Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Image Clustering using Restricted Boltzman Machine

Abraham Woubie, Enoch Solomon, Eyael Solomon Emiru

TL;DR

The paper addresses unsupervised image clustering when labeled data is limited by introducing a two-stage RBM-based representation: first train a universal RBM (URBM) on extensive training data, then adapt a per-image RBM to capture class-specific features, and finally form fixed-length embeddings by concatenating the adapted weights and biases. These RBM embeddings are clustered with bottom-up Agglomerative Hierarchical Clustering using cosine similarity and PLDA scoring, with a threshold-based stopping rule. Across MS-Celeb-1M and DeepFashion, the proposed AHC-RBM approach outperforms strong baselines, demonstrating robust, scalable performance for face and fashion image clustering in open-set contexts. The combination of URBM learning, per-image adaptation, and RBM-based embeddings provides a powerful unsupervised representation for clustering tasks in large-scale image datasets.

Abstract

In various verification systems, Restricted Boltzmann Machines (RBMs) have demonstrated their efficacy in both front-end and back-end processes. In this work, we propose the use of RBMs to the image clustering tasks. RBMs are trained to convert images into image embeddings. We employ the conventional bottom-up Agglomerative Hierarchical Clustering (AHC) technique. To address the challenge of limited test face image data, we introduce Agglomerative Hierarchical Clustering based Method for Image Clustering using Restricted Boltzmann Machine (AHC-RBM) with two major steps. Initially, a universal RBM model is trained using all available training dataset. Subsequently, we train an adapted RBM model using the data from each test image. Finally, RBM vectors which is the embedding vector is generated by concatenating the visible-to-hidden weight matrices of these adapted models, and the bias vectors. These vectors effectively preserve class-specific information and are utilized in image clustering tasks. Our experimental results, conducted on two benchmark image datasets (MS-Celeb-1M and DeepFashion), demonstrate that our proposed approach surpasses well-known clustering algorithms such as k-means, spectral clustering, and approximate Rank-order.

Image Clustering using Restricted Boltzman Machine

TL;DR

The paper addresses unsupervised image clustering when labeled data is limited by introducing a two-stage RBM-based representation: first train a universal RBM (URBM) on extensive training data, then adapt a per-image RBM to capture class-specific features, and finally form fixed-length embeddings by concatenating the adapted weights and biases. These RBM embeddings are clustered with bottom-up Agglomerative Hierarchical Clustering using cosine similarity and PLDA scoring, with a threshold-based stopping rule. Across MS-Celeb-1M and DeepFashion, the proposed AHC-RBM approach outperforms strong baselines, demonstrating robust, scalable performance for face and fashion image clustering in open-set contexts. The combination of URBM learning, per-image adaptation, and RBM-based embeddings provides a powerful unsupervised representation for clustering tasks in large-scale image datasets.

Abstract

In various verification systems, Restricted Boltzmann Machines (RBMs) have demonstrated their efficacy in both front-end and back-end processes. In this work, we propose the use of RBMs to the image clustering tasks. RBMs are trained to convert images into image embeddings. We employ the conventional bottom-up Agglomerative Hierarchical Clustering (AHC) technique. To address the challenge of limited test face image data, we introduce Agglomerative Hierarchical Clustering based Method for Image Clustering using Restricted Boltzmann Machine (AHC-RBM) with two major steps. Initially, a universal RBM model is trained using all available training dataset. Subsequently, we train an adapted RBM model using the data from each test image. Finally, RBM vectors which is the embedding vector is generated by concatenating the visible-to-hidden weight matrices of these adapted models, and the bias vectors. These vectors effectively preserve class-specific information and are utilized in image clustering tasks. Our experimental results, conducted on two benchmark image datasets (MS-Celeb-1M and DeepFashion), demonstrate that our proposed approach surpasses well-known clustering algorithms such as k-means, spectral clustering, and approximate Rank-order.
Paper Structure (15 sections, 2 equations, 1 figure, 4 tables)

This paper contains 15 sections, 2 equations, 1 figure, 4 tables.

Table of Contents

  1. Introduction
  2. Proposed Method
  3. RBM Vector Representation
  4. URBM Training
  5. RBM Adaptation
  6. RBM Vector Extraction
  7. Image Clustering
  8. Experimental Setup and Dataset
  9. Experimental Setup
  10. Datasets
  11. Evaluation Metrics
  12. Results
  13. Results on MS-Celeb-1M dataset
  14. Results on DeepFashion dataset
  15. Conclusion

Figures (1)

  • Figure 1: Block diagram showing different stages of the Restricted Boltzmann Machines vector extraction and its input to Bottom-up Agglomerative Hierarchical Clustering.