Deep Subspace Clustering Networks

By Pan Ji, T. Zhang, Hongdong Li, Mathieu Salzmann, Ian David Reid. Neural Information Processing Systems, 2017

586 citations recorded by the source corpus.

We present a novel deep neural network architecture for unsupervised subspace clustering. This architecture is built upon deep auto-encoders, which non-linearly map the input data into a latent space. Our key idea is to introduce a novel self-expressive layer between the encoder and the decoder to mimic the "self-expressiveness" property that has proven effective in traditional subspace clustering. Being differentiable, our new self-expressive layer provides a simple but effective way to learn pairwise affinities between all data points through a standard back-propagation procedure. Being nonlinear, our neural-network based method is able to cluster data points having complex (often nonlinear) structures. We further propose pre-training and fine-tuning strategies that let us effectively learn the parameters of our subspace clustering networks. Our experiments show that the proposed method significantly outperforms the state-of-the-art unsupervised subspace clustering methods.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/da0ae07cafd76210fed537e4a776cb52b3c50174

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

Deep Comprehensive Correlation Mining for Image Clustering (2019, methodology): “[52] adopts the K-means loss, and [23, 42, 41] incorporate the self-representation based subspace clustering loss.”
Generalized Latent Multi-View Subspace Clustering (2020, methodology): “In contrast to thesemethods, which learn the latent representation by linearly [37] or non-linearly [43], [44] mapping the original single-view data, our method jointly recovers the latent multi-view representation and the mappings corresponding to different views to encode the intrinsic complementa”
Structured AutoEncoders for Subspace Clustering (2018, background): “One of pioneer works in deep subspace clustering is [38] which is remarkably different from our method.”
OLE: Orthogonal Low-rank Embedding, A Plug and Play Geometric Loss for Deep Learning (2017, methodology): “In the unsupervised learning domain, a very recent family of methods proposes to enforce a locally linear structure in the deep representations, such that Subspace Clustering [33] can be later applied to the deep representations [13, 23].”
Multi-Scale Fusion Subspace Clustering Using Similarity Constraint (2020, methodology): “The self-expression module is used to obtain a self-expression coefﬁcient matrix of the input latent variable Z , which models the self-expression property of data into a fully-connected layer without any activation and bias [13, 42, 43].”
ClusterGAN : Latent Space Clustering in Generative Adversarial Networks (2018, unknown): “Architectures for deep unsupervised subspace clustering have also been built on the encoder-decoder framework [15].”
Structural Deep Clustering Network (2020, background): “a representation. Variational deep embedding [9] is able to model the data generation process and clusters jointly by using a deep variational autoencoder, so as to achieve better clustering results. [8] proposes deep subspace clustering networks, which uses a novel self-expressive layer between the”
A Survey of Clustering With Deep Learning: From the Perspective of Network Architecture (2018, background): “• Deep Subspace Clustering Networks (DSC-Nets): DSC-Nets [37] introduces a novel autoencoder architecture to learn an explicit non-linear mapping that is 39504 VOLUME 6, 2018 with a reconstruction loss.”
Deep Spectral Clustering Using Dual Autoencoder Network (2019, methodology): “In [16], a deep au-toencoder is trained to minimize a reconstruction loss together with a self-expressive layer.”
Structured Graph Learning for Scalable Subspace Clustering: From Single View to Multiview (2021, unknown): “To enjoy the benefit of discriminative features brought by deep neural works, some deep subspace clustering networks have recently been proposed [13], [14].”
C2AE: Class Conditioned Auto-Encoder for Open-Set Recognition (2019, methodology): “Recent advancements in computer vision have resulted in significant improvements for tasks such as image classification [16], [24], [17], [48], Detection [41], [40], [4], [49], Clustering [19], [1], [8], [2], etc.”