M. Tanveer

Indian Institute of Technology Indore

Unclaimed · AI-structured profileORCID OpenAlex

IIT Indore machine learning researcher: robust randomized networks, noisy-label imbalance, medical VLM audits

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

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—

Output is unscored: no publication-count or productivity metrics exist in this snapshot. This axis would measure total and per-year output volume across venues; the four catalog preprints alone cannot support an estimate.

Impact—

Impact is unscored: the snapshot carries no citation counts, h-index or venue-influence data. This axis would measure how widely the work is cited and taken up, which cannot be estimated here.

Focus0/100

Focus scores 0 because the field distribution splits perfectly evenly — machine learning, machine learning (robust randomized neural networks), and medical imaging machine learning (evaluation methodology) each hold 0.333 across the four catalog papers. That reads as three parallel, equally weighted research lines rather than deep concentration in a single specialty.

Collaboration39/100

The collaboration axis scores 39 on this snapshot, built on a small co-author base: 6 distinct co-authors are recorded, with a recurring circle of Mushir Akhtar, Mohd. Arshad, Akarsh J., A. Quadir, A. Rahaman and P. N. Suganthan. All four catalog papers place Tanveer in a middle or last author position, none as corresponding author — a senior-contributor pattern within a compact, stable team rather than a large distributed network.

Activity—

Activity is unscored: all four catalog items are 2026 preprints and no multi-year publication record is available, so research tempo cannot be computed. This axis would measure publishing cadence and recent-year momentum.

Reproducibility55/100

Reproducibility scores 55 — the mean readiness across 3 reviewed papers, each rated medium. Evolve's verdicts credit the medical VLM audit as rigorous and honest, but flag grid-best test reporting that inflates gains on IFW-BLS and missing noisy long-tailed baselines on BARGE, so protocols look plausible yet not fully turnkey.

Research summary

M. Tanveer is a machine learning researcher at the Indian Institute of Technology Indore (ORCID 0000-0002-5727-3697). This profile is a full-tier, name-resolved view built from public bibliographic data and four 2026 arXiv preprints in the Evolve catalog. The work splits into three roughly equal lines — the snapshot's field distribution assigns 0.333 to each of machine learning, robust randomized neural networks, and medical imaging machine learning (evaluation methodology). One line develops robust randomized-network models: IFW-BLS couples intuitionistic fuzzy sample weighting with a bounded asymmetric wave loss for the broad learning system, while GBFRVFL brings granular-ball computing to fuzzy random vector functional link networks. A second line tackles imbalanced learning under noisy labels, notably BARGE's bounded prior-adjusted score with reliability-guided embedding geometry. The third audits medical vision-language models: a chest X-ray tuberculosis screening study finds that rankings, reliability and thresholds often fail to port across cohorts and prompts. Evolve has reviewed three of the four papers, rating each reproducibility medium for a reproducibility score of 55, while output, impact and activity metrics remain unavailable. Frequent co-authors include Mushir Akhtar, Mohd. Arshad, Akarsh J., A. Quadir, A. Rahaman and P. N. Suganthan.

Collaboration pattern

Collaboration on this snapshot is compact: 6 distinct co-authors underpin a collaboration score of 39. The recurring circle includes Mushir Akhtar, Mohd. Arshad, Akarsh J., A. Quadir, A. Rahaman and P. N. Suganthan. Across the four catalog papers Tanveer appears in middle or last author positions, none as corresponding author, consistent with a senior contributor working within a small, stable group rather than leading large multi-institution teams.

Research keywords

long-tailed learninglearning with noisy labelsclass imbalancelogit adjustmentproper scoring rulesbroad learning systemrandomized neural networksrobust loss functionsvision-language modelszero-shot classificationchest X-ray tuberculosis screeninggranular-ball computing

Highlights

Auditing medical VLMs for chest X-ray TB screening
Four medical vision-language models are audited for zero-shot chest X-ray tuberculosis screening. Evolve calls it a rigorous, honest audit: rankings, reliability and thresholds often fail to port across cohorts and prompts, with prompt sensitivity and negative-spectrum behaviour examined. Reproducibility rated medium.
BARGE: bounded adjustment for imbalanced learning with noisy labels
BARGE combines a bounded prior-adjusted score with reliability-guided embedding geometry for long-tailed, label-noise settings. Evolve notes it trails logit adjustment on clean data, its noisy-setting edge is modest (70.0 vs 72.3 MBE avg), and noisy long-tailed baselines are absent. Reproducibility medium.
IFW-BLS: dual-robust broad learning system
Pairs intuitionistic fuzzy sample weighting with a bounded asymmetric wave loss to make the broad learning system robust to label noise. Evolve judges the method sound but flags grid-best test reporting that inflates reported gains. Reproducibility medium.
GBFRVFL: granular-ball fuzzy random vector functional link network
Applies granular-ball computing within a fuzzy random vector functional link network, extending the randomized-network research line. This paper is in the catalog but not yet reviewed by Evolve, so no verdict, keywords or repro-readiness rating exist.

Timeline

2026Posted an audit of four medical vision-language models for chest X-ray tuberculosis screening, finding that rankings, reliability and thresholds often fail to port across cohorts and prompts.
2026Proposed BARGE for imbalanced learning with noisy labels, pairing a bounded prior-adjusted score with reliability-guided embedding; the review notes a modest noisy-setting edge (70.0 vs 72.3 MBE avg).
2026Introduced IFW-BLS, a dual-robust broad learning system pairing intuitionistic fuzzy sample weighting with a bounded asymmetric wave loss.
2026Published GBFRVFL, applying granular-ball computing to a fuzzy random vector functional link network; not yet reviewed by Evolve.

Papers in the Evolve catalog

Frequent co-authors

Mushir Akhtar×3 Mohd. Arshad×2 Akarsh J.×1 A. Quadir×1 A. Rahaman×1 P. N. Suganthan×1

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.

Dimension scores come from a sparse public bibliometric snapshot; output, impact and activity lack underlying data and are left unscored.
Evidence covers four 2026 arXiv preprints in the Evolve catalog; peer-reviewed venue records and the full publication history are not represented.
One catalog paper (GBFRVFL) awaits Evolve review, so it has no verdict, keywords, field tag or repro-readiness rating.
Profile is name-resolved from public bibliographic data (ORCID 0000-0002-5727-3697); country and career details are absent from the source data.