AI & Mathematics — Deakin University

Bao Minh
Tran.

I teach machines to reason over graphs — and I'm chasing brain-like models that move us closer to genuinely human-like AI. Optimization is where it all begins.

Portrait of Bao Minh Tran Open to research & roles
01

About

Who / Why

I'm an undergraduate in Artificial Intelligence at Deakin University, driven by curiosity and a deep love of mathematics. My work lives where optimization meets graph neural networks — the geometry of how information flows across structured data.

Increasingly, I'm drawn to neuroscience-inspired models such as brain-like neural networks, which I believe are indispensable to building AI that genuinely resembles human reasoning.

Before university, I trained as a competitive programmer, picking up a strong foundation in data structures and algorithms — and a habit of chasing problems all the way to their mathematical core.

94/100WAM
Deakin University
2027Expected
graduation
3Olympiad
awards
PyTorch·HPCResearch
toolkit
02

What I work on

Research interests
Optimization Graph Neural Networks Brain-like Neural Networks Graph Theory Linear & Convex Optimization Multi-objective Optimization Distribution Shift / OOD Computer Vision Medical ML Mathematical Modelling

A full breakdown of the questions I find worth chasing lives on the research page.

03

Selected work

2025 — 2026

Choquet Adaptive Scalarisation (CAS)

Aug — Sep 2026
SIT316 · Deakin University · Supervised by Prof. Gleb Beliakov
  • Designed a nonlinear multi-objective optimiser that scalarises objectives with the Choquet integral.
  • Captures non-convex and disconnected Pareto fronts; benchmarked on ZDT and DTLZ problems.
PythonpymooMulti-objective Optimization
Details →

Sign Language Recognition with X3D

Aug — Sep 2026
SIT332 · Deakin University
  • Fine-tuned three X3D video models at six freezing depths on AUTSL (226 signs) with Slurm on A100s.
  • Reached 82.3% top-1 with X3D-M at 4.9 GFLOPs per clip.
PyTorchVideoComputer VisionSlurm / HPC
Details →

Out-of-Distribution Data Fitting

Nov 2025 — Feb 2026
ADR Summer Project 2025 · Deakin University
  • Designed splitting strategies that emulate real-world distribution shift.
  • Benchmarked state-of-the-art ML/DL models under shifted data regimes.
PyTorchBenchmarkingOOD
Details →