{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:OQFFAOHPL6WWCJQCQ4XW3RHNCS","short_pith_number":"pith:OQFFAOHP","schema_version":"1.0","canonical_sha256":"740a5038ef5fad612602872f6dc4ed149fc310609e606d4c79718d12fa8b7971","source":{"kind":"arxiv","id":"1905.10843","version":8},"attestation_state":"computed","paper":{"title":"Asymptotic learning curves of kernel methods: empirical data v.s. Teacher-Student paradigm","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG"],"primary_cat":"stat.ML","authors_text":"Mario Geiger, Matthieu Wyart, Stefano Spigler","submitted_at":"2019-05-26T17:29:11Z","abstract_excerpt":"How many training data are needed to learn a supervised task? It is often observed that the generalization error decreases as $n^{-\\beta}$ where $n$ is the number of training examples and $\\beta$ an exponent that depends on both data and algorithm. In this work we measure $\\beta$ when applying kernel methods to real datasets. For MNIST we find $\\beta\\approx 0.4$ and for CIFAR10 $\\beta\\approx 0.1$, for both regression and classification tasks, and for Gaussian or Laplace kernels. To rationalize the existence of non-trivial exponents that can be independent of the specific kernel used, we study "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"1905.10843","kind":"arxiv","version":8},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2019-05-26T17:29:11Z","cross_cats_sorted":["cond-mat.dis-nn","cs.LG"],"title_canon_sha256":"253f4639c76cde956186f013953596a5498f9ca9863ecca2b3f4d747ce1c8b4e","abstract_canon_sha256":"3b2902f026353e19e86ea2531bf29920c987d0f6044d074d1ab21f37db312c79"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:11:38.441267Z","signature_b64":"I1O93wpt4a/uEmaCIWX8XT/azn979OOf5JwIpdgVoWV728ZCMbTS4I2ooCBay28xL/vaLYEYYdLM+kOIgp9CCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"740a5038ef5fad612602872f6dc4ed149fc310609e606d4c79718d12fa8b7971","last_reissued_at":"2026-07-05T02:11:38.440849Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:11:38.440849Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Asymptotic learning curves of kernel methods: empirical data v.s. Teacher-Student paradigm","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG"],"primary_cat":"stat.ML","authors_text":"Mario Geiger, Matthieu Wyart, Stefano Spigler","submitted_at":"2019-05-26T17:29:11Z","abstract_excerpt":"How many training data are needed to learn a supervised task? It is often observed that the generalization error decreases as $n^{-\\beta}$ where $n$ is the number of training examples and $\\beta$ an exponent that depends on both data and algorithm. In this work we measure $\\beta$ when applying kernel methods to real datasets. For MNIST we find $\\beta\\approx 0.4$ and for CIFAR10 $\\beta\\approx 0.1$, for both regression and classification tasks, and for Gaussian or Laplace kernels. To rationalize the existence of non-trivial exponents that can be independent of the specific kernel used, we study "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1905.10843","kind":"arxiv","version":8},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1905.10843/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"1905.10843","created_at":"2026-07-05T02:11:38.440903+00:00"},{"alias_kind":"arxiv_version","alias_value":"1905.10843v8","created_at":"2026-07-05T02:11:38.440903+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.10843","created_at":"2026-07-05T02:11:38.440903+00:00"},{"alias_kind":"pith_short_12","alias_value":"OQFFAOHPL6WW","created_at":"2026-07-05T02:11:38.440903+00:00"},{"alias_kind":"pith_short_16","alias_value":"OQFFAOHPL6WWCJQC","created_at":"2026-07-05T02:11:38.440903+00:00"},{"alias_kind":"pith_short_8","alias_value":"OQFFAOHP","created_at":"2026-07-05T02:11:38.440903+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.19680","citing_title":"Feature learning is decoupled from generalization in high capacity neural networks","ref_index":59,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS","json":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS.json","graph_json":"https://pith.science/api/pith-number/OQFFAOHPL6WWCJQCQ4XW3RHNCS/graph.json","events_json":"https://pith.science/api/pith-number/OQFFAOHPL6WWCJQCQ4XW3RHNCS/events.json","paper":"https://pith.science/paper/OQFFAOHP"},"agent_actions":{"view_html":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS","download_json":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS.json","view_paper":"https://pith.science/paper/OQFFAOHP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1905.10843&json=true","fetch_graph":"https://pith.science/api/pith-number/OQFFAOHPL6WWCJQCQ4XW3RHNCS/graph.json","fetch_events":"https://pith.science/api/pith-number/OQFFAOHPL6WWCJQCQ4XW3RHNCS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS/action/storage_attestation","attest_author":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS/action/author_attestation","sign_citation":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS/action/citation_signature","submit_replication":"https://pith.science/pith/OQFFAOHPL6WWCJQCQ4XW3RHNCS/action/replication_record"}},"created_at":"2026-07-05T02:11:38.440903+00:00","updated_at":"2026-07-05T02:11:38.440903+00:00"}