Statistical learning theorist at TU Braunschweig: kernel methods, Rademacher bounds, minimax early stopping.
2
papers here
2
reviewed by Evolve
0
first-author
0
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—
Not yet scored: no publication-count data is available in the underlying record. This axis would measure scholarly productivity — papers per year and cumulative output across venues — and requires a fuller bibliographic snapshot.
Impact—
Not yet scored: no citation counts, h-index, or venue-level impact metrics exist in the current data. This axis would measure uptake of the researcher's work — citations, citing papers, and influence on follow-up theory.
Focus0/100
Scores 0/100 in the current radar. The axis measures topical spread of cataloged output. The raw field distribution is nearly singular: 0.5 statistical learning theory (kernel methods, nonparametric regression) and 0.5 statistical learning theory — two overlapping labels for what is effectively one research program on generalization theory for kernels and neural networks. On the field split alone, this reads as a highly concentrated, single-theme profile.
Collaboration28/100
Scores 28/100, computed from 3 distinct coauthors across the cataloged record — a small, tight-knit collaboration footprint. The recurring partners are Mike Nguyen and Mahdi Mohammadigohari, the only repeated names in the coauthor record. On both cataloged papers Mücke is the last-listed author but not the corresponding author, consistent with senior or supervisory positioning within compact two-to-three-author teams rather than large consortium-style collaborations.
Activity—
Not yet scored: the data covers only two 2026 arXiv papers, so recent publication cadence cannot be established. This axis would measure ongoing research activity — annual posting rate, preprint-to-publication flow, and continuity of output over time.
Reproducibility55/100
Scores 55/100, the mean readiness rating across the 2 papers Evolve has reviewed — both rated medium. The pattern is consistent: strong, carefully argued theory paired with limited practical artifacts. The Brownian-heads paper is judged convincing but its code remains unreleased, and the discrepancy-principle paper carries strong theory with thin experiments; releasing code and fuller empirical sections would lift this axis.
Research summary
Nicole Mücke is a statistical learning theorist affiliated with Technische Universität Braunschweig whose cataloged work centers on rigorous generalization theory for kernel and neural-network models. The Evolve catalog holds two 2026 arXiv papers, on both of which Mücke is the last-listed author (not marked corresponding). The first, 'A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks,' develops a multi-scale discrepancy rule yielding adaptive early stopping with minimax rates for kernel methods, random features, and NTK-based networks; Evolve's review calls the theory strong but the experiments thin. The second, 'Brownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample Selection,' derives sharp activation-mass Rademacher bounds for Brownian heads and quantifies the cost of same-sample selection; the verdict is convincing theory with code unreleased. Both papers carry a medium reproducibility-readiness rating, and the reproducibility dimension scores 55 as the mean across those two reviews. The field distribution is tightly concentrated — statistical learning theory, split evenly between kernel-methods/nonparametric-regression work and the broader field — and collaboration is small-scale, with three distinct coauthors (score 28), including recurring partners Mike Nguyen and Mahdi Mohammadigohari. Output, impact, and activity metrics are not yet available in the underlying data.
Collaboration pattern
Collaboration is compact: 3 distinct coauthors yield a score of 28/100. Mike Nguyen and Mahdi Mohammadigohari are the recurring partners in the record, while Mücke takes the last author position — without corresponding authorship — on both 2026 cataloged papers, a pattern typical of senior supervisory roles in small theory groups rather than large multi-institution teams.
Research keywords
Highlights
Minimax adaptive early stopping via a smoothed discrepancy principle
Sharp activation-mass Rademacher bounds for Brownian heads
Timeline
- 2026Posts 'A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks' to arXiv, establishing adaptive early stopping with minimax rates for kernels, random features, and NTK networks.
- 2026Posts 'Brownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample Selection' to arXiv, deriving sharp activation-mass Rademacher bounds for deep ReLU representations.
- 2026Enters the Evolve catalog with two last-author theory papers, both rated medium for reproducibility readiness, yielding a reproducibility score of 55 and a three-coauthor collaboration footprint.
Papers in the Evolve catalog
Frequent co-authors
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.
- Bibliometric dimensions are computed from a small snapshot: only 2 reviewed papers underpin the reproducibility score, and 3 distinct coauthors underpin the collaboration score.
- Output, impact, and activity axes are missing — no publication counts, citation metrics, or h-index data exist in the input; nothing on those axes is estimated.
- Profile is name-resolved from public bibliographic data via ORCID 0000-0002-5708-1820; affiliation is limited to Technische Universität Braunschweig and no country field is provided.
- The supplied frequent-coauthor list includes the researcher's own name, likely a de-duplication artifact; only Nguyen and Mohammadigohari are treated as external collaborators.