This paper first probes MCoT faithfulness by measuring how much the prediction changes when its visual and textual thoughts are intervened, and proposes Sufficient-Component Cause Model (SCCM) learning, which encourages the MCoT to generate sufficient yet minimal visual components that are independently capable of lead...
Zujing Liu, Jun-Wen Pan, Qi She et al.· arXiv.org· 15 citations· ⚡2
This work proposes MLLMEraser, an input-aware, training-free framework for test-time unlearning that consistently outperforms state-of-the-art MLLM unlearning baselines, achieving stronger forgetting performance with lower computational cost and minimal utility degradation.
A lightweight extension that turns any pretrained backbone with spatial feature maps into a saccadic classifier through glimpse-induced inference is proposed, which improves consistently over its backbones.
Johann Schmidt, S. Stober, Joachim Denzler et al.· 5 citations
A rule-to-image synthesis pipeline that extracts and expands puzzle rules from seed questions and generates executable rendering programs for puzzle assembly to tackle the challenge of image synthesis with grounded answers is proposed.
Offline RL (ORL) promises safe and sample-efficient deployment but existing methods rely on undocumented online interactions for hyperparameter tuning and lack reliable fully offline estimates of initial online performance. We introduce SOReL, a fully offline Bayesian model-based RL method that learns a posterior over...
Mattie Fellows, Clarisse Wibault, Uljad Berdica et al.· 0 citations
Bayesian optimization (BO) is a sequential decision-making tool widely used for optimizing expensive black-box functions. Recently, Large Language Models (LLMs) have shown remarkable adaptability in low-data regimes, making them promising tools for black-box optimization by leveraging contextual knowledge to propose hi...
Chih-Yu Chang, Milad Azvar, Chinedum Okwudire et al.· 0 citations
Scaling test-time compute is crucial for enhancing the reasoning capabilities of large language models (LLMs). Existing approaches typically employ reinforcement learning (RL) to maximize a verifiable reward obtained at the end of reasoning traces. However, such methods optimize only the final performance under a large...
Penghui Qi, Zichen Liu, Tianyu Pang et al.· 0 citations
Despite advancements in Computer-Aided Diagnosis (CAD) systems, breast cancer remains one of the leading causes of cancer-related deaths among women worldwide. Recent breakthroughs in Artificial Intelligence (AI) have shown significant promise in development of advanced Deep Learning (DL) architectures for breast cance...
Farnoush Bayatmakou, Reza Taleei, Milad Amir Toutounchian et al.· 0 citations
Neural graph collaborative filtering is a powerful tool for recommendation due to its superb capability of capturing the user-item interaction. Despite the substantial progress, the noisy nature of interaction data remains a significant challenge in collaborative filtering, especially for data collected from implicit f...
Xinyu He, Jose Sepulveda, Mostafa Rahmani et al.· 0 citations
Generative recommendation with semantic IDs (SIDs) represents each item as a sequence of discrete tokens and autoregressively predicts the next item's SID. Despite strong recommendation performance, repeated decoding over multiple SID positions incurs substantial inference cost. Knowledge distillation transfers knowled...
Fengshi Zhang, Xiyan Ding, Shuxiao Xie et al.· 0 citations
Single-document corpus poisoning under black-box access poses a severe threat to Retrieval-Augmented Generation (RAG). However, existing attacks predominantly rely on explicit factual refutations or imperative instruction overrides. Modern aligned LLMs often arbitrate such conflicts conservatively, rejecting overt cont...
Hang-Yi Zhao, Bo-Nan Zhang, Fu-Sen Guo et al.· 0 citations
Equation discovery solvers often use different data interfaces, equation representations and evaluation criteria, making their results hard to compare. We present LegoSR, a unified framework for algebraic symbolic regression and differential equation discovery. LegoSR integrates seven solvers through a common equation...
