Yuichi Takano

University of Tsukuba · Japan

Unclaimed · AI-structured profileOpenAlex

Tsukuba researcher in optimization-based machine learning for portfolio optimization and A/B experiment design.

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Output is unscored: this snapshot lacks a complete, identity-resolved publication list, with only 9 recent-window works linked via OpenAlex. The axis would measure total publication volume, venues, and output growth across the full career.

Impact—

Impact is unscored: h-index and i10-index read zero here, most plausibly reflecting incomplete author disambiguation for a common Japanese name rather than a verified citation record. This axis would measure citations, h-index, and influence trajectories.

Focus0/100

Focus scores 0/100, consistent with the OpenAlex field distribution, which splits evenly (0.5 / 0.5) between machine learning and operations research for quantitative finance and optimization-based machine learning for experimental design and survey sampling. Topic spread — risk and portfolio optimization (4), statistical inference, stock market forecasting, and privacy-preserving technologies — reinforces this dual-track profile; it signals breadth across two distinct application domains rather than concentration.

Collaboration36/100

Collaboration scores 36/100, based on a small network of 5 distinct co-authors recorded in the recent window. The Evolve catalog identifies Shunnosuke Ikeda, Kensei Nosaka, and Tomoka Takei as recurring co-authors, and Takano appears as last author on both reviewed 2026 papers without corresponding-author status — a pattern consistent with a senior supervisory role over a compact, stable team rather than large-scale or international collaboration.

Activity67/100

Activity scores 67/100. The OpenAlex snapshot records 9 works in the recent three-year window, with countsByYear showing 9 works attributed to 2026 alone — a concentrated recent burst. Both cataloged papers are 2026 arXiv preprints, and topic counts (risk and portfolio optimization: 4; four further topics at 2 each) indicate several parallel research threads active in the same year. Older-year counts are absent, so long-term cadence cannot be verified.

Reproducibility78/100

Reproducibility scores 78/100, computed as the mean over 2 scored papers (reproMean 78, reproPapers 2). The stratified-sampling decision-tree paper is rated high readiness — exact binary-optimization formulation with offline datasets — while the decision-focused learning portfolio paper is rated medium, its ETF backtests lacking transaction costs, error bars, and significance tests. A mean near 80 across a small set signals strong methodological transparency overall, with the finance line the weaker of the two.

Research summary

Yuichi Takano is a researcher at the University of Tsukuba, Japan, whose work sits at the intersection of operations research and machine learning. The OpenAlex snapshot splits his field distribution evenly (0.5 / 0.5) between machine learning and operations research for quantitative finance and optimization-based machine learning for experimental design and survey sampling. In 2026 he posted two arXiv preprints reviewed in the Evolve catalog, in both cases as last author. The first formulates multi-way stratification for online controlled experiments as an exact binary-optimization decision-tree problem for A/B variance reduction; Evolve rates its reproducibility high while noting validation on only two offline datasets. The second proposes a KKT-based single-level decision-focused learning reformulation for constrained mean-variance portfolio optimization — claimed as a first — with ETF backtests reporting the best SR/FW/CDL results, though without transaction costs, error bars, or significance tests. OpenAlex topics are led by risk and portfolio optimization (4), alongside statistical inference, stock market forecasting, and privacy-preserving technologies in data. The snapshot records 9 works in the recent three-year window, 5 distinct co-authors, and a reproducibility mean of 78 across 2 scored papers. Citation-level impact cannot be assessed: h-index and i10-index read zero, most plausibly reflecting incomplete identity resolution for this common Japanese name rather than a verified record.

Collaboration pattern

Collaboration is compact: the OpenAlex window records 5 distinct co-authors, with Shunnosuke Ikeda, Kensei Nosaka, and Tomoka Takei recurring across work. On both reviewed 2026 arXiv papers Takano is the last author and not the corresponding author, a positioning typical of a senior group lead guiding junior researchers. The two reviewed outputs also map onto his two field clusters, suggesting the same small team supports both the quantitative finance and experimental design lines.

Research keywords

decision-focused learningpredict-then-optimizemean-variance portfolio optimizationbilevel optimizationKKT conditionsoptimal decision treesstratified samplingonline controlled experiments (A/B testing)variance reductionbinary optimizationstock market forecastingrisk and portfolio optimization

Highlights

Exact multi-way stratification trees for A/B variance reduction
As last author, Takano's 2026 arXiv paper formulates multi-way stratification for online controlled experiments as exact binary optimization, building decision trees that reduce stratified-sample variance in A/B testing. Evolve rates reproducibility high but notes validation is limited to two offline datasets with no field test.
KKT-based decision-focused learning for portfolio optimization
The second 2026 paper proposes a KKT-based single-level reformulation of decision-focused learning for constrained mean-variance portfolio optimization, claimed as a first. ETF backtests report the best SR/FW/CDL results, though Evolve flags the absence of transaction costs, error bars, and significance tests.

Timeline

2026Evolve catalog review added for 'Optimal Multi-way Decision Trees for Stratified Sampling in Online Controlled Experiments' (arXiv, last author), extending optimization-based machine learning into A/B experiment design.
2026'Decision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based Reformulation' posted to arXiv (last author), advancing the quantitative finance line with a claimed-first single-level DFL formulation.
2026OpenAlex attributes 9 works to this year (countsByYear 2026: 9; recent3yWorks 9), with topic activity led by risk and portfolio optimization (4 mentions).

Papers in the Evolve catalog

Frequent co-authors

Shunnosuke Ikeda×2 Kensei Nosaka×1 Shunnosuke Ikeda×1 Yuichi Takano×1 Tomoka Takei×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometrics are drawn from OpenAlex only; this profile is name-resolved from public bibliographic data and no ORCID is linked, so identity for the common Japanese name 'Yuichi Takano' carries residual ambiguity.
h-index and i10-index are recorded as 0 in this snapshot, most plausibly reflecting incomplete author disambiguation; the output and impact dimensions are therefore left unscored rather than estimated.
countsByYear contains only 2026 (9 works); no multi-year publication history is available, so career-long activity and impact trends cannot be assessed.
Both reviewed papers are 2026 arXiv preprints assessed through Evolve review only; peer-reviewed versions or venue acceptances are not reflected.