Postdoc, York University
1 paper at NeurIPS 2025
We show that adaptive optimizers like RMSProp lead to fairer minima than SGD, both theoretically and empirically.
