Research map / Learning theory and algorithms / Clustering
Deep Clustering: On the Link Between Discriminative Models and K-Means
Mohammed Jabi, Marco Pedersoli, Amar Mitiche, Ismail Ben Ayed
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace direct mutual information maximization (MI-D) by ADM (MI-ADM) (ablation) | mutual information clustering · FRGC clustering | ACC: 0.322 -> 0.431 | Improved |
| replace direct mutual information maximization (MI-D) by ADM (MI-ADM) (ablation) | mutual information clustering · CMU-PIE clustering | ACC: 0.705 -> 0.858 | Improved |
| replace direct mutual information maximization (MI-D) by ADM (MI-ADM) (ablation) | mutual information clustering · MNIST-full clustering | ACC: 0.967 -> 0.969 | No clear effect |
| replace direct mutual information maximization (MI-D) by ADM (MI-ADM) (ablation) | mutual information clustering · USPS clustering | ACC: 0.979 -> 0.979 | No clear effect |
| replace direct mutual information maximization (MI-D) by ADM (MI-ADM) (ablation) | mutual information clustering · YTF clustering | ACC: 0.616 -> 0.606 | Mixed |
| use discriminative MI-ADM instead of generative SR-K-means | clustering · MNIST-full clustering | ACC: 0.939 -> 0.969 | Improved |
| use discriminative MI-ADM instead of generative SR-K-means | clustering · FRGC clustering | NMI: 0.487 -> 0.580 | Improved |
| use MI-ADM instead of SR-K-means on unbalanced data | clustering · YTF clustering | ACC: 0.605 -> 0.606 | No clear effect |
| use MI-ADM versus DEPICT | clustering · MNIST-full clustering | ACC: 0.971 -> 0.969 | No clear effect |
| add entropy term H(Q) to the objective (ablation) | MI-ADM · six clustering datasets | ACC | No clear effect |
| use regularized soft K-means instead of DCN hard K-means | SR-K-means · MNIST-full clustering | ACC: 0.83 -> 0.939 | Improved |
Closest papers by meaning
- Towards K-means-friendly Spaces: Simultaneous Deep Learning and Clustering (ICML 2017 · similarity 0.90)
- Deep Clustering via Joint Convolutional Autoencoder Embedding and Relative Entropy Minimization (ICCV 2017 · similarity 0.89)
- DeepDPM: Deep Clustering With an Unknown Number of Clusters (CVPR 2022 · similarity 0.89)
- Samples Are Not Equal: A Sample Selection Approach for Deep Clustering (ICLR 2026 · similarity 0.88)
- From Logits to Hierarchies: Hierarchical Clustering made Simple (ICML 2025 · similarity 0.87)
- You Can Trust Your Clustering Model: A Parameter-free Self-Boosting Plug-in for Deep Clustering (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
