Studies how quantization and pruning reshape adversarial robustness in EEG decoders and vision-language models
2
papers here
0
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-09-27) 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: this axis would measure total publication volume, venue mix, and output rhythm across the full career record. Only two reviewed 2026 arXiv papers are currently catalogued, so no reliable output score can be computed yet.
Impact—
Insufficient data: this axis would measure citation counts, h-index, and downstream influence of the work. Neither citation aggregates nor an ORCID-linked record is available in the snapshot, so no impact score is reported and none is estimated.
Focus—
Insufficient data: this axis would measure how research topics and venues distribute and shift over time. The two catalogued papers both fall in 2026 and share a compression-versus-reliability theme, but a two-paper window cannot establish a focus profile.
Collaboration14/100
The collaboration axis scores 14 out of 100, computed from real bibliometric data showing just 1 distinct coauthor in the current snapshot. Saim Rehman is the one recurring partner, appearing across this profile's catalogued work. Both papers place Shafique in the last author position, a senior-authorship pattern, though neither is flagged corresponding. The narrow network partly reflects a thin snapshot: with only two catalogued papers, collaboration breadth is inevitably understated.
Activity—
Insufficient data: this axis would measure publishing cadence and year-over-year momentum. Both known papers are dated 2026, so recent output is visible, but earlier years are absent from the snapshot and no activity score can be computed.
Reproducibility—
Insufficient data: this axis would measure artifact release, code availability, and reproducibility signals across the record. The only current signal is Evolve's review tagging both papers medium on repro readiness; a profile-level score is not yet computable.
Research summary
Muhammad Shafique appears in Evolve's catalog as a last-author contributor to two 2026 arXiv studies that share one research question: when neural networks are compressed in accuracy-preserving ways, what silently changes beyond the headline score? The first, AERIAL, adversarially evaluates low-precision EEG decoders for brain-computer interfaces; Evolve's review calls it a well-controlled single-dataset study showing that accuracy-preserving compression alters adversarial transfer rather than direct robustness, with claims largely supported. The second, GHOST-Q, examines grounding hallucinations in quantized vision-language models, finding that post-training quantization preserves headline scores while shifting item-level grounding outcomes; the review praises rigorous paired statistics while noting small effect sizes and a bitsandbytes-only scope. Together the two papers sketch a coherent agenda spanning brain-computer interfaces, adversarial machine learning, and efficient multimodal AI. In the current bibliometric snapshot the collaboration footprint is narrow: one distinct coauthor, Saim Rehman, yields a collaboration dimension score of 14 out of 100. No institution, country, or ORCID is attached to this profile, so it is name-resolved from public bibliographic data, and the other five radar dimensions lack sufficient data. This is an early-stage profile that will sharpen as more of the publication record is catalogued and reviewed.
Collaboration pattern
Both catalogued 2026 papers place Muhammad Shafique in the last author position — a senior-authorship pattern — though neither is flagged corresponding. The snapshot records exactly 1 distinct coauthor, Saim Rehman, who is also the sole frequent collaborator, giving a collaboration score of 14 out of 100. This low score signals a narrow visible network rather than isolation: with only two reviewed papers, broader co-authorship simply has not been captured yet.
Research keywords
Highlights
AERIAL: compression shifts adversarial transfer in EEG decoders
GHOST-Q: quantized VLMs keep scores but shift grounding outcomes
Timeline
- 2026AERIAL posted to arXiv: adversarial evaluation of robustness in accuracy-preserving low-precision EEG decoders, with Shafique in the last author position.
- 2026GHOST-Q posted to arXiv: grounding hallucinations overlooked under same-score trade-offs in quantized VLMs, again as last author.
- 2026Both papers enter the Evolve catalog with published verdicts and a medium repro-readiness tag; only the collaboration dimension has enough data to score.
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 dimension scores are computed from public bibliometric data only; five of six axes (output, impact, focus, activity, reproducibility) currently lack sufficient data and are unscored.
- Muhammad Shafique is a common name; this profile is name-resolved from public bibliographic data, and no institution, country, or ORCID was available to confirm identity.
- The evidence base is two 2026 arXiv preprints reviewed by Evolve; the broader publication record, including peer-reviewed venues, is not yet reflected.
- Both papers carry a medium repro-readiness tag from Evolve's review; artifact and code availability beyond that has not been assessed.