TASK	DATASET	MODEL	METRIC NAME	METRIC VALUE	GLOBAL RANK
Image Clustering	CIFAR-10	CoHiClust	Accuracy	0.839	# 14
Image Clustering	CIFAR-10	CoHiClust	NMI	0.779	# 10
Image Clustering	CIFAR-10	CoHiClust	Train set	Train	# 1
Image Clustering	CIFAR-10	CoHiClust	ARI	0.731	# 12
Image Clustering	CIFAR-10	CoHiClust	Backbone	ResNet-50	# 1
Image Clustering	CIFAR-100	CoHiClust	Accuracy	0.437	# 13
Image Clustering	CIFAR-100	CoHiClust	NMI	0.467	# 10
Image Clustering	CIFAR-100	CoHiClust	ARI	0.299	# 11
Image Clustering	Fashion-MNIST	CoHiClust	Accuracy	0.65	# 5
Image Clustering	ImageNet-10	CoHiClust	Accuracy	0.953	# 6
Image Clustering	ImageNet-10	CoHiClust	NMI	0.907	# 3
Image Clustering	ImageNet-10	CoHiClust	ARI	0.899	# 6
Image Clustering	ImageNet-10	CoHiClust	Backbone	ResNet-50	# 1
Image Clustering	Imagenet-dog-15	CoHiClust	Accuracy	0.355	# 13
Image Clustering	Imagenet-dog-15	CoHiClust	NMI	0.411	# 12
Image Clustering	Imagenet-dog-15	CoHiClust	ARI	0.232	# 12
Image Clustering	Imagenet-dog-15	CoHiClust	Backbone	ResNet-50	# 1
Image Clustering	MNIST	CoHiClust	Accuracy	0.99	# 1
Image Clustering	STL-10	CoHiClust	Accuracy	0.613	# 17
Image Clustering	STL-10	CoHiClust	NMI	0.584	# 14
Image Clustering	STL-10	CoHiClust	ARI	0.474	# 7
Image Clustering	STL-10	CoHiClust	Backbone	ResNet-50	# 1

Badge	Markdown
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-mnist)](https://paperswithcode.com/sota/image-clustering-on-mnist?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-imagenet-10)](https://paperswithcode.com/sota/image-clustering-on-imagenet-10?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-fashion-mnist)](https://paperswithcode.com/sota/image-clustering-on-fashion-mnist?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-cifar-10)](https://paperswithcode.com/sota/image-clustering-on-cifar-10?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-cifar-100)](https://paperswithcode.com/sota/image-clustering-on-cifar-100?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-imagenet-dog-15)](https://paperswithcode.com/sota/image-clustering-on-imagenet-dog-15?p=contrastive-hierarchical-clustering)`
	`[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/contrastive-hierarchical-clustering/image-clustering-on-stl-10)](https://paperswithcode.com/sota/image-clustering-on-stl-10?p=contrastive-hierarchical-clustering)`

Contrastive Hierarchical Clustering

3 Mar 2023 · Michał Znaleźniak, Przemysław Rola, Patryk Kaszuba, Jacek Tabor, Marek Śmieja ·

Deep clustering has been dominated by flat models, which split a dataset into a predefined number of groups. Although recent methods achieve an extremely high similarity with the ground truth on popular benchmarks, the information contained in the flat partition is limited. In this paper, we introduce CoHiClust, a Contrastive Hierarchical Clustering model based on deep neural networks, which can be applied to typical image data. By employing a self-supervised learning approach, CoHiClust distills the base network into a binary tree without access to any labeled data. The hierarchical clustering structure can be used to analyze the relationship between clusters, as well as to measure the similarity between data points. Experiments demonstrate that CoHiClust generates a reasonable structure of clusters, which is consistent with our intuition and image semantics. Moreover, it obtains superior clustering accuracy on most of the image datasets compared to the state-of-the-art flat clustering models.

PDF Abstract

Code

Add Remove Mark official

michalznalezniak/contrastive-hierar… official

Tasks

Add Remove

Clustering

Deep Clustering

Image Clustering

Self-Supervised Learning

Datasets

CIFAR-10

CIFAR-100

MNIST

Fashion-MNIST

STL-10

Results from the Paper

Edit

Ranked #1 on Image Clustering on MNIST

Get a GitHub badge

Task	Dataset	Model	Metric Name	Metric Value	Global Rank	Benchmark
Image Clustering	CIFAR-10	CoHiClust	Accuracy	0.839	# 14	Compare
			NMI	0.779	# 10	Compare
			Train set	Train	# 1	Compare
			ARI	0.731	# 12	Compare
			Backbone	ResNet-50	# 1	Compare
Image Clustering	CIFAR-100	CoHiClust	Accuracy	0.437	# 13	Compare
			NMI	0.467	# 10	Compare
			ARI	0.299	# 11	Compare
Image Clustering	Fashion-MNIST	CoHiClust	Accuracy	0.65	# 5	Compare
Image Clustering	ImageNet-10	CoHiClust	Accuracy	0.953	# 6	Compare
			NMI	0.907	# 3	Compare
			ARI	0.899	# 6	Compare
			Backbone	ResNet-50	# 1	Compare
Image Clustering	Imagenet-dog-15	CoHiClust	Accuracy	0.355	# 13	Compare
			NMI	0.411	# 12	Compare
			ARI	0.232	# 12	Compare
			Backbone	ResNet-50	# 1	Compare
Image Clustering	MNIST	CoHiClust	Accuracy	0.99	# 1	Compare
Image Clustering	STL-10	CoHiClust	Accuracy	0.613	# 17	Compare
			NMI	0.584	# 14	Compare
			ARI	0.474	# 7	Compare
			Backbone	ResNet-50	# 1	Compare

Methods

Add Remove

1x1 Convolution • Average Pooling • BASE • Batch Normalization • Bottleneck Residual Block • Convolution • Global Average Pooling • Kaiming Initialization • Max Pooling • ReLU • Residual Block • Residual Connection • ResNet

Edit Social Preview

Contrastive Hierarchical Clustering

Code Edit Add Remove Mark official

Tasks Edit Add Remove

Datasets Edit

Results from the Paper Edit

Methods Edit Add Remove

Code

Add Remove Mark official

Tasks

Add Remove

Datasets

Results from the Paper

Edit

Methods

Add Remove