Yanchen Jiang

Google (United States) · United States

Unclaimed · AI-structured profileOpenAlex

Online learning theorist at Google: anytime expert-prediction regret and horizon-uniform O(d) inverse optimization.

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—

Publication-volume data for this axis is not yet available in the snapshot, so no score is given. Output would measure total peer-reviewed and preprint production over time, venue distribution, and authorship role across the career record.

Impact—

Citation-based impact is not yet measurable: OpenAlex reports h-index 0 and i10-index 0, and all recorded works date to 2026, so no citation window exists. This axis would track citation accumulation and downstream influence over time.

Focus0/100

Evolve's focus axis scores 0/100 on a very thin record. The underlying OpenAlex field distribution is in fact narrow, splitting evenly (0.5/0.5) between 'online learning / learning theory' and 'online learning and optimization theory', two adjacent fields, while topic mass concentrates in Advanced Bandit Algorithms (2) and Optimization and Search Problems (2), with smaller mass in Stochastic Gradient Optimization Techniques (1) and Machine Learning and Algorithms (1).

Collaboration50/100

The collaboration score is 50/100. OpenAlex records 11 distinct coauthors, and the frequent-coauthor list forms a dense theory cluster: Yang Cai and Di Wang each appear twice, alongside Anupam Gupta, Aranyak Mehta, Renato Paes Leme, Guru Guruganesh, Christopher Liaw, Vineet Gupta, and Grigoris Velegkas. Both reviewed papers put Jiang in a middle author position without corresponding status, indicating team-based theory work rather than solo-led projects in the visible sample.

Activity32/100

The activity score is 32/100. The OpenAlex counts-by-year profile contains a single entry — 2 works in 2026 — and recent-3-year works also equals 2, so the visible record spans one publication year. Both works are arXiv preprints reviewed in the Evolve catalog; earlier output is not captured in this snapshot, which keeps the activity reading low.

Reproducibility100/100

Reproducibility scores 100/100, the maximum, with a mean of 100 computed over 2 reviewed papers. Both carry Evolve's 'high' reproducibility readiness: the anytime-regret paper's proofs were checked and confirmed, and the online inverse optimization paper's short, self-contained proof was verified correct on extraction. For pure-theory work this reflects verifiable, self-contained mathematics rather than artifact-sharing pipelines.

Research summary

Yanchen Jiang is a researcher at Google (United States) whose reviewed work sits in online learning theory. This profile is resolved from OpenAlex bibliographic data, and the Evolve catalog holds two 2026 arXiv preprints, both middle-authorship, non-corresponding papers that passed human-verified correctness review. The first, 'Prediction with Expert Advice: Anytime Regret with Many Experts Matches the Fixed-Time Constant', gives an anytime algorithm whose minimax regret matches the fixed-horizon constant, removing a √2 overhead; the proofs check out, with the caveat that the overhead is asymptotic and vanishes very slowly. The second, 'Efficient Online Inverse Optimization with O(d) Regret', establishes deterministic O(d) regret, uniform in the horizon T, for proper online inverse optimization via a trace-power potential; the proof is short, self-contained, and verified correct on extraction. Both papers rate 'high' on reproducibility readiness, and the snapshot's reproducibility mean is 100 over 2 reviewed papers. The field distribution splits evenly between online learning/learning theory and online learning and optimization theory, with topic mass in advanced bandit algorithms and optimization and search problems. OpenAlex lists 11 distinct coauthors, with Yang Cai and Di Wang recurring; h-index and i10-index both read 0, consistent with a preprint-stage record — an early-stage, theory-focused profile.

Collaboration pattern

Research keywords

online learninglearning theoryprediction with expert adviceanytime regretminimax regretmultiplicative weights updateparameter-free algorithmsonline inverse optimizationregret bounds uniform in horizonvariable-metric online gradient descentcutting-plane methodsbandit algorithms

Highlights

Anytime many-expert regret matches the fixed-time constant
Evolve's human-verified review confirms this 2026 preprint builds an anytime prediction-with-expert-advice algorithm whose minimax regret matches the fixed-horizon constant, removing a √2 overhead, with parameter-free operation under unknown horizons; the proofs check out, while the overhead is asymptotic and vanishes very slowly.
Deterministic O(d)-regret online inverse optimization, uniform in T
The second 2026 preprint claims and verifies deterministic O(d) regret, uniform in horizon T, for proper online inverse optimization via a trace-power potential tr(H^{-1/2}), variable-metric online gradient descent, and cutting-plane methods; the proof is short, self-contained, and correct on extraction, with high reproducibility readiness.

Timeline

2026The arXiv preprint 'Prediction with Expert Advice: Anytime Regret with Many Experts Matches the Fixed-Time Constant' enters Evolve's review catalog and passes verification, matching the fixed-horizon regret constant in the many-experts setting.
2026The second reviewed preprint, 'Efficient Online Inverse Optimization with O(d) Regret', joins the catalog; its deterministic, horizon-uniform O(d) regret proof is verified correct on extraction.
2026Snapshot state: OpenAlex counts 2 works in 2026 (recent-3-year works = 2), both reviewed papers rate high on reproducibility, giving a reproducibility mean of 100.

Papers in the Evolve catalog

Frequent co-authors

Vineet Gupta×2 Christopher Liaw×2 Aranyak Mehta×2 Grigoris Velegkas×2 Yang Cai×1 Di Wang×1 Yang Cai×1 Anupam Gupta×1 Guru Guruganesh×1 Renato Paes Leme×1 Di Wang×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.

Identity is name-resolved from OpenAlex public bibliographic data (no ORCID provided); 'Yanchen Jiang' is a shared name, so record boundaries may include or omit homonymous authors.
Bibliometrics come from OpenAlex only; the visible record spans a single publication year (2026, 2 works), so volume-, citation-, and career-level metrics are incomplete.
Citation impact cannot be assessed: OpenAlex reports h-index 0 and i10-index 0 on this preprint-stage record.
Coauthor affiliations are not verified in the input data; the frequent-coauthor list is taken from the snapshot as-is, including its repeated entries.