Graph and topological signal processing researcher at TU Delft; theory-driven first-author work, high reproducibility
2
papers here
2
reviewed by Evolve
2
first-author
0
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—
No score yet: the snapshot lacks publication-volume data. This axis would measure total paper count, venue mix, and author-position share once a fuller bibliometric record is available.
Impact—
Not scored: citation metrics are absent from this snapshot. This axis would track citations, h-index, and downstream uptake; both catalogued papers are recent 2026 preprints, so citation evidence has not yet accumulated.
Focus0/100
Scores 0/100, consistent with an even field split: the distribution is 50% 'graph signal processing and machine learning for time series' and 50% 'topological signal processing / graph machine learning'. The two catalogued papers sit one in each — Kronecker-covariance forecasting architectures versus unoriented simplicial operators — indicating breadth across adjacent graph-structured directions rather than concentration in a single niche.
Collaboration39/100
Scores 39/100 with a compact network: 6 distinct coauthors recorded. Both catalogued papers are first-author works, and the recurring core includes Elvin Isufi, Athanasios Georgoutsos, Varun Sarathchandran and Geert Leus. The pattern suggests a small, stable circle around two 2026 projects rather than a broad distributed network; neither paper lists this author as corresponding.
Activity—
Not scored: cadence data such as per-year output and recency-adjusted volume is insufficient in this snapshot. The axis would measure publishing rhythm and momentum; only two 2026 papers are visible here, which cannot establish a trend.
Reproducibility100/100
Scores 100/100, the profile's strongest axis: both reviewed papers (reproPapers = 2) reach a reproducibility mean of 100, and each carries a 'high' reproducibility rating from Evolve's review. The Kronecker-covariance forecasting model and the unoriented topological framework are both judged straightforward to reproduce — a consistent, verifiable working style across the record.
Research summary
Andrea Cavallo is a researcher at Delft University of Technology whose Evolve profile, name-resolved from public bibliographic data under ORCID 0009-0006-1226-3926, centres on graph-based signal processing and machine learning. The catalogued record contains two 2026 first-author arXiv papers splitting the field distribution evenly (50% each): 'Spatiotemporal Kronecker Covariance Neural Networks' and 'Topological Signal Processing With Unoriented Operators'. The first proposes KVNN, which decouples spatiotemporal covariance through Kronecker sums; Evolve's review finds the theory sound and forecasting parameter-efficient, while noting dated baselines and modest margins. The second introduces an unoriented topological signal processing framework with a proven interaction-order decomposition, whose regularizers win on four small real datasets. A distinguishing trait is verifiability: both papers are rated high for reproducibility, giving a reproducibility mean of 100 across the two reviewed works — the strongest scored axis on this radar. Collaboration is comparatively compact, with six distinct coauthors and a recurring core including Elvin Isufi, Athanasios Georgoutsos, Varun Sarathchandran and Geert Leus. Output, impact and activity axes lack sufficient underlying data and are left unscored; readers should treat this as an early-stage profile built from a small, recent record.
Collaboration pattern
The collaboration footprint is compact: 6 distinct coauthors, with a recurring core of Elvin Isufi, Geert Leus, Athanasios Georgoutsos and Varun Sarathchandran. Both 2026 catalogued papers are first-author (non-corresponding) works, consistent with an early-stage researcher embedded in a small, stable signal-processing group rather than leading large multi-team consortia. The even 50/50 field distribution suggests the small team spans both the forecasting and the simplicial-operator directions.
Research keywords
Highlights
KVNN: decoupling spatiotemporal covariance with Kronecker sums
Unoriented operators bring provable structure to topological signal processing
Perfect reproducibility across the reviewed record
Both catalogued papers earn a 'high' reproducibility rating, giving a reproducibility mean of 100 over 2 reviewed papers — the strongest scored axis on this researcher's radar.
Timeline
- 2026First-author arXiv preprint 'Spatiotemporal Kronecker Covariance Neural Networks' appears, proposing KVNN for parameter-efficient spatiotemporal forecasting.
- 2026First-author arXiv preprint 'Topological Signal Processing With Unoriented Operators' appears, with a proven interaction-order decomposition for simplicial data.
- 2026Both 2026 papers enter the Evolve catalog with 'high' reproducibility ratings, lifting the reproducibility mean to 100 across 2 reviewed papers.
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.
- Radar scores derive from a public bibliometric snapshot; the output, impact and activity axes lack sufficient underlying data and are unscored.
- Both catalogued papers are 2026 arXiv preprints; no peer-reviewed venue record appears in this snapshot.
- Country-level affiliation is not recorded; the profile carries institution-level attribution (Delft University of Technology) only.
- The frequent-coauthor input contains duplicates and the author's own name; this profile uses the deduplicated list of four recurring coauthors.