Mining On Manifolds: Metric Learning Without Labels

Abstract

In this work we present a novel unsupervised framework for hard training example mining. The only input to the method is a collection of images relevant to the target application and a meaningful initial representation, provided e.g. by pre-trained CNN. Positive examples are distant points on a single manifold, while negative examples are nearby points on different manifolds. Both types of examples are revealed by disagreements between Euclidean and manifold similarities. The discovered examples can be used in training with any discriminative loss. The method is applied to unsupervised fine-tuning of pre-trained networks for fine-grained classification and particular object retrieval. Our models are on par or are outperforming prior models that are fully or partially supervised.

Stats

• citations80
• S2 citations—
• github stars0
• HF likes0
• heat score14.31
• arxiv keyiscen2018mining

Related papers

• Piecewise-linear Manifolds For Deep Metric Learning (2024)0.00
• Metric Learning On Manifolds (2019)0.00
• For Manifold Learning, Deep Neural Networks Can Be Locality Sensitive Hash Functions (2021)0.00
• Dynamic Sampling For Deep Metric Learning (2020)5.84
• Cross-modal Manifold Learning For Cross-modal Retrieval (2016)0.00
• MIC: Mining Interclass Characteristics For Improved Metric Learning (2019)11.93
• Self-taught Metric Learning Without Labels (2022)9.41
• Quadric Hypersurface Intersection For Manifold Learning In Feature Space (2021)0.00