Zerrin Dağlı

Mersin Üniversitesi · Türkiye

Unclaimed · AI-structured profileORCID OpenAlex

Mersin University researcher proposing Mandelbrot-derived zero-storage neural networks and edge-AI triage engines

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-08) 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 not yet scoreable: the OpenAlex record lists only h-index 2 and i10-index 0 alongside 21 works in the last three years, so no reliable total-output or venue-quality measure is available; this axis would capture publication volume and consistency.

Impact—

Impact cannot be scored yet: with an h-index of 2 and i10-index of 0 in the OpenAlex record, there are too few citations to ground a citation-based measure; this axis would capture citation uptake and scholarly influence of the work.

Focus0/100

Radar assigns focus 0, the lowest reading: field distribution splits exactly 50/50 between edge AI/efficient inference systems and machine learning (neural network parameterization)/nonlinear dynamics, and indexed topics spread across five clusters — led by Advanced Memory and Neural Computing (10) and Neural Networks and Applications (9) — with no dominant single direction.

Collaboration36/100

Collaboration scores 36 on the radar, consistent with a small co-author pool: the OpenAlex record lists 5 distinct co-authors over the recent window, and the Evolve-reviewed papers recur within one surname cluster — Volkan Dağlı and Dağhan Dağlı — rather than an expanding external network; both entries are middle-author, non-corresponding.

Activity90/100

Activity is the strongest axis at 90: countsByYear shows single works in 2019, 2021, 2022 and 2025, then a sharp acceleration to 20 works dated 2026, with 21 works over the last three years — a publication rate concentrated almost entirely in the current year.

Reproducibility55/100

Reproducibility scores 55, a mean computed across 2 Evolve-reviewed papers, both rated medium reproducibility readiness. The reviews temper this: headline mechanisms in the zero-storage paper are untested beyond one Two-Moons toy test, and the triage paper's 93.1% figure risks in-sample seed tuning — reproducible in outline, not yet at headline-claim level.

Research summary

Zerrin Dağlı is a researcher at Mersin Üniversitesi (Mersin University, Turkey), ORCID 0000-0001-9490-6425, whose profile — name-resolved from OpenAlex bibliographic data — centers on an unconventional program: replacing stored neural-network weights with procedurally generated, Mandelbrot-set-derived parameters. Two 2026 arXiv preprints carry this line. The first proposes zero-storage neural synthesis framed by self-organized criticality; Evolve's review finds the evidence limited to a single Two-Moons toy test against a 16-byte baseline, with headline mechanisms untested. The second applies Mandelbrot escape dynamics to a weight-free edge-triage engine, claiming 93.1% decision-map accuracy and 0% prompt-injection bypass; Evolve flags possible in-sample seed tuning and a security claim covering only 10 attack classes. Bibliometrically, the OpenAlex record reports an h-index of 2 and i10-index of 0 against 21 works in the last three years — 20 of them dated 2026 — with 5 distinct co-authors, led topically by Advanced Memory and Neural Computing (10) and Neural Networks and Applications (9). Field distribution splits evenly between edge AI/efficient inference systems and machine learning (neural network parameterization)/nonlinear dynamics. Radar scores: activity 90, reproducibility 55 (mean across 2 reviewed papers, both medium readiness), collaboration 36, focus 0; output and impact axes lack sufficient data.

Collaboration pattern

The record shows a compact, same-surname collaboration pattern: 5 distinct co-authors in the OpenAlex recent window, dominated by Volkan Dağlı and Dağhan Dağlı, who recur across the profile's topics. Both Evolve-reviewed papers list middle authorship with no corresponding role, and the researcher's own name appears among the frequent co-author strings — a pattern that can indicate family co-publication or entity-resolution overlap in OpenAlex rather than a broad external network.

Research keywords

zero-storage neural networksMandelbrot setprocedural weight generationparameter compressionedge of chaosself-organized criticalityzero-tensor inferenceMandelbrot escape dynamicsedge triageprompt injection resistancenonlinear dynamicsefficient inference

Highlights

Zero-storage neural synthesis from Mandelbrot-set seeds
A 2026 arXiv preprint proposes generating network weights procedurally from Mandelbrot-set seeds, framed with self-organized criticality and edge-of-chaos concepts; Evolve's review finds the evidence rests on a single Two-Moons test against a 16-byte baseline, leaving headline mechanisms untested.
Weight-free Mandelbrot-boundary engine for edge triage
A second 2026 preprint applies Mandelbrot escape dynamics to real-time edge triage across heterogeneous domains, claiming 93.1% decision-map accuracy and 0% prompt-injection bypass; Evolve notes the headline figure risks in-sample seed tuning and the security claim covers only 10 attack classes.
Sharp 2026 output surge in a young OpenAlex record
The OpenAlex record shows single works in 2019, 2021, 2022 and 2025, then 20 works dated 2026 (21 over the last three years) against an h-index of 2 — an unusually steep ramp concentrated in one year that shapes the entire profile.

Timeline

2019First work appears in the OpenAlex record (1 work indexed), the start of a sparse early trail.
2021One indexed work, keeping the record's pre-2025 output at single-work years.
2022A single indexed work continues the low-rate period before the 2026 surge.
2025One indexed work marks the last pre-surge year in the record.
202620 works dated 2026 appear in the record, including both Evolve-reviewed arXiv preprints on Mandelbrot-based zero-storage networks and edge triage.

Papers in the Evolve catalog

Frequent co-authors

Volkan Dağlı×2 Dağhan Dağlı×2 Volkan Dağlı×1 Zerrin Dağlı×1 Dağhan Dağlı×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 come from an OpenAlex record only (h-index 2, i10-index 0, countsByYear, topics); no Scopus, Web of Science, or citation-level data is included.
The frequent co-author list contains the researcher's own name and repeats one surname, so the OpenAlex entity may conflate records; this profile is name-resolved from public bibliographic data.
Only 2 papers carry full Evolve reviews (both 2026 arXiv preprints, middle authorship); the remaining works in the 2026 spike are unreviewed and cannot be verified from this input.
Output and impact radar axes are missing due to insufficient data; no venue quality, affiliation history, or funding information was provided.