Neighborhood Contrastive Learning for Novel Class Discovery

Last update: Dec 9, 2022

Related tags

Overview

Neighborhood Contrastive Learning for Novel Class Discovery

This repository contains the official implementation of our paper:

Neighborhood Contrastive Learning for Novel Class Discovery, CVPR 2021
Zhun Zhong, Enrico Fini, Subhankar Roy, Zhiming Luo, Elisa Ricci, Nicu Sebe

Requirements

PyTorch >= 1.1

Data preparation

We follow AutoNovel to prepare the data

By default, we save the dataset in ./data/datasets/ and trained models in ./data/experiments/.

For CIFAR-10 and CIFAR-100, the datasets can be automatically downloaded by PyTorch.
For ImageNet, we use the exact split files used in the experiments following existing work. To download the split files, run the command: sh scripts/download_imagenet_splits.sh . The ImageNet dataset folder is organized in the following way:
```
ImageNet/imagenet_rand118 #downloaded by the above command
ImageNet/images/train #standard ImageNet training split
ImageNet/images/val #standard ImageNet validation split
```

Pretrained models

We use the pretrained models (self-supervised learning and supervised learning) provided by AutoNovel. To download, run:

sh scripts/download_pretrained_models.sh

If you would like to train the self-supervised learning and supervised learning models by yourself, please refer to AutoNovel for more details.

After downloading, you can go to perform our neighbor contrastive learning below.

Neighborhood Contrastive Learning for Novel Class Discovery

CIFAR10/CIFAR100

Without Hard Negative Generation (w/o HNG)

# Train on CIFAR10
CUDA_VISIBLE_DEVICES=0 sh scripts/ncl_cifar10.sh ./data/datasets/CIFAR/ ./data/experiments/ ./data/experiments/pretrained/supervised_learning/resnet_rotnet_cifar10.pth

# Train on CIFAR100
CUDA_VISIBLE_DEVICES=0 sh scripts/ncl_cifar100.sh ./data/datasets/CIFAR/ ./data/experiments/ ./data/experiments/pretrained/supervised_learning/resnet_rotnet_cifar100.pth

With Hard Negative Generation (w/ HNG)

# Train on CIFAR10
CUDA_VISIBLE_DEVICES=0 sh scripts/ncl_hng_cifar10.sh ./data/datasets/CIFAR/ ./data/experiments/ ./data/experiments/pretrained/supervised_learning/resnet_rotnet_cifar10.pth

# Train on CIFAR100
CUDA_VISIBLE_DEVICES=0 sh scripts/ncl_hng_cifar100.sh ./data/datasets/CIFAR/ ./data/experiments/ ./data/experiments/pretrained/supervised_learning/resnet_rotnet_cifar100.pth

Note that, for cifar-10, we suggest to train the model w/o HNG, because the results of w HNG and w/o HNG for cifar-10 are similar. In addition, the model w/ HNG sometimes will collapse, but you can try different seeds to get the normal result.

ImageNet

Without Hard Negative Generation (w/o HNG)

# Subset A
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --unlabeled_subset A --model_name resnet_imagenet_ncl

# Subset B
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --unlabeled_subset B --model_name resnet_imagenet_ncl

# Subset C
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --unlabeled_subset C --model_name resnet_imagenet_ncl

With Hard Negative Generation (w/o HNG)

# Subset A
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --hard_negative_start 3 --unlabeled_subset A --model_name resnet_imagenet_ncl_hng

# Subset B
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --hard_negative_start 3 --unlabeled_subset B --model_name resnet_imagenet_ncl_hng

# Subset C
CUDA_VISIBLE_DEVICES=0 python ncl_imagenet.py --hard_negative_start 3 --unlabeled_subset C --model_name resnet_imagenet_ncl_hng

Acknowledgement

Our code is heavily designed based on AutoNovel. If you use this code, please also acknowledge their paper.

Citation

We hope you find our work useful. If you would like to acknowledge it in your project, please use the following citation:

@InProceedings{Zhong_2021_CVPR,
      author    = {Zhong, Zhun and Fini, Enrico and Roy, Subhankar and Luo, Zhiming and Ricci, Elisa and Sebe, Nicu},
      title     = {Neighborhood Contrastive Learning for Novel Class Discovery},
      booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
      month     = {June},
      year      = {2021},
      pages     = {10867-10875}
}

Contact me

If you have any questions about this code, please do not hesitate to contact me.

Zhun Zhong

Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction".

GNN_PPI Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction". Lear

2 Dec 14, 2022

Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"

Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis Abstract: This work targets at using a general deep lea

163 Dec 14, 2022

A python implementation of Physics-informed Spline Learning for nonlinear dynamics discovery

PiSL A python implementation of Physics-informed Spline Learning for nonlinear dynamics discovery. Sun, F., Liu, Y. and Sun, H., 2021. Physics-informe

8 Jul 13, 2022

This is the implementation of "SELF SUPERVISED REPRESENTATION LEARNING WITH DEEP CLUSTERING FOR ACOUSTIC UNIT DISCOVERY FROM RAW SPEECH" submitted to ICASSP 2022

CPC_DeepCluster This is the implementation of "SELF SUPERVISED REPRESENTATION LEARNING WITH DEEP CLUSTERING FOR ACOUSTIC UNIT DISCOVERY FROM RAW SPEEC

2 Sep 15, 2022

Pytorch implementation of paper "Learning Co-segmentation by Segment Swapping for Retrieval and Discovery"

SegSwap Pytorch implementation of paper "Learning Co-segmentation by Segment Swapping for Retrieval and Discovery" [PDF] [Project page] If our project

41 Dec 10, 2022

GeneDisco is a benchmark suite for evaluating active learning algorithms for experimental design in drug discovery.

22 Dec 12, 2022

Open source repository for the code accompanying the paper 'Non-Rigid Neural Radiance Fields Reconstruction and Novel View Synthesis of a Deforming Scene from Monocular Video'.

Non-Rigid Neural Radiance Fields This is the official repository for the project "Non-Rigid Neural Radiance Fields: Reconstruction and Novel View Synt

296 Dec 29, 2022

Official PyTorch implementation for Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers, a novel method to visualize any Transformer-based network. Including examples for DETR, VQA.

PyTorch Implementation of Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers 1 Using Colab Please notic

489 Jan 7, 2023

A novel method to tune language models. Codes and datasets for paper ``GPT understands, too''.

P-tuning A novel method to tune language models. Codes and datasets for paper ``GPT understands, too''. How to use our code We have released the code

562 Dec 27, 2022

Comments

Reproducing the results on CIFAR100

Dear authors,

Thank you for your great work and clean code.

I was trying to reproduce the results from the paper (w/ HNG) on CIFAR10 and CIFAR100 datasets, however, I achieved significantly lower results on CIFAR100:

| CIFAR10 | CIFAR100 | |--|--| | 93.4±0.2% | 81.6±0.8% |

CIFAR10 results match the paper's nearly exactly, while the CIFAR100 scores reported in the paper are 86.6±0.4% - 5% higher.

To obtain the results, I ran the exact commands (w/ HNG) from the ReadMe 3 times. The training on CIFAR100 was pretty stable, but I picked the peak results anyway.

Just for reference, the final command that the ncl_hng_cifar100.sh gets executed is: ncl_cifar.py --dataset_root ./data/datasets/CIFAR/ --exp_root ./experiments --warmup_model_dir ./data/experiments/pretrained/supervised_learning/resnet_rotnet_cifar100.pth --lr 0.1 --gamma 0.1 --weight_decay 1e-4 --step_size 170 --batch_size 128 --epochs 200 --rampup_length 150 --rampup_coefficient 50 --num_labeled_classes 80 --num_unlabeled_classes 20 --dataset_name cifar100 --seed 3 --model_name resnet_cifar100_ncl_hng --mode train --hard_negative_start 3 --bce_type cos

Could you please check that the script provided for CIFAR100 is correct? E.g. I noticed that the utils/ramps.py was missing from the RankStats repository, so maybe the pushed version misses some final changes.

opened by vlfom 5
CVE-2007-4559 Patch

Patching CVE-2007-4559

Hi, we are security researchers from the Advanced Research Center at Trellix. We have began a campaign to patch a widespread bug named CVE-2007-4559. CVE-2007-4559 is a 15 year old bug in the Python tarfile package. By using extract() or extractall() on a tarfile object without sanitizing input, a maliciously crafted .tar file could perform a directory path traversal attack. We found at least one unsantized extractall() in your codebase and are providing a patch for you via pull request. The patch essentially checks to see if all tarfile members will be extracted safely and throws an exception otherwise. We encourage you to use this patch or your own solution to secure against CVE-2007-4559. Further technical information about the vulnerability can be found in this blog.

If you have further questions you may contact us through this projects lead researcher Kasimir Schulz.

opened by TrellixVulnTeam 0

Neighborhood Contrastive Learning for Novel Class Discovery

Related tags

Overview

Neighborhood Contrastive Learning for Novel Class Discovery

Requirements

Data preparation

Pretrained models

Neighborhood Contrastive Learning for Novel Class Discovery

CIFAR10/CIFAR100

Without Hard Negative Generation (w/o HNG)

With Hard Negative Generation (w/ HNG)

ImageNet

Without Hard Negative Generation (w/o HNG)

With Hard Negative Generation (w/o HNG)

Acknowledgement

Citation

Contact me

You might also like...

Codes and models for the paper "Learning Unknown from Correlations: Graph Neural Network for Inter-novel-protein Interaction Prediction".

Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"

A python implementation of Physics-informed Spline Learning for nonlinear dynamics discovery

This is the implementation of "SELF SUPERVISED REPRESENTATION LEARNING WITH DEEP CLUSTERING FOR ACOUSTIC UNIT DISCOVERY FROM RAW SPEECH" submitted to ICASSP 2022

Pytorch implementation of paper "Learning Co-segmentation by Segment Swapping for Retrieval and Discovery"

GeneDisco is a benchmark suite for evaluating active learning algorithms for experimental design in drug discovery.

Open source repository for the code accompanying the paper 'Non-Rigid Neural Radiance Fields Reconstruction and Novel View Synthesis of a Deforming Scene from Monocular Video'.

Official PyTorch implementation for Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers, a novel method to visualize any Transformer-based network. Including examples for DETR, VQA.

A novel method to tune language models. Codes and datasets for paper ``GPT understands, too''.

Comments

Reproducing the results on CIFAR100

CVE-2007-4559 Patch

Patching CVE-2007-4559

Owner

Zhun Zhong

TraND: Transferable Neighborhood Discovery for Unsupervised Cross-domain Gait Recognition.

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.

Code for KDD'20 "An Efficient Neighborhood-based Interaction Model for Recommendation on Heterogeneous Graph"

An implementation of "MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing" (ICML 2019).

Code for our NeurIPS 2021 paper 'Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation'

Contrastive learning of Class-agnostic Activation Map for Weakly Supervised Object Localization and Semantic Segmentation (CVPR 2022)

SUPERVISED-CONTRASTIVE-LEARNING-FOR-PRE-TRAINED-LANGUAGE-MODEL-FINE-TUNING - The Facebook paper about fine tuning RoBERTa with contrastive loss

Re-implementation of the Noise Contrastive Estimation algorithm for pyTorch, following "Noise-contrastive estimation: A new estimation principle for unnormalized statistical models." (Gutmann and Hyvarinen, AISTATS 2010)

Stereo Radiance Fields (SRF): Learning View Synthesis for Sparse Views of Novel Scenes

PyToch implementation of A Novel Self-supervised Learning Task Designed for Anomaly Segmentation