Introduction to Neuromorphic Computing
Karla Daniela Chávez Degollado FISI 4961 · 4 credits 2026 materials coming
From the von Neumann bottleneck to memristive oxides: the physics of resistive switching and its use as a substrate for brain-inspired computing.
A sixteen-week tutorial course delivered as a series of thirty-minute lectures in the group’s weekly seminar.
The course runs in two parts. The first covers the neuromorphic paradigm itself — the energy cost of separating memory from processing, leaky integrate-and-fire and Hodgkin–Huxley neuron models, spike-timing-dependent plasticity, the memristor and its pinched hysteresis loop, crossbar arrays for in-memory computing, spiking neural networks in hardware, and reservoir computing.
The second turns to the materials. Atomic and electronic structure of oxides, defect chemistry and oxygen vacancies, valence-change memory in HfO₂, Ta₂O₅ and TiO₂, electrochemical metallization, and the Mott transition in VO₂ and NbO₂ — the “Mott memristor” as a neuristor. It closes on electrical characterization and the figures of merit that decide whether a device is viable: endurance, retention, linearity and variability.
Assessment is 40% the weekly seminar lectures, 30% a mid-semester written review, and 30% a final project presented to the group.