Publications

Conference and Journal Papers

All-in-One MRI Reconstruction Manuscript in Progress

All-in-One MRI Reconstruction with Cascaded Transformers

Read more

Cascaded transformer U-Nets with Frequency Mining for multi-contrast MRI reconstruction from k-space (MICCAI-style benchmarks). Emphasizes robustness across contrasts and acceleration factors with artifact mitigation and strong generalization.

Diffusion-Driven Airfoil Optimization Manuscript in Progress

Diffusion-Driven Airfoil Optimization

Read more

Latent-space diffusion + neural multi-objective BO for inverse-designing 2D airfoils, balancing lift–drag while encouraging novelty; large-batch acquisition improves Pareto diversity.

Deep RL for Stock Trading IEEE ICCKE 2024

Deep Reinforcement Learning for Stock Trading

Read more

FinBERT sentiment fused with PPO-LSTM; reached 134.39% cumulative return and 1.46 Sharpe, outperforming DJIA, Ensemble, and PPO baselines, with risk-aware reward shaping and rolling-window evaluation.

SincNet for Parkinson’s Gait Under Review — AIHC

SincNet for Parkinson’s Disease Detection

Armin Salimi-Badr, Mahan Veisi, Sadra Berangi
Read more

SincNet-based gait classification with post-hoc filter clustering to surface medically relevant frequency bands; achieved 98.7% accuracy with interpretable components.

RoboCup Soccer Simulation RoboCup Proc.

Intelligent Agents for RoboCup Soccer Simulation

Read more

Role-optimized agents (goalkeeping, defense, teamwork) in the 2D Simulation League; team placed 2nd at IranOpen 2023.