KIIT University ML researcher bridging machine unlearning evaluation and training-free time series forecasting
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
22
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: this snapshot lacks the normalized publication-volume and completeness measures Evolve's output dimension requires. The axis would measure cumulative and recent publication output; h-index 22 and i10-index 28 are on file but not converted into a score.
Impact—
Not yet scored. This axis would measure citation-based influence — citations per work, field-normalized rates, and uptake over time. The snapshot records h-index 22 and i10-index 28 but provides no per-work citation detail to compute it.
Focus0/100
Scores 0 for concentration, reflecting an unusually even field split — machine learning and time series forecasting each carry 0.5 of recent work — and topic counts spread across five clusters: neural network applications 24, privacy-preserving technologies 22, adversarial robustness 19, topic modeling 11, image enhancement 11. The record reads as methodologically broad rather than single-niche.
Collaboration41/100
The collaboration score is 41. Across 49 works in the last three years the snapshot records only 7 distinct co-authors, and the frequent list names seven people — Aaryaman Kalani, Dhruv Kumar, Mohan Kankanhalli, Yash Sinha, Karthik Sridhar, Aaditya Jain, Saurabh Deshpande. That is a small, stable writing circle relative to output volume; both 2026 catalog papers carry middle, non-corresponding authorship.
Activity100/100
Activity scores 100, the ceiling of the scale. The counts-by-year series climbs from 1 work in 2015 to 10 in 2020 and 12 in 2021, holds at 7–8 through 2022–2024, then rises to 11 in 2025 and 30 recorded for 2026 — a snapshot year that may be partial. Recent three-year output is 49 works, consistent with sustained high-volume publishing.
Reproducibility55/100
Reproducibility scores 55, a mean over the 2 papers Evolve has reviewed, each rated medium readiness. Event Signature Transfer is training-free and quantile-only but relies on hand-set controls with no prospective test; The BatchNorm Illusion runs convincing controls yet reports key tables single-seed with no code release. Medium readiness on both sides sets the current baseline.
Research summary
Murari Mandal is a machine learning researcher at KIIT University (India) whose OpenAlex profile spans two evenly weighted fields — machine learning and time series forecasting, each 0.5 of recent work — with topic mass in advanced neural network applications (24), privacy-preserving technologies (22), adversarial robustness (19), topic modeling (11), and image enhancement (11). The snapshot reports an h-index of 22, an i10-index of 28, 49 works over the last three years, and 7 distinct recent co-authors. Evolve has reviewed two of this researcher's 2026 arXiv preprints, both in middle authorship: Event Signature Transfer, a training-free, quantile-only operator that transfers historical event effects onto forecasts (judged promising but entirely in-sample, with hand-set controls and no prospective test), and The BatchNorm Illusion, which claims a weight-free BatchNorm pass reverses apparent forgetting in machine unlearning evaluation by up to 78 percentage points (convincing controls, but key tables single-seed and no code release). Both papers are rated medium reproducibility readiness, giving a reproducibility mean of 55 over 2 reviewed papers. The activity dimension tops the scale, while output and impact dimensions are not yet scored.
Collaboration pattern
Recent output is produced within a compact circle: 49 works over the last three years with only 7 distinct co-authors recorded. The frequent list names Aaryaman Kalani, Dhruv Kumar, Mohan Kankanhalli, Yash Sinha, Karthik Sridhar, Aaditya Jain, and Saurabh Deshpande. Both 2026 catalog papers sit in middle authorship without a corresponding role. The collaboration dimension scores 41, mid-range — steady repeat partners rather than a wide ad-hoc network.
Research keywords
Highlights
Training-free event-to-forecast scenario transfer
A weight-free BatchNorm pass that reverses apparent unlearning
Timeline
- 2015First recorded output in the OpenAlex snapshot: 1 work, followed by 1–2 works per year through 2018.
- 2020Annual output jumps to 10 works (from 4 in 2019), the start of a sustained ramp.
- 2021Early peak of 12 works, the highest full-year total before 2026.
- 2022Output settles at 7 works, holding at 7–8 per year through 2024.
- 2025Output rises again to 11 works.
- 202630 works recorded in the snapshot year, including the two arXiv preprints Evolve reviewed: Event Signature Transfer and The BatchNorm Illusion.
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 bibliometric figures come from a single OpenAlex snapshot (identity: openalex); no other databases were audited, and the profile is name-resolved from public bibliographic data.
- Output and impact dimensions are unscored in this snapshot; h-index 22 and i10-index 28 are on file but not converted into dimension scores.
- Evolve's qualitative review covers only 2 papers — both 2026 arXiv preprints with middle, non-corresponding authorship — so judgments may not represent the full portfolio.
- The 2026 count (30 works) covers a snapshot year in progress and may be partial.