![Yuanbang Liang](https://profiles-cardiff.imgix.net/attachments/7532?w=200&h=200&auto=format&crop=faces&fit=crop&q=90")
Yuanbang Liang
(e/fe)
Timau a rolau for Yuanbang Liang
Cydymaith Ymchwil
Cyhoeddiad
2026
- Wang, Y. et al., 2026. GRADE: probing knowledge gaps in LLMs through gradient subspace dynamics. Presented at: COLM 2026: The Third Annual Conference on Language Modeling San Francisco, California 6 - 9 October 2026. COLM 2026: The Third Annual Conference on Language Modeling. Open Review:
2025
- Liang, Y. 2025. Hubness awareness sampling for deep generative models in generation and evaluation. PhD Thesis , Cardiff University.
2024
- Liang, Y. et al. 2024. Deep generative model based rate-distortion for image downscaling assessment. Presented at: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024 Seattle, WA, USA 17-21 June 2024. Proceedings of the Conference on Computer Vision and Pattern Recognition. IEEE. , pp.19363-19372. (10.1109/CVPR52733.2024.01832)
- Liang, Y. et al. 2024. Efficient precision and recall metrics for assessing generative models using hubness-aware sampling. Presented at: The Forty-first International Conference on Machine Learning (ICML) Vienna, Austria 21-27 July 2024. Vol. 235., pp.29682-29699.
2023
- Song, S. et al. 2023. Feature proliferation — the "cancer" in StyleGAN and its treatments. Presented at: International Conference on Computer Vision (ICCV) 2023 Paris, France October 1 - 6, 2023. Proceedings of IEEE/CVF International Conference on Computer Vision. IEEE. , pp.2360-2370. (10.1109/ICCV51070.2023.00224)
2022
- Liang, Y. et al. 2022. Exploring and exploiting hubness priors for high-quality GAN latent sampling. Presented at: The 39th International Conference on Machine Learning (ICML 2022) Baltimore, Maryland USA 17-23 July 2022. Vol. 162.ML Research Press. , pp.13271-13284.
Cynadleddau
- Wang, Y. et al., 2026. GRADE: probing knowledge gaps in LLMs through gradient subspace dynamics. Presented at: COLM 2026: The Third Annual Conference on Language Modeling San Francisco, California 6 - 9 October 2026. COLM 2026: The Third Annual Conference on Language Modeling. Open Review:
- Liang, Y. et al. 2024. Deep generative model based rate-distortion for image downscaling assessment. Presented at: The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024 Seattle, WA, USA 17-21 June 2024. Proceedings of the Conference on Computer Vision and Pattern Recognition. IEEE. , pp.19363-19372. (10.1109/CVPR52733.2024.01832)
- Liang, Y. et al. 2024. Efficient precision and recall metrics for assessing generative models using hubness-aware sampling. Presented at: The Forty-first International Conference on Machine Learning (ICML) Vienna, Austria 21-27 July 2024. Vol. 235., pp.29682-29699.
- Song, S. et al. 2023. Feature proliferation — the "cancer" in StyleGAN and its treatments. Presented at: International Conference on Computer Vision (ICCV) 2023 Paris, France October 1 - 6, 2023. Proceedings of IEEE/CVF International Conference on Computer Vision. IEEE. , pp.2360-2370. (10.1109/ICCV51070.2023.00224)
- Liang, Y. et al. 2022. Exploring and exploiting hubness priors for high-quality GAN latent sampling. Presented at: The 39th International Conference on Machine Learning (ICML 2022) Baltimore, Maryland USA 17-23 July 2022. Vol. 162.ML Research Press. , pp.13271-13284.
Gosodiad
- Liang, Y. 2025. Hubness awareness sampling for deep generative models in generation and evaluation. PhD Thesis , Cardiff University.
Ymchwil
Mae fy niddordebau ymchwil yn canolbwyntio ar ddysgu peiriannau a'i gymwysiadau mewn gweledigaeth gyfrifiadurol, graffeg gyfrifiadurol a chynhyrchu cynnwys. Mae fy ymchwil PhD, sy'n cael ei ariannu'n llawn gan yr EPSRC, yn canolbwyntio ar y ffenomen hubness i ddatgelu'r berthynas rhwng y dosbarthiad hyper-dimensiwn a'r modelau cynhyrchiol.
Nawr mae fy ymchwil ar esboniadwy modelau cynhyrchiol i wneud canlyniadau a gynhyrchir gan fodel yn fwy cyd-fynd â ffeithiau gwrthrychol yn her graidd mewn AI, y cyfeirir ato yn aml fel aliniad AI. Y nod yw sicrhau bod ymddygiad a phenderfyniadau system AI yn cyd-fynd â bwriad, gwerthoedd a realiti gwrthrychol dynol.
Contact Details
Themâu ymchwil
Arbenigeddau
- AI & Dysgu Peiriant
- Dysgu peiriant Adversarial
- Golwg cyfrifiadurol
- AI cynhyrchiol a LLMs