CMSA New Technologies in Mathematics Seminar: Local learning needs dendrites

---

I will present a learning scheme for recurrent networks in which nonlinear dendrites enable learning by generating and stabilizing task-relevant attractors within a representation manifold. The scheme is local in space and time and matches the performance of backpropagation through time on the same architecture, while following a fundamentally different trajectory in weight space. Statistical physics reveals how dendritic gating and the geometry of the representation manifold support this alternative route to learning.

Zoom link:https://harvard.zoom.us/j/95406961810  Passcode: cmsa