Post-hoc Concept Bottleneck Models (ICLR 2023 Spotlight)

The code for the paper Post-hoc Concept Bottleneck Models. This paper got an ICLR2023 Spotlight(notable-top-25%)!

This is the joint work of Mert Yuksekgonul, Maggie Wang, and James Zou.

Update: Reproducibility Study. A recent study titled "On the Reproducibility of Post-Hoc Concept Bottleneck Models" recently got accepted to TMLR. It is a very detailed study, and I recommend that you check out their work and GitHub repository! Thank you Nesta, Gregory, Diego, Ioana, and Satchit!!

Here is an overview of our work, and you can find more in our Preprint (soon to be updated).

Data Instructions

Below we list the data sources to download the datasets we used to train / evaluate PCBMs. Once you download a dataset, please go to data/constants.py and specify the corresponding paths where you downloaded the datasets.

Downloading the Data

Dataset	Description	URL
CUB	Bird Classification dataset	Codalab Bundle from CBM repo
CUB	CUB Metadata and splits	Codalab Bundle from CBM repo
Derm7pt	Dermatology Concepts Dataset	Get access here
HAM10k	Skin lesion classification dataset	Kaggle Link
CIFAR10, CIFAR100	Standard CIFAR datasets	These are automatically downloaded via torchvision.
Broden Concept Dataset	This dataset is mostly inherited from the Broden Dataset, thanks to Amirata Ghorbani and Abubakar Abid.	Can be downloaded from this gdrive folder.
SkinCon Concept Dataset	TODO will add. We later proposed this dataset for dermatology concepts.	Please see our NeurIPS22 paper and the dataset website.

Downloading the backbones

Please see models/model_zoo.py for the backbones we used. Some models rely on external dependencies (e.g. pytorchcv for the CUB backbone, OpenAI repo for the CLIP backbone.) or will be downloaded (e.g. HAM1000 model from the DDI repo). If you want to add your own model, please edit models/model_zoo.py.

Learning Concepts

In our paper, we propose two different ways to learn concepts activations vectors to build concepts banks.

1- Learning concepts with a concept dataset

To learn concepts in this way, we need to have a concept dataset where each concept should have a set of positive and negative images. We use the CAV methodology (Kim et al. 2018).

Concept Dataset Implementations: We provide the code to extract concept data loaders in data/concept_loaders.py. In there, you could find the loaders for Broden, CUB, and derm7pt datasets to extract concept loaders. If you'd like to use custom concept datasets, you could implement your own loader and place there.

Obtainin concept vectors: Once you have the concept data loaders, you could use the learn_concepts_dataset.py script to learn the concept vectors. As examples, you can run the following scripts (once you obtain the corresponding datasets):

OUTPUT_DIR=/path/where/you/save/conceptbanks/
# Learning CUB Concepts
python3 learn_concepts_dataset.py --dataset-name="cub" --backbone-name="resnet18_cub" --C 0.001 0.01 0.1 1.0 10.0 --n-samples=100 --out-dir=$OUTPUT_DIR

# Learning Derm7pt Concepts
python3 learn_concepts_dataset.py --dataset-name="derm7pt" --backbone-name="ham10000_inception" --C 0.001 0.01 0.1 1.0 10.0 --n-samples=50 --out-dir=$OUTPUT_DIR

# Learning Broden Concepts
python3 learn_concepts_dataset.py --dataset-name="broden" --backbone-name="clip:RN50" --C 0.001 0.01 0.1 1.0 10.0 --n-samples=50 --out-dir=$OUTPUT_DIR

Limitation:

1. This approach relies on the existence of a concept dataset. These may be hard to get, depending on the application.
2. Learning concepts with the CAV way could inherit the potential biases in the concept datasets. One should be careful about how the concept dataset is constructed, and what it means to learn that concept.

2- Learning concepts via prompting with multimodal models

What if we do not have a concept dataset? We could leverage multimodal models, such as CLIP! i.e. We can simply prompt the text encoder with the concept name, and obtain the concept vector in the shared embedding space.

We provide the code to do this in learn_concepts_multimodal.py. You can run the following script to learn the concept vectors:

python3 learn_concepts_multimodal.py --backbone-name="clip:RN50" --classes=cifar10 --out-dir=$OUTPUT_DIR --recurse=1

Currently, we support CIFAR10/CIFAR100 for this approach. You can very easily add the set of class names in the script and obtain the concept bank for your own purpose.

Limitation: This approach is limited to the multimodal models that have a shared embedding space. Existing multimodal models that are not specialized may not do very well with domain-specific concepts (e.g. healthcare concepts).

Training PCBMs

Once you have a concept bank and a backbone, you are ready to train your PCBM! We provide the code to train PCBMs in train_pcmb.py. You can run the following script to train a PCBM on CUB:

python3 train_pcbm.py --concept-bank="${OUTPUT_DIR}/cub_resnet18_cub_0.1_100.pkl" --dataset="cub" --backbone-name="resnet18_cub" --out-dir=$OUTPUT_DIR --lam=2e-4

Please see the train_pcbm.py file for the arguments / where the models are saved.

Limitation: There is a tradeoff between the regularization and how sparse/"interpretable" (yes, hard to define what exactly this means) the linear module is. This hyperparameter selection can be a bit tedious. We can play around with the lam parameter and alpha parameter to observe the concept coefficients and understand what seems like a good tradeoff. Good thing is, we can simply monitor concept weights, and since concepts are more meaningful, we may have a better say here.

Training PCBM-h

Once you have the PCBM, you can train the PCBM-h model by running the following script:

pcbm_path="/path/to/pcbm_cub__resnet18_cub__cub_resnet18_cub_0__lam:0.0002__alpha:0.99__seed:42.ckpt"
python3 train_pcbm_h.py --concept-bank="${OUTPUT_DIR}/cub_resnet18_cub_0.1_100.pkl" --pcbm-path=$pcbm_path --out-dir=$OUTPUT_DIR --dataset="cub"

Notes

This repository is WIP! The quickest way to reach out to me is via merty@stanford.edu.

Citation

If you find this code useful, please consider citing our paper:

@inproceedings{
yuksekgonul2023posthoc,
title={Post-hoc Concept Bottleneck Models},
author={Mert Yuksekgonul and Maggie Wang and James Zou},
booktitle={The Eleventh International Conference on Learning Representations },
year={2023},
url={https://openreview.net/forum?id=nA5AZ8CEyow}
}