Weizheng Wang 王维政
Portrait of Weizheng Wang

Weizheng Wang 王维政

Master of Artificial Intelligence and Machine Learning graduate

I study multimodal UAV decision evaluation, visual memory, efficient computer vision, and edge deployment. From February to August 2026, I worked remotely as a Research Assistant at Zayed University on protocol-conditioned evaluation of multimodal UAV agents.

Before Adelaide, I completed a B.Eng. in Computer Science and Technology at North China Institute of Science and Technology, where I worked on UAV monitoring, object detection, and practical perception systems.

Research interests: multimodal evaluation, visual episodic memory, efficient vision, structured pruning, and UAV/edge AI.

Selected Publications

View all

  1. TRIM: Dual-Budget Structured Pruning with Information-Modulated Saliency for CNN, Transformer, and Mamba Vision Models

    W. Wang†, J. Wu†, Q. Tian, and L. Tian. Submitted to IEEE ICASSP 2027, 2026. Repository

  2. When Retrieval Is Not Binding: A Diagnostic Benchmark for Visual Episodic Memory

    W. Wang†, J. Wu†, L. Tian, and Q. Tian. Submitted to IEEE ICASSP 2027, 2026. Repository

  3. A Lightweight Thermal Denoising and Occlusion-Robust Infrared Detection Model for Substation Equipment

    J. Wu†, W. Wang†, Y. Niu, X. Shi, and L. Tian. IEEE ICIP, 2026. Code

† Equal contribution. Submitted manuscripts are not yet accepted.

Updates

  • Submitted TRIM, visual episodic-memory diagnostics, ManiSeg, and EmergUAV-Bench to IEEE ICASSP 2027.
  • Released the MulRobBench preprint; submitted to Applied Soft Computing.
  • Joined Zayed University as a remote Research Assistant in February.

Research Experience

View all

  1. Zayed University

    Research Assistant · February–August 2026 · Remote

    Co-developed MulRobBench: 3,024 UAV decision samples across 17 task-taxonomy nodes, with evaluation pipelines for 17 multimodal models.

  2. Huawei MindSpore Open-Source Community

    Research Intern · February–June 2025

    Contributed to MindNLP model integration and fine-tuning workflows for Autoformer, BEiT, and ALBERT, with attention to training stability and reproducibility.

  3. Automatic Inspection and Rescue Based on Drone Nest

    Research Intern · January–July 2024

    Developed UAV inspection pipelines using DeepSORT and YOLOv8, with ONNX export for real-time edge inference.

Education

  • University of Adelaide / Adelaide University
    Master of Artificial Intelligence and Machine Learning
    Completed 18 August 2026; degree conferred 7 October 2026
    Testamur issued by Adelaide University
  • North China Institute of Science and Technology
    B.Eng. in Computer Science and Technology, GPA 4.02

Selected Honors

View all
  • University of Adelaide Global Citizens 30% International Scholarship
  • Principal Investigator, National Undergraduate Innovation Training Program
  • Second Prize, Challenge Cup Science and Technology Invention and Creation Track