Software testing, model checking and neuro-symbolic learning for trustworthy AI at Simula Research Laboratory
3
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
8
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—
Not yet scored. This axis would measure total publication volume and venue quality across the full career, but the underlying data in this snapshot is insufficient; only per-year OpenAlex counts and the three Evolve-reviewed papers are available, so no output score is given.
Impact—
Not yet scored. Impact would capture citation uptake, normalized citation influence and adoption of methods by other groups; the available raw footprint (h-index 8, i10-index 6) is early-stage, so the platform leaves this axis unscored rather than estimating it.
Focus100/100
Scored 100 on focus. OpenAlex topic assignments concentrate on software testing and debugging (32 works), explainable AI (23), formal methods in verification (19), reinforcement learning in robotics (18) and adversarial robustness (16) — one coherent programme on testing, verifying and explaining learned systems, with the field distribution listing a single field.
Collaboration32/100
Scored 32 on collaboration. The OpenAlex window records 60 works in the last three years but only 4 distinct coauthors, a dense small-team pattern rather than a broad network; recurring partners are Dennis Gross, Arnaud Gotlieb, Nassim Belmecheri and Nadjib Lazaar, across the constraint-acquisition and verification lines of work.
Activity100/100
Scored 100 on activity. Recorded works climbed from 2 in 2015 and 9 in 2021 to 18 in 2024 and 28 in 2025 — the highest in the series — with 14 already recorded in 2026 and 60 works across the trailing three years, indicating a sustained and sharply accelerating publication pace.
Reproducibility100/100
Scored 100 on reproducibility, with a mean readiness of 100 across assessed papers. The base is narrow: only 1 paper has received a reproducibility assessment so far, so the perfect score should be read as one strong data point on a well-documented 2026 paper, not a long-term track record.
Research summary
Helge Spieker is a researcher at Simula Research Laboratory in Norway whose work sits at the junction of software-engineering rigor and modern machine learning. Profiled from OpenAlex bibliographic data and three 2026 papers in the Evolve catalog, his research clusters around software testing and debugging (32 topic-assigned works), explainable AI (23), formal verification methods (19), reinforcement learning in robotics (18), and adversarial robustness (16). In 2026 he first-authored 'Probabilistic Model Checking of Autoregressive Neural Sequence Models', converting autoregressive generators into PRISM-checkable DTMCs with certified under-approximation bounds, and last-authored a neuro-symbolic approach that learns symbolic constraint representations, where a Transformer oracle plus FastCA automate human-free constraint acquisition. A third 2026 paper proposes automated testing of LLM-based post hoc explainers using model checking as an oracle. His recorded output has climbed from 2 works in 2015 to 18 in 2024 and 28 in 2025 — 60 works in the trailing three years — while his citation footprint (h-index 8, i10-index 6) remains early-stage. Recent collaboration is compact: 4 distinct coauthors, including Dennis Gross, Arnaud Gotlieb, Nassim Belmecheri and Nadjib Lazaar.
Collaboration pattern
Recent collaboration is compact: OpenAlex lists 60 works in the last three years against only 4 distinct coauthors, consistent with the low-moderate collaboration score of 32. The recurring partners are Dennis Gross, Arnaud Gotlieb, Nassim Belmecheri and Nadjib Lazaar, spanning the constraint-acquisition and neural-network-verification threads; the 2026 Evolve-catalog papers place Spieker in first-author and last-author roles.
Research keywords
Highlights
Certified probabilistic model checking for autoregressive neural sequence models
Human-free constraint acquisition via a Transformer oracle
Model checking as an oracle for testing LLM-based explainers
A 2026 last-authored paper proposing automated testing of LLM-based post hoc explainers with model checking as the oracle — an early signal of the testing-meets-explainable-AI direction, not yet reviewed by Evolve.
Timeline
- 2015First year in the OpenAlex counts-by-year series, with 2 recorded works.
- 2019Output steadies at 5 recorded works (6 in 2020), starting a multi-year climb.
- 20219 recorded works, the highest annual count to that point.
- 202418 recorded works — a step change from 6 in 2023.
- 202528 recorded works, the highest annual count in the series.
- 202614 works recorded so far; three papers in the Evolve catalog as first or last author.
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.
- All bibliometrics come from an OpenAlex snapshot (identity: openalex); the profile is name-resolved from public bibliographic data, and no biographical facts beyond institution and country are used.
- The output and impact radar dimensions are unscored because the underlying data is insufficient in this snapshot.
- The reproducibility score of 100 rests on a single assessed paper (reproPapers = 1).
- One catalog paper ('Automated Testing of LLM-Based Post Hoc Explainers Using Model Checking as an Oracle') has no Evolve verdict, keywords, or reproducibility assessment yet.