Deep learning researcher at Seoul National University: weak supervision for wafer defects and efficient GNN inference
3
papers here
0
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
1
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-17) 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—
This axis would measure total publication volume and productivity trajectory, but the current OpenAlex snapshot does not yield a reliable output score. Only nine works over the last three years are recorded, too short a record to normalize against field baselines.
Impact—
Impact would capture citation-based influence such as h-index and i10-index, yet the snapshot reports h-index 1 and i10-index 0 with no citation aggregates. This is typical for an early-career profile, so no score is assigned here.
Focus—
Focus would quantify how concentrated the research agenda is across topics. Five topic clusters appear in the snapshot, led by Advanced Neural Network Applications (7), but no normalized concentration score is available yet, so this axis is left unscored.
Collaboration53/100
A collaboration score of 53/100 reflects moderate network breadth: the snapshot records 13 distinct coauthors across the nine works of the last three years. Frequent partners include Kanghyun Choi, Sunjong Park, Jinho Lee, and Hyeyoon Lee. Authorship roles show first-author leadership on the two 2026 first-position papers and a middle position on STARLINC, suggesting work within small, project-based teams rather than large consortia.
Activity67/100
An activity score of 67/100 is grounded in a steady ramp-up: countsByYear shows 1 work in 2024, 5 in 2025, and 3 in 2026, for 9 works in the recent three-year window. All three Evolve-cataloged papers appeared in 2026, two as first author, indicating a phase of intense submission activity atop an early, rapidly growing record.
Reproducibility—
Reproducibility would assess whether released papers ship code, data, and sufficient implementation detail for replication. The three cataloged arXiv papers carry no Evolve repro-readiness verdicts yet, so there is no evidence base to score this axis.
Research summary
Dain Kwon is a researcher at Seoul National University whose work centers on applied deep learning: making neural networks practical in industrial and scientific imaging settings. The OpenAlex-linked snapshot (identity: openalex) records 9 works over the recent three-year window — 1 in 2024, 5 in 2025, and 3 in 2026 — with an h-index of 1 and an i10-index of 0, consistent with an early-career profile scaling up quickly. Topic modeling attributes 7 works to Advanced Neural Network Applications and 5 to Domain Adaptation and Few-Shot Learning, with smaller clusters in numerical methods, data classification, and VLSI design. Three 2026 arXiv papers appear in the Evolve catalog. Two are first-authored: SePArate, which segments wafer defect patterns using weak supervision, and TopGQ, a topology-aware post-training quantization method for graph neural networks. A third, STARLINC, removes satellite trail artifacts via inter-frame correlation, with a middle authorship position. Together they span industrial inspection, efficient graph learning, and astronomical image restoration. Collaboration data shows 13 distinct coauthors in three years (collaboration score 53/100), including recurring partners such as Kanghyun Choi, Sunjong Park, Jinho Lee, and Hyeyoon Lee. Impact, output, focus, and reproducibility dimensions remain unscored pending more evidence; this should be read as an early-stage, name-resolved profile built from public bibliographic data.
Collaboration pattern
In three years, Dain Kwon collaborated with 13 distinct coauthors (collaboration score 53/100). The frequent-coauthor list includes Kanghyun Choi, Sunjong Park, Jinho Lee, Hyeyoon Lee, Shingeon Kim, and Mi-Ryang Kim, among others. Authorship roles point to small, project-based teams: first-author leadership on SePArate and TopGQ and a middle-author, supporting position on STARLINC, rather than large-consortium authorship.
Research keywords
Highlights
First-authored SePArate: weak-supervision wafer defect segmentation
As first author, Dain Kwon's 2026 arXiv paper SePArate segments defect patterns from wafer manufacturing images using weak supervision, reducing reliance on manually labeled industrial inspection data.
First-authored TopGQ: topology-aware GNN quantization
TopGQ (2026, first author) proposes fast post-training quantization for graph neural networks that leverages topology information, targeting efficient deployment of graph learning models.
Co-authored STARLINC: satellite trail artifact removal
On the 2026 arXiv paper STARLINC, Dain Kwon contributes as a middle-position author to a method that removes satellite trail artifacts using inter-frame correlation, extending applied deep learning into astronomical image restoration.
Timeline
- 2024First visible publications appear in the OpenAlex record, with 1 indexed work in 2024 marking the start of the publication trail.
- 2025Most productive year in the snapshot, with 5 indexed works — a fivefold step-up from 2024.
- 2026Three arXiv preprints reach the Evolve catalog — SePArate and TopGQ as first author, STARLINC as middle author — alongside 3 indexed works in total for the year.
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 indicators (h-index 1, i10-index 0, countsByYear, topics, 13 distinct coauthors) are derived from OpenAlex only; other databases were not consulted.
- All three cataloged papers are 2026 arXiv preprints; no peer-reviewed venue, field label, or Evolve review verdict is recorded yet.
- The output, impact, focus, and reproducibility dimensions are unscored because the underlying evidence (citation aggregates, normalized output, concentration scores, and repro-readiness reviews) is insufficient at snapshot time.
- This profile is name-resolved from public bibliographic data; no biographical details, awards, or career history beyond the snapshot are included.