Dağhan Dağlı

Toros University · Türkiye

Unclaimed · AI-structured profileORCID OpenAlex

Zero-storage fractal neural weights and edge-AI triage: high-activity, early-record researcher at Toros University

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—

Insufficient data to score. This axis would measure cumulative publication volume, venue mix, and year-over-year consistency; the OpenAlex record for this profile is too recent and sparse for a defensible estimate, so it is left unscored rather than guessed.

Impact—

Insufficient data to score. Impact would measure citation uptake, h-index trajectory, and influence on downstream work; the observed h-index of 2 with an i10-index of 0 reflects an early record too thin to support any defensible impact score.

Focus0/100

Focus scores 0/100. The field distribution splits evenly — 0.5 for edge AI / efficient inference systems and 0.5 for machine learning (neural network parameterization) / nonlinear dynamics — with no dominant field. OpenAlex topic counts span five clusters (Advanced Memory and Neural Computing 10, Neural Networks and Applications 9, Complex Systems and Dynamics 5, Advanced Neural Network Applications 4, Artificial Intelligence Applications 4), confirming breadth without a single concentrated thread.

Collaboration36/100

The collaboration score is 36/100 with 5 distinct coauthors on record. The frequent-coauthor list is dominated by same-surname collaborators (Volkan Dağlı, Zerrin Dağlı) and even repeats the profile name itself, suggesting a very small, tightly-knit — possibly internal or family-based — author cluster rather than broad external networks. Both cataloged 2026 preprints place Dağlı in the last author position without a corresponding-author role.

Activity89/100

Activity scores 89/100, the profile's strongest axis. OpenAlex countsByYear shows a single work in 2020 and a jump to 20 works in 2026, with recent3yWorks also at 20 — essentially all recorded output is concentrated in the current cycle. This indicates a highly active, bursty publishing cadence, currently expressed through arXiv preprints rather than established venues.

Reproducibility55/100

Reproducibility scores 55/100, computed as a mean over 2 reviewed papers (reproPapers: 2). Both cataloged preprints carry a medium repro-readiness rating: procedural descriptions exist in principle, but Evolve's verdicts note headline mechanisms remain untested beyond toy evidence — a Two-Moons comparison against a 16-byte baseline, and a 93.1% headline figure at risk of in-sample seed tuning.

Research summary

Dağhan Dağlı is a researcher at Toros University (TR) whose OpenAlex-indexed profile centers on fractal parameterization of neural networks and zero-storage, edge-AI inference. The bibliometric record is very recent: it lists an h-index of 2 and an i10-index of 0, one work recorded in 2020, and a jump to 20 works in 2026 (recent3yWorks: 20). OpenAlex topic tags concentrate on Advanced Memory and Neural Computing (10), Neural Networks and Applications (9), and Complex Systems and Dynamics (5), while the field distribution splits evenly, 0.5/0.5, between edge AI / efficient inference systems and machine learning (neural network parameterization) / nonlinear dynamics. Evolve has reviewed two 2026 arXiv preprints, both with Dağlı in the last author position and neither corresponding: a weight-free Mandelbrot-boundary engine for real-time edge triage, and a proposal for Mandelbrot-seed zero-storage neural weights. Evolve's verdicts flag overreach in both cases — headline mechanisms untested beyond a Two-Moons toy test against a 16-byte baseline, and a headline 93.1% figure at risk of in-sample seed tuning — while rating replication readiness as medium (reproducibility mean 55 across 2 papers). Activity (89/100) is the strongest axis; collaboration is thin (36/100) with 5 distinct coauthors, dominated by same-surname collaborators.

Collaboration pattern

The coauthor footprint is small: 5 distinct coauthors on record. The frequent-coauthor list is dominated by same-surname collaborators — Volkan Dağlı and Zerrin Dağlı — and even includes Dağhan Dağlı itself, an artifact suggesting a tightly-knit internal or family-based author cluster and warranting identity-resolution caution. Both cataloged 2026 preprints place Dağlı last among the authors without a corresponding-author role, consistent with a contributing rather than leading position within this cluster.

Research keywords

zero-storage neural networksprocedural weight generationMandelbrot-set computationparameter compressionedge AIefficient inference systemsreal-time edge triageprompt injection resistanceself-organized criticalityedge-of-chaos dynamics

Highlights

Weight-free Mandelbrot-boundary engine for real-time edge triage
The 2026 arXiv preprint Universal Fractal Natural Language Decision Map proposes a zero-tensor, weight-free triage engine built on Mandelbrot escape dynamics, targeting real-time decisions across heterogeneous domains at the edge. Evolve's review cautions that the headline 93.1% figure risks in-sample seed tuning and that the 0% prompt-injection bypass claim rests on only 10 attack classes; repro-readiness is rated medium.
Mandelbrot-seed zero-storage neural weights — overreach flagged
The 2026 preprint Orbital Error Dynamics claims neural weights synthesized procedurally from Mandelbrot-set seeds, eliminating stored parameters altogether. Evolve finds the supporting evidence limited to a Two-Moons toy test against a 16-byte baseline, with headline mechanisms — self-organized criticality and ephemeral parameter resonance — untested; the claims are judged to overreach, with medium repro-readiness.
A burst 2026 agenda linking fractal parameterization to efficient inference
OpenAlex records 20 works in 2026 versus 1 in 2020, with topic counts spanning five clusters and the field distribution split 0.5/0.5 between edge AI / efficient inference and ML parameterization / nonlinear dynamics. The result is a fast-moving but not yet concentrated research agenda; both reviewed papers place Dağlı last among the authors.

Timeline

2020Earliest recorded output in the OpenAlex record: countsByYear logs 1 work in 2020.
2026Publishes Universal Fractal Natural Language Decision Map on arXiv, proposing a weight-free Mandelbrot-boundary edge-triage engine; Dağlı is last author, non-corresponding.
2026Publishes Orbital Error Dynamics on arXiv, claiming Mandelbrot-seed zero-storage neural weight synthesis; again last author, non-corresponding.
2026Output peaks: OpenAlex countsByYear logs 20 works in 2026 (recent3yWorks: 20), against an h-index of 2 and i10-index of 0.

Papers in the Evolve catalog

Frequent co-authors

Volkan Dağlı×2 Zerrin 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.

Bibliometric figures are drawn from OpenAlex only (identity: openalex) and have not been cross-checked against other bibliographic databases.
Output and impact dimensions are unscored due to insufficient data; the observed h-index of 2 and i10-index of 0 reflect a very early record.
Evolve has reviewed 2 of the works recorded for 2026 (20 works logged); the verdicts apply only to those two arXiv preprints.
Frequent coauthors share the surname Dağlı and the list repeats the profile name itself; this profile is name-resolved from public bibliographic data (ORCID 0009-0003-2492-8313), so same-name conflation is possible.