Deepcode-Interpretability-Higher

Overview

This repository contains the codes for the paper "Higher-order Interpretations of DeepCode, a Learned Feedback Codes". It is a non-linear interpretable model with higher-order error correction for AWGN channel with feedback.

Structure

• deepcode.py
  Implementation of Deepcode using PyTorch, based on TensorFlow Deepcode.

• encx_deepdec5.py
  Ablation study on different encoders with a Deepcode decoder.

• enc3decxsingle.py
  Encoder with third-order error correction.
  Decoder that considers future x-1 parity bits.
  Single-stage decoder.

• enc3decxtwofix.py
  Encoder with third-order error correction.
  Decoder that considers future x-1 parity bits.
  Two-stage decoder with fixed knee points (noiseless feedback).

• enc3decxtwovary.py
  Encoder with third-order error correction.
  Decoder that considers future x-1 parity bits.
  Two-stage decoder with varying knee points (noisy feedback).

The logs folder contains the parameters for all models.

original TensorFlow Deepcode: https://github.com/hyejikim1/Deepcode

previous interpretable model: https://github.com/zyy-cc/Deepcode-Interpretability