ZSL-KG Experiments

ZSL-KG is a general-purpose zero-shot learning framework with a novel transformer graph convolutional network (TrGCN) to learn class representation from common sense knowledge graphs.

This is the codebase for all the experiments mentioned in Zero-shot Learning with Common Sense Knowledge graphs.

Code Organization

The code is organized by task, namely:

1. intent_classification
2. fine_grained_entity_typing
3. object_classification
4. graph_utils

Refer to the individual directories (intent_classification, fine_grained_entity_typing, and object_classification) to run the zero-shot experiments.

Querying the ConceptNetDB

In the graph_utils directory, we include example code to query and preprocess the 2-hop neighbourhood for the ImageNet classes from the conceptnetdb. While the preprocessing the graph is relatively simple, setting up the initial ConceptNet database with the official guide could be time-consuming. To easily reproduce our experiments, we're releasing the all the knowledge graph-related data for the experiments on google drive.

Setup

conda create --name zsl_kg python=3.7
conda activate zsl_kg
pip install -r zsl-kg-requirements.txt

You can go to the respective directory and run the experiments.

Data

We include all the knowledge graph-related data for the experiments on google drive.

Citation

Please cite the following paper if you are using our framework.

@article{nayak:tmlr22,
  Author = {Nayak, N. V. and Bach, S. H.},
  Title = {Zero-Shot Learning with Common Sense Knowledge Graphs},
  Journal = {Transactions on Machine Learning Research (TMLR)},
  Year = {2022}}