dorsal/arxiv
View SchemaAmbiguous model learning made unambiguous with 1/f priors
| Authors | Gurinder Singh Atwal, William Bialek |
|---|---|
| Categories | |
| ArXiv ID | q-bio/0512040 |
| URL | https://arxiv.org/abs/q-bio/0512040 |
| Journal | NIPS 16, MIT Press, Cambridge (2004) |
Abstract
What happens to the optimal interpretation of noisy data when there exists more than one equally plausible interpretation of the data? In a Bayesian model-learning framework the answer depends on the prior expectations of the dynamics of the model parameter that is to be inferred from the data. Local time constraints on the priors are insufficient to pick one interpretation over another. On the other hand, nonlocal time constraints, induced by a $1/f$ noise spectrum of the priors, is shown to permit learning of a specific model parameter even when there are infinitely many equally plausible interpretations of the data. This transition is inferred by a remarkable mapping of the model estimation problem to a dissipative physical system, allowing the use of powerful statistical mechanical methods to uncover the transition from indeterminate to determinate model learning.
{
"annotation_id": "473cf388-aae4-4502-ac33-6d01fa7b4393",
"date_created": "2026-03-02T18:01:35.126000Z",
"date_modified": "2026-03-02T18:01:35.126000Z",
"file_hash": "70d8456be2315356fbf35ca569b15dae0083e31ff094be300776bc1fe600a426",
"private": false,
"record": {
"abstract": "What happens to the optimal interpretation of noisy data when there exists\nmore than one equally plausible interpretation of the data? In a Bayesian\nmodel-learning framework the answer depends on the prior expectations of the\ndynamics of the model parameter that is to be inferred from the data. Local\ntime constraints on the priors are insufficient to pick one interpretation over\nanother. On the other hand, nonlocal time constraints, induced by a $1/f$ noise\nspectrum of the priors, is shown to permit learning of a specific model\nparameter even when there are infinitely many equally plausible interpretations\nof the data. This transition is inferred by a remarkable mapping of the model\nestimation problem to a dissipative physical system, allowing the use of\npowerful statistical mechanical methods to uncover the transition from\nindeterminate to determinate model learning.",
"arxiv_id": "q-bio/0512040",
"authors": [
"Gurinder Singh Atwal",
"William Bialek"
],
"categories": [
"q-bio.OT",
"q-bio.NC"
],
"journal_ref": "NIPS 16, MIT Press, Cambridge (2004)",
"title": "Ambiguous model learning made unambiguous with 1/f priors",
"url": "https://arxiv.org/abs/q-bio/0512040"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "bc6255bc-8158-42f2-9e4e-2018d107779a",
"id": "arXiv Dataset IDs",
"type": "Model",
"variant": "snapshot-2026-03-01",
"version": "0.1.0"
},
"user_id": 1000002
}