Streaming-ML researcher bridging anomaly detection and tabular foundation models for adaptive edge-cloud data streams.
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
5
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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—
The output axis measures cumulative publication volume and career trajectory. It is unscored here because the underlying work-level records are incomplete in this profile's data; only partial signals (15 works in the last three years, 8 in 2026) are on file, so no score is given.
Impact—
The impact axis measures citation-based influence: citations received, field-weighted exposure and uptake by later work. Scoring data is not yet available here; the only signals on file are an h-index of 5 and an i10-index of 2, so no score is given.
Focus0/100
The focus score is 0 because output splits exactly evenly (0.5/0.5) between two ML subfields: streaming data mining/anomaly detection and stream learning with tabular deep learning. Topic counts reinforce the spread — data stream mining (11), machine fault diagnosis (10), railway engineering and dynamics (6), fault detection and control systems (6), IoT and edge/fog computing (4) — an application-spanning rather than single-track profile.
Collaboration36/100
Collaboration scores 36 on the strength of five distinct coauthors recorded in OpenAlex. The recurring circle is compact and stable — Goreti Marreiros, Diogo Risca, Ricardo Martins, Vitor Crista and Diogo Martinho — suggesting a small core team at Universidade Técnica do Porto rather than a broad external network. In both Evolve-reviewed 2026 preprints the researcher holds a middle, non-corresponding position, consistent with contributing inside a group.
Activity81/100
Activity is the strongest axis at 81. OpenAlex year counts climb from 2 works in 2022 and 4 in 2023, dip to 2 in 2024, then accelerate to 5 in 2025 and 8 in 2026 — the highest year on file — for 15 works across the last three years. Both Evolve-reviewed arXiv preprints landed in 2026, matching the surge.
Reproducibility38/100
Reproducibility averages 38 across the two papers Evolve has scored. The hierarchical TabPFN streaming paper rates medium — six streams with consistent error/communication trade-offs, though Evolve flags untested statistics and latency. The collaborative anomaly detection paper rates low: the evidence rests on one loom and three episodes with no baselines. The pattern suggests working systems whose evaluations are not yet benchmark-grade.
Research summary
Afonso Lourenço is a machine learning researcher at Universidade Técnica do Porto (Portugal) working at the intersection of streaming data mining and tabular deep learning. The OpenAlex-resolved profile splits evenly (0.5/0.5) between two strands: streaming anomaly detection and stream learning with tabular foundation models. Topic counts show a broad applied footprint — data stream mining (11 works), machine fault diagnosis (10), railway engineering and dynamics (6), fault detection and control systems (6), and IoT and edge/fog computing (4). Activity is the strongest axis (81/100): recorded works rose from 2 in 2022 and 4 in 2023, dipped to 2 in 2024, then accelerated to 5 in 2025 and 8 in 2026 — the highest year on file — with 15 works across the last three years. Two 2026 arXiv preprints have been reviewed by Evolve: a collaborative streaming anomaly detection system combining four detectors with rule-based surrogates and interactive human refinement, and a hierarchical kNN-gated TabPFN offloading scheme showing consistent error/communication trade-offs across six streams. Citation signals remain early-stage (h-index 5, i10-index 2), and the output and impact axes are unscored for lack of data. Collaboration is modest (36/100, five distinct coauthors), and Evolve's reproducibility scoring across the two reviewed papers averages 38, with readiness rated low and medium respectively.
Collaboration pattern
Collaboration is compact: five distinct coauthors are on record, with Goreti Marreiros, Diogo Risca, Ricardo Martins, Vitor Crista and Diogo Martinho recurring most often — a stable in-group cluster around Universidade Técnica do Porto. Both Evolve-reviewed 2026 preprints list the researcher as a middle author without corresponding status, indicating shared, group-driven projects rather than solo-led ones.
Research keywords
Highlights
Hierarchical TabPFN inference moves tabular foundation models onto data streams
Human-in-the-loop streaming anomaly detection with ensemble consensus
A two-strand research identity backed by a 2026 output surge
The profile splits evenly between streaming anomaly detection and tabular stream learning (0.5/0.5 field distribution), and recorded works hit 8 in 2026 — the highest year in the OpenAlex window — within a broader applied span covering fault diagnosis, railway dynamics and edge/fog computing.
Timeline
- 2022Earliest year in the OpenAlex output window: 2 recorded works.
- 20234 works recorded, an early step up in publication cadence.
- 2024A dip to 2 works before the late acceleration.
- 20255 works recorded — output picks up again and the recent three-year run begins.
- 20268 works, the highest year on file, including two arXiv preprints reviewed by Evolve on streaming anomaly detection and tabular foundation models.
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.
- Bibliometrics are computed from OpenAlex records only (identity: openalex; ORCID 0000-0002-3465-3419); other bibliographic databases are not aggregated.
- The output and impact radar axes are unscored because the underlying data is insufficient; the quoted citation signals (h-index 5, i10-index 2) are the only impact figures on file and likely lag the 2026 preprints.
- Paper-level evidence covers only two 2026 arXiv preprints with middle-author, non-corresponding positions; first-author or corresponding-author work is not represented in the reviewed set.
- The profile is name-resolved from public bibliographic data and remains early-stage with thin citation coverage, so conclusions may shift as records accumulate.