DeepDiffusion: Unsupervised Learning of Retrieval-Adapted Representations via Diffusion-Based Ranking on Latent Feature Manifold.
Takahiko FuruyaRyutarou OhbuchiPublished in: IEEE Access (2022)
Keyphrases
- unsupervised learning
- supervised learning
- manifold ranking
- information retrieval
- learning to rank
- ranking algorithm
- ranked retrieval
- retrieval functions
- document retrieval
- image database
- expert finding
- information retrieval systems
- ranking functions
- retrieval quality
- vector representation
- web search
- anisotropic diffusion
- manifold learning
- image features
- feature ranking
- document ranking
- computer vision
- retrieval systems
- test collection
- dimensionality reduction
- semi supervised
- feature vectors
- latent variables
- diffusion process
- expectation maximization
- deep learning
- tensor field
- image retrieval
- training set
- feature space
- heat diffusion