Machine learning researcher isolating mechanisms: noise optima in continual learning and state tracking in linear RNNs.
2
papers here
1
reviewed by Evolve
2
first-author
1
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-28) 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 scoreable yet. This axis would measure total publication volume and its growth over time, but the OpenAlex record has no reliable cumulative output count and its year-by-year data includes an implausible 1916 entry, so volume cannot be established.
Impact—
Not scoreable yet. Only an h-index of 1 and an i10-index of 0 are recorded — far too thin to characterize influence. This axis would measure citations, their trajectory, and uptake of the two 2026 papers.
Focus100/100
OpenAlex assigns this profile exactly one field — machine learning (deep learning / sequence models) — and both reviewed 2026 papers sit inside it: rehearsal-free continual learning and state tracking in Householder linear RNNs. Across both, the questions target mechanism isolation (dose-response curves, causal ablation), giving a maximal focus score of 100 on a narrow, self-consistent agenda.
Collaboration14/100
The record is near-solo: OpenAlex counts 1 distinct coauthor, and the frequent-coauthor list contains only the researcher's own name. Both reviewed 2026 papers are first-authored, one also corresponding. The collaboration score of 14 reflects this single-author pattern in the data, not any judgment of teamwork quality.
Activity100/100
Activity is the strongest axis: OpenAlex logs 51 works in the last three years, all dated 2026 in the year-by-year counts, and the Evolve catalog adds two first-authored arXiv papers in the same year. The score of 100 captures this concentrated burst, though with no earlier years on record the longer trajectory cannot yet be read.
Reproducibility100/100
Signals are positive but thin: the mean reproducibility-readiness score is 100 across 1 scored paper. Evolve rates the pre-registered Householder RNN ablation high, while the continual-learning paper is medium — well-isolated simulations, but MNIST-scale with code unreleased. Methodological discipline is evident; independent replication is not yet on record.
Research summary
Gunner Levi Howe is an early-stage machine learning researcher whose profile is resolved from public bibliographic data (OpenAlex, ORCID 0009-0003-8417-578X); no institution or country is on record. The Evolve catalog holds two 2026 arXiv papers, both first-authored — one also corresponding — and both in machine learning. The first, 'A Noise Optimum in Rehearsal-Free Continual Learning,' isolates an inverted-U dose-response between noise injection and rehearsal-free consolidation, attributing it to a σ²-coupled anchor gain; Evolve rates the simulations well-isolated but MNIST-scale, with code unreleased. The second, 'The Automaton Underneath,' uses a pre-registered ablation to show that deleting the additive input pathway flips shortcut collapse into exact state tracking in Householder linear RNNs, and is rated high for reproducibility readiness. The bibliometric record is extremely recent and compressed: 51 works logged over the last three years, all dated 2026, with an h-index of 1, an i10-index of 0, and 1 distinct coauthor — the frequent-coauthor list containing only the researcher's own name, indicating near-solo work. OpenAlex's field distribution lists a single field, machine learning (deep learning / sequence models), while the topic profile spans broader areas. The overall pattern is a mechanism-first agenda — dose-response curves, causal ablation, pre-registration — at the very beginning of its citation life.
Collaboration pattern
This is a near-solo record: OpenAlex counts 1 distinct coauthor, and the frequent-coauthor list contains only the researcher's own name, indicating single-author work. Both reviewed 2026 papers are first-authored, one also corresponding. The 14/100 collaboration score reflects this independent mode; no lab, group, or co-author network is visible in the data yet.
Research keywords
Highlights
An inverted-U noise optimum in rehearsal-free continual learning
Deleting the additive input pathway restores exact state tracking in Householder RNNs
A mechanism-first toolkit spanning both 2026 studies
Timeline
- 2026Posted the first-authored, corresponding arXiv paper 'A Noise Optimum in Rehearsal-Free Continual Learning: Isolation, Mechanism, and Scope' on continual learning.
- 2026Posted the first-authored arXiv paper 'The Automaton Underneath' on state tracking in Householder linear RNNs, rated high for reproducibility readiness.
- 2026Both reviewed papers land in a single year — OpenAlex logs 51 works in 2026 — making it the burst year that defines this early-stage profile.
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.
- All bibliometrics come from the OpenAlex record only (identity: openalex); there is no cross-check against other citation databases.
- No institution or country is on record; the profile is name-resolved from public bibliographic data tied to ORCID 0009-0003-8417-578X.
- The OpenAlex year-by-year counts include an implausible 1916 entry (1 work), and the topic list spans areas outside the reviewed ML papers (e.g., ferroelectric and negative capacitance devices, 10 works) — possible record noise or unreviewed output.
- Output and impact axes are unscored: citation uptake of the two 2026 arXiv papers is not yet observable, and an h-index of 1 with i10-index of 0 is too thin to rank.