{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/arxiv-2605-11246","title":"Support-Proximity Augmented Diffusion Estimation for Offline Black-Box Optimization","arxiv_id":"2605.11246","date":"2026-05-11","proceeding":"ICML","authors":["Yonghan Yang","Ye Yuan","Zipeng Sun","Linfeng Du","Bowei He","Haolun Wu","Can Chen","Xue Liu"],"abstract":"Offline black-box optimization aims to discover novel designs with high property scores using only a static dataset, a task fundamentally challenged by the out-of-distribution (OOD) extrapolation problem. Existing approaches typically bifurcate into inverse methods, which struggle with the ill-posed nature of mapping scores to designs, and forward methods, which often lack the distributional expressivity to quantify uncertainty effectively. In this work, we propose SPADE (Support-Proximity Augmented Diffusion Estimation), a novel framework that reimagines forward surrogate modeling through the lens of conditional generative modeling. SPADE models the forward likelihood p(y|x) using a diffusion model, but with two critical enhancements to tailor it for optimization: (1) a Calibrated Diffusion Estimation module that enforces global consistency in statistical moments and pairwise rankings, and (2) a Support-Proximity Regularization mechanism that implicitly internalizes the data manifold constraint p(x) via kNN-based density estimation. Theoretically, we prove that our regularization is first-order equivalent to maximizing a Bayesian posterior with a valid design prior. Empirically, SPADE achieves state-of-the-art performance across Design-Bench tasks and an LLM data mixture optimization benchmark.","url_abs":"https://arxiv.org/abs/2605.11246","url_pdf":"https://arxiv.org/pdf/2605.11246","source":{"archive":null,"snapshot":"2025-07-28","note":"not in the Papers with Code archive (frozen at the snapshot)","row_kind":"graph","title_abstract_authors_date":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)"},"code_links":[],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2605.11246","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2605.11246"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"mentioned_in_github":null,"is_official":null,"provenance":"deterministic:regex_extraction","mentioned_in_paper":null,"url":"https://github.com/HarryYoung2018/spade","reach":null}],"summary":{"ran":4,"ran_honours":6,"ran_draft_wrong":4,"ran_fixture":1,"unverified":5},"by_repo_kind":{"found_in_text":{"samples":20,"ran":15,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"97a9f68b697701d9","entry":"CondEpsNet","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"97a9f68b697701d9"}},{"code_sha256_prefix":"9f70f55965dff86a","entry":"KnnDensityHelper","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"9f70f55965dff86a"}},{"code_sha256_prefix":"681458c13d0167a2","entry":"ScalarDiffusionSurrogate","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"681458c13d0167a2"}},{"code_sha256_prefix":"4cf430efa36c0fb8","entry":"TimeEmbedding","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4cf430efa36c0fb8"}},{"code_sha256_prefix":"80a38ab4a9fb9601","entry":"_as_2d","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"80a38ab4a9fb9601"}},{"code_sha256_prefix":"ab5cef763a973407","entry":"_ensure_2d","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"ab5cef763a973407"}},{"code_sha256_prefix":"4db6557dac754a47","entry":"acquisition_from_samples","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4db6557dac754a47"}},{"code_sha256_prefix":"855e6bffce251fe6","entry":"apply_support_transform","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"855e6bffce251fe6"}},{"code_sha256_prefix":"71870850a265ad4e","entry":"build_knn_helper","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"71870850a265ad4e"}},{"code_sha256_prefix":"d1fe61e95e0ab9cf","entry":"clip_and_renorm","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d1fe61e95e0ab9cf"}},{"code_sha256_prefix":"d01936d34e41a35c","entry":"clip_to_bounds","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"d01936d34e41a35c"}},{"code_sha256_prefix":"36bd257d56f94f61","entry":"evolutionary_optimize","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"36bd257d56f94f61"}},{"code_sha256_prefix":"146db3f324ec3821","entry":"make_beta_schedule","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"146db3f324ec3821"}},{"code_sha256_prefix":"80a3da03a6fc9ab7","entry":"z_score","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"OUTPUT_MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"80a3da03a6fc9ab7"}},{"code_sha256_prefix":"f39bb60e2d4fae07","entry":"z_score_denorm","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"f39bb60e2d4fae07"}},{"code_sha256_prefix":"1fe070d1ecb71daf","entry":"Dataset","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1fe070d1ecb71daf"}},{"code_sha256_prefix":"42cd3d3a49f9da3c","entry":"OptimizationResult","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"42cd3d3a49f9da3c"}},{"code_sha256_prefix":"4d8039a217216ef4","entry":"SpadeConfig","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"4d8039a217216ef4"}},{"code_sha256_prefix":"47bdbaeb1cc4a7f4","entry":"mc_samples","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"47bdbaeb1cc4a7f4"}},{"code_sha256_prefix":"3ddf62b9dc3c9a75","entry":"optimize_spade","repo":"HarryYoung2018/spade","repo_kind":"found_in_text","path":"spade/optimize.py","file_url":"https://github.com/HarryYoung2018/spade/blob/HEAD/spade/optimize.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"3ddf62b9dc3c9a75"}}]},"arxiv_metadata":{"licence":"arXiv metadata, CC0 1.0 (https://info.arxiv.org/help/license)","fields":["title","abstract","authors","date"],"primary_category":"cs.LG","source":"arxiv_2026.jsonl"},"syntology_extracted_results":null}