. We find that training the model is difficult, but it can be greatly improved by adding random noise to the gradient. On a fairly complex synthetic table-comprehension dataset, traditional recurrent networks and attentional models perform poorly while Neural Programmer typically obtains nearly perfect accuracy.