ARXIV:2604.12325 · OPTIMIZATION AI · SUBMITTED 15 APR · 16:59 UTC · FRESHNESS STALE

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks

Azza Fadhel · The Hung Tran · Trong Nghia Hoang · Jana Doppa · arXiv

A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

Ship in 2-4 weeks›Score7.0Evidence unverified

Opportunity summary

Pain A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

Evidence 0 refs | 3 sources | 50% coverage

Blocker Evidence unverified

Open Build Read PDF Signal Canvas Track

PROBLEM

A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets. A key challenge in this setting is data scarcity: in many scientific applications, only small or poor-quality datasets are…

METHOD

Full abstract

We consider the problem of offline black-box optimization, where the goal is to discover optimal designs (e.g., molecules or materials) from past experimental data. A key challenge in this setting is data scarcity: in many scientific applications, only small or poor-quality datasets are available, which severely limits the effectiveness of existing algorithms. Prior work has theoretically and empirically shown that performance of offline optimization algorithms depends on how well the surrogate model captures the optimization bias (i.e., ability to rank input designs correctly), which is challenging to accomplish with limited experimental data. This paper proposes Surrogate Learning with Optimization Bias via Synthetic Task Generation (OptBias), a meta-learning framework that directly tackles data scarcity. OptBias learns a reusable optimization bias by training on synthetic tasks generated from a Gaussian process, and then fine-tunes the surrogate model on the small data for the target task. Across diverse continuous and discrete offline optimization benchmarks, OptBias consistently outperforms state-of-the-art baselines in small data regimes. These results highlight OptBias as a robust and practical solution for offline optimization in realistic small data settings.

RESULT

ScienceToStartup currently rates this 7.0/10 on the public viability pass. These results highlight OptBias as a robust and practical solution for offline optimization in realistic small data settings. Code availability is flagged in the…

WHY NOW

Optimization AI moved forward this cycle; last verified April 2026. Public score 7.0/10. Production flags indicate code availability.

Continue into Read for claims, analysis, references, and neighboring papers.

Opportunity summary

Score7.0

PainA meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

Evidence0 refs | 3 sources | 50% coverage

Blockerno shell-level blocker reported

Analysis summary

A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

VerifiedSource: PDF linkedVerifiedPaperPack: citation fields availablePartialProof: unverified proof status

Competitive landscape

A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

Segment

Optimization AI

Adoption evidence

No public code link in the paper record yet

Commercial read

7.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

{ "contract_version": "paper-r2", "paper_id": "65b25cb2-cf32-4c3f-a42c-5a663b9be849", "arxiv_id": "2604.12325", "canonical_route": "/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "active_tab": "synced from current hash by the drawer client", "selected_artifact": "black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "endpoints": { "paper_pack": "/api/v1/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks/paper-pack", "build_passport": "/api/v1/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks/build-passport", "mcp_resource": "sciencetostartup://surfaces/paper-workspace" } }

{ "surface": "paper", "mode": "paper", "query": "Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks", "normalized_query": "2604.12325", "route": "/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "paper_ref": "black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "topic_slug": null, "benchmark_ref": null, "dataset_ref": null }

{ "@context": "https://schema.org", "@graph": [ { "@type": "WebPage", "@id": "https://sciencetostartup.com/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks#webpage", "url": "https://sciencetostartup.com/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "name": "Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks", "description": "A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.", "isPartOf": { "@id": "https://sciencetostartup.com/#website" } }, { "@type": "ScholarlyArticle", "@id": "https://sciencetostartup.com/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks#scholarlyArticle", "headline": "Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks", "description": "A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.", "url": "https://sciencetostartup.com/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks", "sameAs": "https://arxiv.org/abs/2604.12325", "identifier": { "@type": "PropertyValue", "propertyID": "arXiv", "value": "2604.12325" }, "isAccessibleForFree": true, "isPartOf": { "@id": "https://sciencetostartup.com/#website" }, "datePublished": "2026-04-14T06:00:30.000Z", "author": [ { "@type": "Person", "name": "Azza Fadhel" }, { "@type": "Person", "name": "The Hung Tran" }, { "@type": "Person", "name": "Trong Nghia Hoang" }, { "@type": "Person", "name": "Jana Doppa" } ], "additionalProperty": [ { "@type": "PropertyValue", "propertyID": "viabilityScore", "value": 7 }, { "@type": "PropertyValue", "propertyID": "researchDomain", "value": "Optimization AI" }, { "@type": "PropertyValue", "propertyID": "commercialReadiness", "value": "code" } ] }, { "@type": "BreadcrumbList", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://sciencetostartup.com" }, { "@type": "ListItem", "position": 2, "name": "Optimization AI", "item": "https://sciencetostartup.com/topics" }, { "@type": "ListItem", "position": 3, "name": "Black-Box Optimization From Small Offline Datasets via Meta ", "item": "https://sciencetostartup.com/paper/black-box-optimization-from-small-offline-datasets-via-meta-learning-with-synthetic-tasks" } ] } ] }

Competitive landscape

A meta-learning framework that generates synthetic tasks to enable black-box optimization from small, offline datasets.

Segment

Optimization AI

Adoption evidence

No public code link in the paper record yet

Commercial read

7.0/10 public viability

Direct

not classified

Adjacent

not classified

Substitute

not classified

Unknown

not classified

Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks

Black-Box Optimization From Small Offline Datasets via Meta Learning with Synthetic Tasks

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline

Claim map

Constellation map

Competitive landscape

Buzz

PDF

REFERENCES

Related Papers

Related Resources

Subscribe to the weekly brief

Build artifacts

Brief

Experiment plan

Validation checklist

Scientific founder

Translational engineer

Domain operator

GTM lead

Regulatory/clinical advisor

Timeline