This project is an undergraduate individual graduation project in computer science, focusing on comparing Inductive Logic Programming (ILP) and Graph Neural Networks (GNN) in solving logical problems. The primary dataset used is CLEVR, which comprises images generated via Blender along with corresponding questions that pose logical queries about objects in the images. The goal is to explore and analyze the effectiveness of ILP and GNN in understanding and answering these logical queries.
- Neural Networks: PyTorch
- Inductive Logic Programming: SWI-Prolog
The project is currently in the technology verification stage. The primary aim at this stage is to separate the work on Computer Vision from ILP and GNN components. This separation ensures that ILP or GNN can function as independent modules without necessarily having their own Computer Vision capabilities.
(Currently not applicable as the project is in the early stages of development.)
(Usage instructions will be provided later as the project progresses.)
This project is licensed under the Apache License.
(To be updated)