Selected Publications
Do VLMs Answer Without Looking? Quantifying Language Prior Bias via Shapley Attribution
Yinong Oliver Wang*, Nicholas Mesa-Cucalon*, Ting-Yao Hu, Raviteja Vemulapalli, Oncel Tuzel, Nicholas Apostoloff, Jerremy Holland, Fernando De la Torre,
Under review at NeurIPS 2026
"A Shapley-based metric that quantifies how much a VLM's answer is driven by the language prior instead of the image"
Yinong Oliver Wang*, Nicholas Mesa-Cucalon*, Ting-Yao Hu, Raviteja Vemulapalli, Oncel Tuzel, Nicholas Apostoloff, Jerremy Holland, Fernando De la Torre,
Under review at NeurIPS 2026
"A Shapley-based metric that quantifies how much a VLM's answer is driven by the language prior instead of the image"
DSO: Direct Steering Optimization for Bias Mitigation
Yinong Oliver Wang*, Lucas Monteiro Paes*, Nivedha Sivakumar*, Masha Fedzechkina Donaldson, Barry Theobald, Luca Zappella, Nick Apostoloff,
CVPR 2026
"Mitigating social bias in vision language models by directly optimizing steering interventions inside the model"
Yinong Oliver Wang*, Lucas Monteiro Paes*, Nivedha Sivakumar*, Masha Fedzechkina Donaldson, Barry Theobald, Luca Zappella, Nick Apostoloff,
CVPR 2026
"Mitigating social bias in vision language models by directly optimizing steering interventions inside the model"
Is Your Model Fairly Certain? Uncertainty-Aware Fairness Evaluation for LLMs
Yinong Oliver Wang, Nivedha Sivakumar, Falaah Arif Khan, Rin Metcalf Susa, Adam Golinski, Natalie Mackraz, Barry Theobald, Luca Zappella, Nick Apostoloff,
ICML 2025
"Evaluating LLM fairness through the lens of model uncertainty rather than point predictions alone"
Yinong Oliver Wang, Nivedha Sivakumar, Falaah Arif Khan, Rin Metcalf Susa, Adam Golinski, Natalie Mackraz, Barry Theobald, Luca Zappella, Nick Apostoloff,
ICML 2025
"Evaluating LLM fairness through the lens of model uncertainty rather than point predictions alone"
On the Fine-Grained Planning Abilities of VLM Web Agents
Yinong Oliver Wang*, Surgan Jandial*, Andrea Bajcsy, Fernando De la Torre,
EMNLP Findings 2025
"A fine-grained evaluation of whether VLM web agents can understand temporal relations and assess plans"
Yinong Oliver Wang*, Surgan Jandial*, Andrea Bajcsy, Fernando De la Torre,
EMNLP Findings 2025
"A fine-grained evaluation of whether VLM web agents can understand temporal relations and assess plans"
Unsupervised Model Diagnosis
Yinong Oliver Wang, Eileen Li, Jinqi Luo, Zhaoning Wang, Fernando De la Torre,
arXiv preprint arXiv:2410.06243 (2024)
"An unsupervised model diagnosis framework to report bias and weakness of computer vision models"
Yinong Oliver Wang, Eileen Li, Jinqi Luo, Zhaoning Wang, Fernando De la Torre,
arXiv preprint arXiv:2410.06243 (2024)
"An unsupervised model diagnosis framework to report bias and weakness of computer vision models"
Domain Gap Embeddings for Generative Dataset Augmentation
Yinong Oliver Wang*, Younjoon Chung*, Chen Henry Wu, Fernando De la Torre,
CVPR 2024
"A plug-and-play embedding that captures a domain gap and steers generative models to augment datasets across it"
Yinong Oliver Wang*, Younjoon Chung*, Chen Henry Wu, Fernando De la Torre,
CVPR 2024
"A plug-and-play embedding that captures a domain gap and steers generative models to augment datasets across it"
More Publications
- Yinong Oliver Wang, Rwiddhi Chakraborty, Fan Bai, Cheng Zhang, Fernando De la Torre. "Towards Temporal Understanding in VLMs with a Closer Look at Single-Frame Bias". Under review at NeurIPS 2026.
- Runkai Zheng, Tanisha Gupta, Yinong Oliver Wang, Nicholas Apostoloff, Oncel Tuzel, Jerremy Holland, Fernando De la Torre. "Spectral Preconditioning of Text Embeddings". Under review at NeurIPS 2026.
- Ashish Asokan, Yinong Oliver Wang, Fernando De la Torre. "Model Gap Discovery in VLMs". In preparation.
- Falaah Arif Khan, Nivedha Sivakumar, Yinong Oliver Wang, Katherine Metcalf, Cezanne Camacho, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff. "Quantifying Risk of Epistemic Harm from the Use of AI Surrogates in Social Science Research". AI4GOOD @ ICML 2026.
- Runkai Zheng, Vishnu Asutosh Dasu, Yinong Oliver Wang, Haohan Wang, Fernando De la Torre. "Improving Noise Efficiency in Privacy-preserving Dataset Distillation". ICCV 2025.
- Falaah Arif Khan, Nivedha Sivakumar, Yinong Oliver Wang, Katherine Metcalf, Cezanne Camacho, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff. "Investigating Intersectional Bias in Large Language Models using Confidence Disparities in Coreference Resolution". COLM 2025.
- Rwiddhi Chakraborty, Yinong Oliver Wang, Jialu Gao, Runkai Zheng, Cheng Zhang, Fernando De la Torre. "Visual Data Diagnosis and Debiasing with Concept Graphs". NeurIPS 2024.
