First-author XAI researcher probing task separability to localize where neural representations encode task information.
1
papers here
1
reviewed by Evolve
1
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-21) 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—
Insufficient data: only one cataloged paper (2026) with no publication-volume metrics in this snapshot. This axis would measure total scholarly output — paper counts, venues, and yearly production volume.
Impact—
Insufficient data: no citation counts, h-index, or influence metrics are available yet. This axis would measure how widely the 2026 TRIPROBE paper and any other work are cited and taken up by the community.
Focus100/100
Focus scores a maximum 100 because the entire cataloged output sits in a single field — explainable AI / machine learning covering representation probing and sEMG signal processing (field distribution: one paper, one field). With one reviewed paper, this concentration reflects a sharply defined agenda of probing-based interpretability, though the score only stays meaningful as more entries accumulate.
Collaboration32/100
The collaboration score of 32 rests on a compact footprint: 4 distinct coauthors across the cataloged work. The frequent-collaborer list — Freek Hens, Aleksa Bokšan, Mohammad Mahdi Dehshibi, and Mahyar Shahsavari — suggests a small, stable working group rather than a large distributed network, and the reviewed paper was first-authored, indicating hands-on leadership within that tight circle.
Activity—
Insufficient data: a single 2026 preprint cannot support a yearly productivity trend. This axis would measure recent research tempo — publications per year and current momentum across venues.
Reproducibility20/100
Reproducibility averages 20 across 1 evaluated paper and is rated low. Evolve's review of TRIPROBE cites concrete gaps: a single 3-class dataset, purely qualitative rankings, unreconciled sample counts, and an unspecified evaluation protocol. These are tractable fixes — releasing code, reconciling counts, and fixing a protocol would move future scores upward — but the current evidence base is thin.
Research summary
Amirhossein Sadough is a researcher working in explainable AI and machine learning, centered on representation probing and its application to sEMG signal processing. Evolve's catalog includes one first-authored paper, TRIPROBE (2026, arXiv), which proposes a three-stage probing framework — foundational, latent, and final — that uses the Maximum Fisher's Discriminant Ratio to localize where task-separability bottlenecks arise in learned representations. Applied to a Roshambo sEMG dataset, its three-class findings align with prior results, which the Evolve review reads as support for the diagnostic idea, though the evidence stays thin: a single 3-class dataset, purely qualitative rankings, unreconciled sample counts, and an unspecified evaluation protocol. Reproducibility readiness is rated low (score 20, from 1 paper). The collaboration footprint is compact — 4 distinct coauthors, with Freek Hens, Aleksa Bokšan, Mohammad Mahdi Dehshibi, and Mahyar Shahsavari as frequent collaborators — while focus is maximal (score 100), with all cataloged output in one field. Output, impact, and activity metrics are not yet available, so this profile is an early-stage, name-resolved view of a research program built around probing-based interpretability.
Collaboration pattern
The observed pattern is a small, tight team: 4 distinct coauthors, with Freek Hens, Aleksa Bokšan, Mohammad Mahdi Dehshibi, and Mahyar Shahsavari recurring as frequent collaborators. Combined with a first-author position on the reviewed paper, this suggests hands-on leadership within a stable group rather than shifting, large-scale consortia; the collaboration score of 32 reflects that compact footprint. All evidence comes from one cataloged paper, so the network picture may undercount real reach.
Research keywords
Highlights
Three-stage probing framework for task separability
Roshambo sEMG results align with prior findings
Low reproducibility readiness with concrete gaps
Timeline
- 2026First-authored TRIPROBE posted to arXiv, introducing a three-stage probing framework that applies the Maximum Fisher's Discriminant Ratio to localize task-separability bottlenecks.
- 2026Evolve reviews TRIPROBE and rates its reproducibility readiness as low (score 20 based on 1 paper), citing a single 3-class dataset and an unspecified evaluation protocol.
- 2026The bibliometric snapshot records 4 distinct coauthors, with Freek Hens, Aleksa Bokšan, Mohammad Mahdi Dehshibi, and Mahyar Shahsavari as frequent collaborators.
- 2026All cataloged output sits in a single field — explainable AI / machine learning (representation probing, sEMG signal processing) — yielding a maximum focus score of 100.
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 radar scores rest on a very small base (1 reviewed paper); output, impact, and activity axes are marked missing rather than estimated.
- No ORCID, affiliation, or country is provided; the profile is name-resolved from public bibliographic data and may conflate same-name researchers.
- The reproducibility score (mean 20) reflects a single evaluated paper (reproPapers 1) and should not be read as a stable rating.
- Co-author and field statistics derive solely from the Evolve-reviewed catalog paper and its snapshot; external publication records are not included.