Posts Tagged ‘eric boyd’
CRV 447
Podcast (crealmvault): Play in new window | Download (Duration: 1:08:55 — 17.1MB)
KMO welcomes Eric Boyd back to the program. Eric runs a weekly machine learning study group in Toronto. This conversation covers recent developments in machine learning and generative algorithms like Stable Diffusion and GPT3. KMO asks Eric to unpack the role of back propagation in current methods for training an generative algorithm and describe "Forward forward," an alternative to back propagation proposed by Geoffrey Hinton.