Skip to main content

Posts

Showing posts from March, 2013

Training tip on Neural Networks

(c. early 2013)

Iteratively put more detail when training:
Represent a low approximation, and train.
Given the best solution there, now approximate the function with a bit more detail, and repeat, starting at the point of the best solution given in the last iteration.

Resignation

Time wastes away in ardent frustration, akin to rules of the useless, wondering if following unbelieved rules is worth sacrificing potential passion.
(Written sometime in Fall 2012)

Multicast

When one addresses a crowd successfully, one is likely attending oneself foremost, as it is probably hard to attend everyone being addressed to. One is speaking one's own truth, and the manner in which the receptors take it is left to the receptors' choice.

Addressing a single person successfully, however, can be made attending either oneself's or the other's needs. Focusing one's entire attention on a person can make them feel listened to, attended, important. This is my experience.