npub15swlxudlhx4ttcgsd4556zuqrl57qndxmt4n3dnzrkqn89nxv6lsjzx855 (npub15sw…x855) It looks like inhibition causes learning to be more stable and accurate, so without it learning would likely take longer to converge to the correct information. The Kohonen paper without inhibition doesn't say anything about how long it takes, and only mentions that "the primary purpose of lateral inhibition ... is to improve spatial acuity in pattern detection, and certain inhibitory connections also tend to standardize the
scales of the excitatory signals." It might be fast enough on input that's not very complex or noisy, and I suppose it could learn something wrong quickly..
Holly A. Gultiano /
npub12g…h55ru
2023-12-05 03:42:46
in reply to nevent1q…cfft
Author Public Key
npub12gvxrj7dnx4ylr6y5uuatp7wlga4n8ctr8t72trvtaaml8a2r3yqzh55ruPublished at
2023-12-05 03:42:46Event JSON
{
"id": "208750b0dfdbc1d2f0b507ef788fe39c49f584333dcef3dd7d1f426f544c2fd4",
"pubkey": "521861cbcd99aa4f8f44a739d587cefa3b599f0b19d7e52c6c5f7bbf9faa1c48",
"created_at": 1701747766,
"kind": 1,
"tags": [
[
"p",
"a41df371bfb9aab5e1106d694d0b801fe9e04da6daeb38b6621d8133966666bf",
"wss://relay.mostr.pub"
],
[
"p",
"32b1255615e9848f3e76314b460d27fada96f9042690c62eadbe35ce742273ef",
"wss://relay.mostr.pub"
],
[
"e",
"c0a8fa649f27f29f5511149b81e61ca89d46e9d0f7ee0b8d904b0bba3cfc46ea",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://synapse.cafe/users/axoaxonic/statuses/111525741645738690",
"activitypub"
]
],
"content": "nostr:npub15swlxudlhx4ttcgsd4556zuqrl57qndxmt4n3dnzrkqn89nxv6lsjzx855 It looks like inhibition causes learning to be more stable and accurate, so without it learning would likely take longer to converge to the correct information. The Kohonen paper without inhibition doesn't say anything about how long it takes, and only mentions that \"the primary purpose of lateral inhibition ... is to improve spatial acuity in pattern detection, and certain inhibitory connections also tend to standardize the\nscales of the excitatory signals.\" It might be fast enough on input that's not very complex or noisy, and I suppose it could learn something wrong quickly..",
"sig": "3d917ec968c0aa1af0426404fa18adf4ce09e5c751858e2a5f3d9b1cd07e0e33fe92ce54ca26070413a8077d14e784cce41d1db2ee9851249d35ed4a32009803"
}