YuQian Li

YuQian Li, Limeng Qiao, Lin Ma

Mask Diffusion-based Vision Language Models (MDVLMs) have achieved remarkable progress in multimodal understanding tasks. However, these models are unable to correct errors in generated tokens, meaning they lack self-correction capability. In this paper, we propose Recursive Introspection Mask Diffusion Vision Language Model (RIV), which equips the model with self-correction ability through two novel mechanisms. The first is Introspection Training, where an Introspection Model is introduced to identify errors within generated sequences. Introspection Training enables the model to detect not only grammatical and spelling mistakes, but more importantly, logical errors. The second is Recursive Inference. Beginning with the standard unmasking step, the learned Introspection Model helps to identify errors in the output sequence and remask them. This alternating ($\text{unmask}\rightarrow\text{introspection}\rightarrow\text{remask}$) process is repeated recursively until reliable results are obtained. Experimental results on multiple benchmarks demonstrate that the proposed RIV achieves state-of-the-art performance, outperforming most existing MDVLMs.

Yuqian Li, Yupei Du, Yufang Liu et al.

Language models excel in various tasks by making complex decisions, yet understanding the rationale behind these decisions remains a challenge. This paper investigates \emph{data-centric interpretability} in language models, focusing on the next-word prediction task. Using representer theorem, we identify two types of \emph{support samples}-those that either promote or deter specific predictions. Our findings reveal that being a support sample is an intrinsic property, predictable even before training begins. Additionally, while non-support samples are less influential in direct predictions, they play a critical role in preventing overfitting and shaping generalization and representation learning. Notably, the importance of non-support samples increases in deeper layers, suggesting their significant role in intermediate representation formation. These insights shed light on the interplay between data and model decisions, offering a new dimension to understanding language model behavior and interpretability.

Yuqian Li, Vincent Conitzer, Dmytro Korzhyk

Algorithms for computing game-theoretic solutions have recently been applied to a number of security domains. However, many of the techniques developed for compact representations of security games do not extend to {\em Bayesian} security games, which allow us to model uncertainty about the attacker's type. In this paper, we introduce a general framework of {\em catcher-evader} games that can capture Bayesian security games as well as other game families of interest. We show that computing Stackelberg strategies is NP-hard, but give an algorithm for computing a Nash equilibrium that performs well in experiments. We also prove that the Nash equilibria of these games satisfy the {\em interchangeability} property, so that equilibrium selection is not an issue.