Deakin computer scientist: computer vision, bandit and optimization ML, and AI-driven digital mental health
4
papers here
1
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
60
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: the per-year output aggregation underlying this axis is not available in the current assessment data, so no output level is inferred. The axis would measure publication volume and its trajectory over time relative to field norms.
Impact—
Not yet scored: citation-based impact aggregates are absent from the current assessment data, so no impact estimate is made. The axis would measure citation uptake and scholarly influence of the work relative to field benchmarks.
Focus100/100
Topic concentration is strong: five themes dominate the corpus — advanced bandit algorithms (90 works), video analysis and summarization (84), multi-objective optimization (69), video surveillance and tracking (67), and AI in healthcare (60). The one field-tagged catalog paper sits in machine-learning methodology for clinical prediction design, consistent with a focused shift from vision toward statistical ML and health applications.
Collaboration62/100
The recent window records 21 distinct co-authors, a mid-range score. The frequent-coauthor list spans machine-learning methodology (Dang Nguyen, Bao Duong, Arun Kumar, Julian Berk), LLM multi-agent research (Dat Phan-Trong, Debopriyo Bal), and digital mental-health teams (Taylor A. Braund, Leonard Hoon, Wu Yi Zheng, Jill Newby, Alexis Whitton), matching the cataloged 2026 work.
Activity100/100
Activity scores at the maximum: 87 works in the recent three-year window, and countsByYear shows sustained late-career volume — 39 works in 2024, 35 in 2025, and 13 already in 2026 — after annual peaks of 61 in 2016 and 47 in 2021, continuing an indexed record that begins in 1989.
Reproducibility20/100
Currently weak and thin: the score of 20 averages a single assessed paper (reproPapers: 1). That paper — the ReLU-region sample-size framework — is rated low readiness, with effect sizes drawn from noiseless proxy surfaces and internal-only validation. More reviewed papers are needed before this axis stabilizes.
Research summary
Svetha Venkatesh is a computer scientist at Deakin University, Australia, whose OpenAlex-indexed record spans 1989 to 2026, with an h-index of 60 and an i10-index of 337. Topic profiling of the corpus shows strong concentration: advanced bandit algorithms (90 works), video analysis and summarization (84), multi-objective optimization (69), video surveillance and tracking (67), and artificial intelligence in healthcare (60) — a career-long arc from computer vision toward statistical machine learning and clinical applications. Activity remains high, with 87 works in the recent three-year window, 21 distinct co-authors, and annual counts that peaked at 61 in 2016 and still run at 35-47 per year. The Evolve catalog carries four 2026 arXiv papers: a statistical framework for estimating machine-learning sample size via ReLU-region decomposition; an empirical study asking whether multi-agent debate suppresses creative diversity in LLMs, judged convincing though informally theorized; and two studies predicting depression-related symptoms in university students through novel uncertainty-aware oversampling. Reproducibility evidence is currently thin and weak — one assessed paper at low readiness — while the radar's output and impact dimensions remain unscored.
Collaboration pattern
Her recent window records 21 distinct co-authors. Sunil Gupta heads the frequent-coauthor list, and the named set spans machine-learning methodology (Dang Nguyen, Bao Duong, Arun Kumar, Julian Berk), LLM multi-agent research (Dat Phan-Trong, Debopriyo Bal), and digital mental-health teams (Taylor A. Braund, Leonard Hoon, Wu Yi Zheng, Jill Newby, Alexis Whitton) — matching the cataloged 2026 work.
Research keywords
Highlights
Statistical sample-size estimation for machine-learning models
Does multi-agent debate suppress creative diversity?
Uncertainty-aware screening for student depression
Two 2026 preprints predict amotivation and anhedonia symptoms and depression remission among university students, introducing novel oversampling methods including SMOTE-VAR, an uncertainty-aware variant. They are cataloged but not yet reviewed by Evolve, so keywords and reproducibility data are pending.
Timeline
- 1989Earliest year in the OpenAlex countsByYear record, with 1 published work — the start of an indexed output span running to 2026.
- 2016Output peak: 61 works indexed in a single year, within a sustained 2014-2021 stretch in which annual counts never fall below 36.
- 202147 works indexed — the highest annual count since the 2016 peak, within a recent run of roughly 39-47 works per year.
- 2026Four arXiv preprints enter the Evolve catalog: machine-learning sample-size estimation, multi-agent debate and creative diversity, and two uncertainty-aware oversampling studies of student depression.
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 counts, topics, and co-author lists come from a single OpenAlex snapshot; citation-based impact is not scored here.
- The output and impact radar dimensions are unscored, although the snapshot lists an h-index of 60 and an i10-index of 337.
- The reproducibility score rests on a single reviewed paper rated low readiness, so it is an early-stage signal.
- Two of the four cataloged 2026 papers lack Evolve review, keywords, and repro-readiness data.