Network Learning with Local Propagation
This work addresses the challenge of efficient and interpretable learning in complex networks, such as those with infinite paths or multiple goals, though it appears incremental in its approach.
The paper tackles the problem of network parameter learning by proposing a locally decoupled approach with local propagation, which enables decoupled and parallel learning in graph-based networks. Preliminary results show advantages in learning time and network size compared to state-of-the-art methods while maintaining comparable recognition accuracy.
This paper presents a locally decoupled network parameter learning with local propagation. Three elements are taken into account: (i) sets of nonlinear transforms that describe the representations at all nodes, (ii) a local objective at each node related to the corresponding local representation goal, and (iii) a local propagation model that relates the nonlinear error vectors at each node with the goal error vectors from the directly connected nodes. The modeling concepts (i), (ii) and (iii) offer several advantages, including (a) a unified learning principle for any network that is represented as a graph, (b) understanding and interpretation of the local and the global learning dynamics, (c) decoupled and parallel parameter learning, (d) a possibility for learning in infinitely long, multi-path and multi-goal networks. Numerical experiments validate the potential of the learning principle. The preliminary results show advantages in comparison to the state-of-the-art methods, w.r.t. the learning time and the network size while having comparable recognition accuracy.