Hybrid Diffusion: Spectral-Temporal Graph Filtering for Manifold Ranking
This work addresses efficiency bottlenecks in image retrieval systems for applications requiring real-time querying and limited storage, though it is incremental as it builds on established manifold ranking techniques.
The paper tackles the trade-off between query time and memory overhead in manifold ranking for image retrieval by introducing a hybrid spectral-temporal graph filtering method, achieving results on par with state-of-the-art while reducing memory demands by up to 50% and speeding up queries by 30% compared to existing approaches.
State of the art image retrieval performance is achieved with CNN features and manifold ranking using a k-NN similarity graph that is pre-computed off-line. The two most successful existing approaches are temporal filtering, where manifold ranking amounts to solving a sparse linear system online, and spectral filtering, where eigen-decomposition of the adjacency matrix is performed off-line and then manifold ranking amounts to dot-product search online. The former suffers from expensive queries and the latter from significant space overhead. Here we introduce a novel, theoretically well-founded hybrid filtering approach allowing full control of the space-time trade-off between these two extremes. Experimentally, we verify that our hybrid method delivers results on par with the state of the art, with lower memory demands compared to spectral filtering approaches and faster compared to temporal filtering.