Yorie Nakahira

Carnegie Mellon University · United States

Unclaimed · AI-structured profileORCID OpenAlex

CMU researcher bridging fractional-order system identification and Bayesian deep learning for dynamical systems

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-02) 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. The snapshot lacks a verified per-paper list (venues, types, authorship roles), which this axis would use to measure publication volume, composition, and productivity over time; only year-level work counts are currently on file.

Impact—

Not yet scored. Per-paper citation counts and uptake measures are absent from the snapshot, so this axis — which would gauge citation impact and scholarly influence — cannot be computed; only aggregate h-index 11 and i10-index 12 are on record.

Focus0/100

Scores 0/100 for focus, reflecting a perfectly even field split: 50% system identification and control theory / statistical learning for dynamical systems, 50% machine learning / Bayesian deep learning. Five topic clusters span fault detection and control systems (16 works), advanced control systems optimization (10), neural dynamics and brain function (8), probabilistic and robust engineering design (8), and motor control and adaptation (8), and the two reviewed papers fall one in each field.

Collaboration51/100

Scores 51/100. The OpenAlex record shows 12 distinct coauthors, and both Evolve-reviewed 2026 papers list Nakahira in a middle, non-corresponding position — an embedded contributor on team projects rather than a solo driver. Recurring partners in the snapshot include José M. F. Moura, Oren Wright, Paul Bogdan, Stephen Tu, Guannan Qu, and Xiaole Zhang.

Activity100/100

Scores 100/100. Recorded works per year rise steadily through the late 2010s (6 in 2018, 7 in 2019) and accelerate sharply after 2023: 15 works in 2024 and 16 in 2025 — the highest annual counts in the snapshot — with 12 already recorded in 2026 and 43 works over the most recent three years.

Reproducibility55/100

Scores 55/100, computed from two Evolve-reviewed papers (reproMean 55). Both are rated medium reproducibility readiness: the FO-GS estimator ships synthetic and EEG experiments whose real-world protocol has notable statistical weaknesses, and the Bayesian-smoothing paper is backed by sound theory but small-scale experiments with an untested distributional-target path.

Research summary

Yorie Nakahira is a researcher at Carnegie Mellon University whose OpenAlex-resolved record sits at the interface of control theory and machine learning: the snapshot splits the portfolio evenly (50/50) between system identification and control theory / statistical learning for dynamical systems, and machine learning / Bayesian deep learning. The two Evolve-reviewed papers make that bridge concrete. A 2026 paper on learning fractional-order dynamics from a single trajectory proposes FO-GS, a row-wise grid-search OLS estimator with the first non-asymptotic O(t^-1/2) guarantees for jointly recovering fractional order and system matrix; the other couples full-covariance moment propagation with Rauch–Tung–Striebel smoothing for online adaptation of Bayesian neural networks. Topic clusters reinforce the dual identity: fault detection and control systems (16 works) and advanced control systems optimization (10) on one side; neural dynamics and brain function (8), probabilistic and robust engineering design (8), and motor control and adaptation (8) on the other. Activity is the standout signal — scored 100/100, with recorded works rising from 9 (2023) to 15 (2024) and 16 (2025), and 43 works in the most recent three years. Aggregate indicators include h-index 11 and i10-index 12 across 12 distinct coauthors, with both reviewed papers in middle authorship. Output and impact dimensions remain unscored pending richer citation data.

Collaboration pattern

The OpenAlex record lists 12 distinct coauthors, and both Evolve-reviewed 2026 papers place Nakahira in a middle, non-corresponding authorship position — an embedded contributor on team projects rather than a solo driver. Recurring partners in the snapshot include José M. F. Moura, Oren Wright, Paul Bogdan, Stephen Tu, Guannan Qu, and Xiaole Zhang. The moderate collaboration score (51/100) points to a compact, repeat set of collaborators rather than a wide rotating network.

Research keywords

System identificationFractional-order systemsNon-Markovian dynamicsLong-range dependenceBayesian neural networksKalman smoothingState-space modelsMoment propagationOnline adaptationFault detection and control systemsProbabilistic robust engineering designNon-asymptotic estimation guarantees

Highlights

First non-asymptotic guarantees for single-trajectory fractional-order identification
The 2026 FO-GS paper proposes a row-wise grid-search OLS estimator that jointly recovers the fractional order and system matrix of fractional-order LTI systems from a single trajectory, with the first non-asymptotic O(t^-1/2) guarantees; Evolve rates the theory sound while flagging statistical weaknesses in the real-world EEG protocol.
Full-covariance smoothing brings Kalman-style inference to Bayesian neural networks
A 2026 paper couples full-covariance moment propagation with Rauch–Tung–Striebel smoothing for online adaptation of Bayesian neural networks through state-space models; Evolve judges the theory sound while noting small-scale experiments and an untested distributional-target path.
A two-front program linking dynamical-systems inference and probabilistic machine learning
The catalog snapshot shows an even 50/50 split between control/system identification and Bayesian deep learning, anchored by topic clusters in fault detection and control systems (16 works) and advanced control systems optimization (10), with the two reviewed papers sitting one on each side of the divide.

Timeline

2011Earliest indexed activity in the OpenAlex snapshot: 1 recorded work in 2011, followed by 4 in 2012.
2019Steady mid-period output: 7 recorded works in 2019, following 6 in 2018, across control and learning topics.
2024Recorded output accelerates to 15 works, up from 9 in 2023.
2025Highest annual count in the snapshot at 16 works; 43 works recorded over the most recent three years.
202612 works recorded to date; Evolve reviews two arXiv papers — single-trajectory fractional-order identification and full-covariance Bayesian smoothing — both with middle authorship.

Papers in the Evolve catalog

Frequent co-authors

Xiaole Zhang×1 Ziyi Zhang×1 Zehao Zhao×1 Stephen Tu×1 Guannan Qu×1 Paul Bogdan×1 Oren Wright×1 Haoming Jing×1 Qiaoan Shen×1 Koichiro Niinuma×1 Yorie Nakahira×1 José M. F. Moura×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 resolved from OpenAlex bibliographic data; all bibliometrics (h-index 11, i10-index 12, works per year, topics) come from that single snapshot.
Output and impact dimensions are unscored because per-paper venue, authorship, and citation data are not present in the snapshot.
Only two papers carry full Evolve reviews, both 2026 arXiv preprints with medium reproducibility readiness, so qualitative depth judgments are early-stage.
The supplied frequent-coauthor list included the researcher's own name, which was excluded from the collaboration analysis.