unsupervised-learning

This paper introduces GRaCE, an unsupervised framework for generating interpretable graph embeddings using rank-based measures, which outperforms existing methods in retrieval, classification, and clustering tasks on textual and image data.

GUIDE is a generative unsupervised framework for Chinese query correction that uses phonetic and visual shared-ID encoding to constrain corrections and adapt to changing vocabularies, outperforming baselines in experiments and online A/B testing.

This survey paper categorizes 80 unsupervised post-training methods for foundation models, organizing them by the internal update signals and presenting a unified framework for selection and evaluation.

This paper investigates unsupervised anomaly detection using flow matching on tabular data, focusing on contaminated training sets and comparing different scoring methods for robustness.

Co-RL is a cooperative multi-agent framework that enables unsupervised reasoning improvement in language and vision-language models by using peer-derived rewards, reducing reliance on ground-truth labels and mitigating training collapse through increased cohort diversity.

The paper proposes CMR-Mamba, a causal mechanism monitoring method using Mamba state-space models for unsupervised industrial fault detection, focusing on coupling faults and evaluated on electromechanical, hydraulic, and cyber-physical systems.

Proposes a dual-primal graph VAE architecture that treats ground-truth labels as latent variables to aggregate noisy crowdsourced labels, achieving state-of-the-art results on crowdsourcing benchmarks without needing a separate classifier.

Presents STCAD, a scalable framework using BERT-based encoding and CURE clustering to perform trajectory clustering and anomaly detection on terabyte-scale AIS maritime data, demonstrating stable clusters and clear separation of anomalous vessel behavior.

This paper proposes an unsupervised fire-zone segmentation method combining watershed detection with K-means clustering to improve short-term wildfire prediction, showing consistent gains over grid-based approaches across multiple French departments and forecasting models.

The paper presents an unsupervised method to extract linguistic metaphors and group them into conceptual metaphors, applying the approach to analyze framing differences in left- vs. right-leaning podcasts.

This paper introduces LUCID, an LLM-guided interpretable and training-free unsupervised community detection method that uses a four-stage pipeline inspired by phase-transition kinetics to achieve state-of-the-art performance on real-world graphs.

freeCodeCamp published a comprehensive handbook on clustering in Python, covering K-means, hierarchical, and DBSCAN clustering with implementations and visualizations.

This paper proposes RHEA, a reliability-aware framework for multimodal-attributed graph clustering that estimates node-specific modality reliability from neighborhood consensus, reconstructs unreliable modalities, and uses reliability-aware fusion and optimal transport clustering. Experiments on four benchmarks show consistent gains, especially under noisy or missing attributes.

This paper investigates deep clustering methods on imbalanced tabular data and proposes two novel ensemble approaches that aggregate clustering assignments across embedding dimensions or via majority voting, outperforming individual methods on 16 datasets.

This paper presents a novel unsupervised data augmentation method combining Gaussian Mixture Models and Large Language Models to improve clustering on imbalanced text datasets by generating synthetic documents for underrepresented clusters.

ConvGRUAutoencoder combines convolutional layers with gated recurrent units to compress and reconstruct video sequences, enabling unsupervised learning from video without human labeling.

Proposes Hybrid Latent-Structural Fusion (HLSF), a weighted anomaly fusion framework combining CP-APR structural anomaly scores with latent-space density scores from normalizing flows, improving cyber anomaly detection on real-world compromised user credentials data.

Latent actions are gaining traction in robotics as a way to learn from unlabeled video without action labels. Recent papers from DeepMind and FAIR demonstrate progress from controlled game environments to in-the-wild internet video, promising scalable training for imitation learning.

FastCentNN is an accelerated variant of the Centroid Neural Network that uses an entropy proxy based on centroid movement per epoch to trigger early splitting, reducing runtime by up to 16% while maintaining clustering quality.

This paper compares clustering algorithms for filtering wind farm SCADA data to retain only normal operation measurements, introducing evaluation metrics for unlabeled data and providing recommendations based on tests on three offshore turbines.