Machine learning is physics (Ep. 211)
Data Science at Home
00:00
The Equilibrium in Machine Learning
equilibrium and I already mentioned something at the beginning of this episode about equilibrium equilibrium is you know the state from which we you know the network the model tends to but it's gonna be more difficult for the network or the model to escape right that's what an equilibrium is. Equilibrium can be stable unstable and we can have several types of equilibria but an equilibrium in a stable equilibrium is essentially hard to move out from there. Even if two networks with different parameters will behave the same on a macroscopic level, their internal parameter values are going to be slightly different. The system forgets what happened in the past and still behaves the same way nowThis is absolutely possible and i believe this
Transcript
Play full episode
