Mr Weijian Deng
Areas of expertise
- Artificial Intelligence 4602
- Computer Vision 460304
- Machine Learning 4611
Research interests
- Machine Learning Safety: Concentrated on the safety of large language models and multimodal models, with a focus on improving resilience and reliability. Emphasis is placed on developing robust models for varied environments and creating monitoring mechanisms to detect misuse and analyze failure patterns.
- 3D Content Modeling & Generation: Focused on the advancement of 3D modeling and generation, specifically for refractive objects. By applying optical principles, it aims to significantly enhance the realism and accuracy of 3D objects and scenes.
PhD Research Topic: Model Generalization Prediction. The primary focus lies in the accurate prediction of model generalization across diverse testing environments without relying on human annotations. The significance of this investigation lies in its potential to identify and diagnose potential failure cases, while also providing valuable guidance for future model training endeavors.
Please check my personal website for my research works: https://weijiandeng.xyz/
Biography
I am a Research Fellow at the Australian National University, working with Prof Stephen Gould. My current research focuses are 3D Content Modeling and Out-of-Distribution Generalization. Previously, I was a PhD student at the Australian National University, where I worked on model generalization prediction. My resume is available at CV.
Researcher's projects
- Model Generalization Prediction
- Vision Fundation Model Assessment
- Transparent Object Modeling
Publications
- Deng, W, Gould, S & Zheng, L 2021, 'What Does Rotation Prediction Tell Us about Classifier Accuracy under Varying Testing Environments?', 38th International Conference on Machine Learning, ed. Marina Meila, Tong Zhang, Cambridge MA: JMLR, USA.
- Deng, W, Zheng, L, Sun, Y, Jiao, J et al. 2021, 'Rethinking Triplet Loss for Domain Adaptation', IEEE Transactions on Circuits and Systems for Video Technology, vol. 31, no. 1, pp. 29-37.