4 ms·

The one that matters is missing: how do you get the best random seeds? /s

by visarga 5y ago

jacquesm 5y ago
4 There you go.
- mellavora 5y ago
  I have to disagree, and this close to thanksgiving I just don't see how you missed the best random seed pumpkin 3.14 Normally I prefer apple, but during the holidays you must adjust.
WithinReason 5y ago
41. Remember to consider variance of results and the usage of random seeds.
go_elmo 5y ago
Great one! Actual question: has anyone looked if pseudo randomnes is worse than full randomness?
- dotancohen 5y ago
  It is pseudoworse.
- 317070 5y ago
  I once (8 years ago) did a comparison between random, pseudorandom and quasirandom. Quasirandom worked ever so slightly better than the other two in speed and final performance, but not enough to warrant the additional complexity of implementation.
oigursh 5y ago
# Sussman attains enlightenment In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. “What are you doing?”, asked Minsky. “I am training a randomly wired neural net to play Tic-Tac-Toe” Sussman replied. “Why is the net wired randomly?”, asked Minsky. “I do not want it to have any preconceptions of how to play”, Sussman said. Minsky then shut his eyes. “Why do you close your eyes?”, Sussman asked his teacher. “So that the room will be empty.” At that moment, Sussman was enlightened. http://www.catb.org/jargon/html/koans.html http://www.catb.org/jargon/html/koans.html
- bruce343434 5y ago
  I'm not actually sure what the morale is here
  - AstralStorm 5y ago
    Intelligence does not start from a random state, but from a pretrained one. Random state is likely to roll an artificial overfit or if it just fits, you won't know why it is better than another state - or rather why your training method sucks from other initial states. Proper methods of bias analysis are network compression or random surgery, and expanding test data set. Proper methods of training are robust to varied initial states.
  - ximeng 5y ago
    If you shut your eyes, you do not see anything in the room. This does not mean there is nothing in the room. If you randomize your seeds, you do not deliberately bias your model. This does not mean your model is unbiased.

The one that matters is missing: how do you get the best random seeds? /s

jacquesm 5y ago

4 There you go.

mellavora 5y ago
I have to disagree, and this close to thanksgiving I just don't see how you missed the best random seed pumpkin 3.14 Normally I prefer apple, but during the holidays you must adjust.

WithinReason 5y ago

41. Remember to consider variance of results and the usage of random seeds.

go_elmo 5y ago

Great one! Actual question: has anyone looked if pseudo randomnes is worse than full randomness?

dotancohen 5y ago
It is pseudoworse.
317070 5y ago
I once (8 years ago) did a comparison between random, pseudorandom and quasirandom. Quasirandom worked ever so slightly better than the other two in speed and final performance, but not enough to warrant the additional complexity of implementation.

oigursh 5y ago

# Sussman attains enlightenment In the days when Sussman was a novice, Minsky once came to him as he sat hacking at the PDP-6. “What are you doing?”, asked Minsky. “I am training a randomly wired neural net to play Tic-Tac-Toe” Sussman replied. “Why is the net wired randomly?”, asked Minsky. “I do not want it to have any preconceptions of how to play”, Sussman said. Minsky then shut his eyes. “Why do you close your eyes?”, Sussman asked his teacher. “So that the room will be empty.” At that moment, Sussman was enlightened. http://www.catb.org/jargon/html/koans.html http://www.catb.org/jargon/html/koans.html

bruce343434 5y ago
I'm not actually sure what the morale is here
- AstralStorm 5y ago
  Intelligence does not start from a random state, but from a pretrained one. Random state is likely to roll an artificial overfit or if it just fits, you won't know why it is better than another state - or rather why your training method sucks from other initial states. Proper methods of bias analysis are network compression or random surgery, and expanding test data set. Proper methods of training are robust to varied initial states.
- ximeng 5y ago
  If you shut your eyes, you do not see anything in the room. This does not mean there is nothing in the room. If you randomize your seeds, you do not deliberately bias your model. This does not mean your model is unbiased.