Early-career ML researcher extending Bayesian networks into federated learning, causal discovery and anomaly detection
5
papers here
5
reviewed by Evolve
4
first-author
0
corresponding
—
works · OpenAlex
3
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—
Not yet scored: publication-volume data is insufficient in this snapshot. This axis would measure total research production — full publication counts, venue types and record completeness — beyond the five reviewed preprints currently on file.
Impact—
Not yet scored: citation-based evidence (citation counts, percentiles, field-weighted influence) has not been ingested. This axis would measure uptake and influence of the work; the current h-index 3 and i10-index 1 are too early-stage to represent it.
Focus0/100
Focus scores 0 because output is spread perfectly evenly: the OpenAlex field distribution assigns 20% to each of five clusters — federated/Bayesian classifiers, causal discovery with federated learning, spacecraft-telemetry anomaly detection, and two probabilistic-graphical-model/evolutionary-computation clusters. The spread is nonetheless methodologically coherent: nearly all five reviewed papers build on Bayesian-network methodology applied to different deployment settings, rather than drifting across unrelated topics.
Collaboration50/100
Collaboration scores 50, matching a moderate footprint: 11 distinct coauthors over the last three years. The recurring core sits at Castilla-La Mancha — José A. Gámez and José M. Puerta appear most often — with Juan A. Aledo, Juan C. Alfaro, Fabio Stella and Óscar Alcarria also recurring. Four of five reviewed papers are first-authored and none lists him as corresponding author, consistent with a junior researcher embedded in an established Bayesian-networks group.
Activity79/100
Activity scores 79 on a clear upward trajectory: OpenAlex counts 2 works in 2023, 5 in 2024, 4 in 2025 and 5 in 2026, totalling 14 works over the last three years. The pipeline is dense in 2026 — five arXiv preprints from that single year entered Evolve's catalog — indicating a strongly active current research phase.
Reproducibility82/100
Reproducibility scores 82, an average across the five reviewed papers. Two — the genetic-algorithm Bayesian-network fusion papers — are rated high readiness, with code released for the structural-fusion work; FedRCD also rates high. FedAnDE and the spacecraft-telemetry paper rate medium. Evolve's verdicts still flag gaps: no significance testing or external baselines in the fusion work, and privacy claims in FedAnDE lacking formal backing — an honest, improvable position.
Research summary
Pablo Torrijos is a machine learning researcher at the University of Castilla-La Mancha (Spain) working primarily on probabilistic graphical models — Bayesian networks and averaged n-dependence estimators — extended into federated learning, causal discovery, and applied anomaly detection. Evolve has reviewed five of his 2026 arXiv preprints, four as first author (none as corresponding author). Two papers develop tractable Bayesian-network fusion under treewidth limits using pre-fusion edge pruning and genetic algorithms, with code released and high reproduction-readiness ratings. Others introduce FedAnDE, described as the first federated framework for discriminative AnDE classifiers, and FedRCD, which federates LiNGAM causal discovery through fourth-order cumulants. A collaborative paper applies label-free anomaly detection with extreme-value-theory thresholds to spacecraft telemetry. OpenAlex metrics show a young but accelerating record: h-index 3, i10-index 1, 14 works in the last three years, with yearly outputs of 2 (2023), 5 (2024), 4 (2025) and 5 (2026). His topic profile is led by Bayesian modeling and causal inference (12 topic assignments), with privacy-preserving technologies second (7). The reproducibility axis scores 82 across the five reviewed papers. Output and impact dimensions are not yet scored due to insufficient data.
Collaboration pattern
Collaboration centers on a Castilla-La Mancha Bayesian-networks cluster: José A. Gámez and José M. Puerta recur most frequently in the co-author list, alongside Juan A. Aledo, Juan C. Alfaro, Fabio Stella, Óscar Alcarria, Rafael Sánchez, Javier Sempere and Juan M. Auñón — 11 distinct coauthors over three years. He leads as first author on 4 of 5 reviewed papers but appears on none as corresponding author, suggesting senior colleagues anchor the projects.
Research keywords
Highlights
Tractable consensus fusion of Bayesian networks via genetic algorithms
Structural fusion of Bayesian networks under treewidth limits
First federated framework for discriminative AnDE classifiers
Federated causal discovery exposes a cumulant caveat
Label-free anomaly detection for spacecraft telemetry
Timeline
- 2026Proposes FedRCD, federating LiNGAM causal discovery via fourth-order cumulants (first author).
- 2026Introduces consensus Bayesian-network fusion via pre-fusion edge pruning with genetic algorithms (first author).
- 2026Co-authors structure-aware unsupervised anomaly detection for spacecraft telemetry with adaptive EVT thresholds (middle author).
- 2026Publishes structural fusion of Bayesian networks with limited treewidth using genetic algorithms, with code released (first author).
- 2026Introduces FedAnDE, the first federated framework for discriminative AnDE classifiers (first author).
Papers in the Evolve catalog
- first author
Federated Learning of AnDE Classifiers
- first author
Structural Fusion of Bayesian Networks with Limited Treewidth Using Genetic Algorithms
Structure-Aware Unsupervised Anomaly Detection for Spacecraft Telemetry with Adaptive EVT Thresholding
- first author
Genetic Algorithms for Tractable Bayesian Network Fusion via Pre-Fusion Edge Pruning
- first author
Federated Causal Discovery via Regression-Directed Cumulants
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 figures (h-index 3, i10-index 1, per-year counts) come from a single OpenAlex identity snapshot; no other databases are aggregated.
- Output and impact dimensions are unscored because citation and publication-volume data in the snapshot were insufficient for reliable estimation.
- All five catalog papers are 2026 arXiv preprints; peer-review status and final publication venues are not reflected.
- This profile is name-resolved from public bibliographic data for 'Pablo Torrijos' (ORCID 0000-0002-8395-3848); homonymous works may be missing or wrongly attributed.