https://github.com/hemile/probabilistic-real-logic
https://github.com/hemile/probabilistic-real-logic
Last synced: over 1 year ago
JSON representation
- Host: GitHub
- URL: https://github.com/hemile/probabilistic-real-logic
- Owner: HEmile
- Created: 2018-03-01T20:53:06.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2018-04-13T08:24:35.000Z (over 8 years ago)
- Last Synced: 2025-02-12T18:25:14.472Z (over 1 year ago)
- Language: Python
- Size: 13.1 MB
- Stars: 4
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# PROBABILISTIC REAL LOGIC
- This repository contains the implementation of PRL. The code is adapted from https://gitlab.fbk.eu/donadello/LTN_IJCAI17 which implemented the paper *Learning and Reasoning in Logic Tensor Networks: Theory and Application for Semantic Image Interpretation* by Serafini Luciano, Donadello Ivan, d'Avila Garcez Artur.
- This code implements the paper *Semi-supervised Classification using Differentiable Reasoning* by Emile van Krieken.
- The code is written using Python 3.6 and Tensorflow 1.3.0, though it should work with newer versions.
- This repository is missing the data. You can download it from https://www.dropbox.com/sh/502aq9u537lrmbv/AAAy00eEtQkIIK5Ytt3Bc2Iha/code/data?dl=0 and then extract it to the `code` folder.
## How to train
```sh
$ python train.py
```
- `code/config.py` contains the parameters and experimental setup, see that file for more information. Multiple models can be trained simultaneously.
- During training, summaries are written to the `code/logging` folder. From that folder, run ```tensorboard --logdir .``` to follow the progress.
- After training, the models are written to `code/models`.