npub1kq0nfpugx8czz3u8d2wljxephk0zmn8p2y00vqhprsw8rfn7sucsvx8hsx (npub1kq0…8hsx) npub1xkwaswarxtyfywx0exf0ve9kheuzj4c6py745r6fvfm0nfz6ctlsne7jwu (npub1xkw…7jwu)
It also strikes me that very distributed representations might create problems for credit assignment: reinforcement would lead to a lot of feature-based overgeneralization.
In other words, it makes sense that conjunctive representations (logical AND like) exist and are relatively localized.
How might one design reinforcement on distributed reps that avoid/minimize such overgeneralization?
Yohan John 🤖🧠 /
npub1x2…2jjtg
2024-01-23 18:18:18
in reply to nevent1q…uwev
Author Public Key
npub1x2cj24s4axzg70nkx995vrf8ltdfd7gyy6gvvt4dhc6uuapzw0hst2jjtgSeen on
wss://relay.nostr.bandPublished at
2024-01-23 18:18:18Event JSON
{
"id": "c7355eefcba12bb52792ed8276f9332f2c51abdacacd61ab484bd1625a3753d9",
"pubkey": "32b1255615e9848f3e76314b460d27fada96f9042690c62eadbe35ce742273ef",
"created_at": 1706033898,
"kind": 1,
"tags": [
[
"p",
"b01f34878831f02147876a9df91b21bd9e2dcce1511ef602e11c1c71a67e8731",
"wss://relay.mostr.pub"
],
[
"p",
"359dd83ba332c89238cfc992f664b6be7829571a093d5a0f496276f9a45ac2ff",
"wss://relay.mostr.pub"
],
[
"p",
"931d80e8806aa8e388c02733a74bf7bbbef68e9350568fe63034ee1c42458353",
"wss://relay.mostr.pub"
],
[
"p",
"54913e7e252bf76591fb2614a9641e83296c8f52013ed45590db89c539ad303d",
"wss://relay.mostr.pub"
],
[
"e",
"5c732dfe504b0bbad97f55078e5c165a04f7d9f72b38e8e178e0b07d74291b01",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://fediscience.org/users/DrYohanJohn/statuses/111806637559581201",
"activitypub"
]
],
"content": "nostr:npub1kq0nfpugx8czz3u8d2wljxephk0zmn8p2y00vqhprsw8rfn7sucsvx8hsx nostr:npub1xkwaswarxtyfywx0exf0ve9kheuzj4c6py745r6fvfm0nfz6ctlsne7jwu \n\nIt also strikes me that very distributed representations might create problems for credit assignment: reinforcement would lead to a lot of feature-based overgeneralization.\n\nIn other words, it makes sense that conjunctive representations (logical AND like) exist and are relatively localized. \n\nHow might one design reinforcement on distributed reps that avoid/minimize such overgeneralization?",
"sig": "ca74cb7839a93413dcb000a91a73a389a508548230d1d53ff5ad1f0f91e90429154e8c7f4628c561c0a58ce887750b67067abebb1ecc751d2d9129fe61fe42e2"
}