Geometric Embedding Alignment via Curvature Matching in Transfer Learning
This addresses transfer learning challenges in domains like molecular analysis by providing a geometry-based integration method, though it appears incremental as it builds on existing differential geometry concepts.
The paper tackles the problem of integrating multiple models in transfer learning by aligning their latent space geometries using Ricci curvature matching, resulting in performance gains of 14.4% and 8.3% over benchmarks on molecular tasks.
Geometrical interpretations of deep learning models offer insightful perspectives into their underlying mathematical structures. In this work, we introduce a novel approach that leverages differential geometry, particularly concepts from Riemannian geometry, to integrate multiple models into a unified transfer learning framework. By aligning the Ricci curvature of latent space of individual models, we construct an interrelated architecture, namely Geometric Embedding Alignment via cuRvature matching in transfer learning (GEAR), which ensures comprehensive geometric representation across datapoints. This framework enables the effective aggregation of knowledge from diverse sources, thereby improving performance on target tasks. We evaluate our model on 23 molecular task pairs sourced from various domains and demonstrate significant performance gains over existing benchmark model under both random (14.4%) and scaffold (8.3%) data splits.