paper-with-me

홈 › Papers

Removing Word-Level Spurious Alignment between Images and Pseudo-Captions in Unsupervised Image Captioning

2021-04-28 · EACL 2021 2 · Ukyo Honda, Yoshitaka Ushiku, Atsushi Hashimoto, Taro Watanabe, Yuji Matsumoto

Unsupervised image captioning is a challenging task that aims at generating captions without the supervision of image-sentence pairs, but only with images and sentences drawn from different sources and object labels detected from the images. In previous work, pseudo-captions, i.e., sentences that contain the detected object labels, were assigned to a given image. The focus of the previous work was on the alignment of input images and pseudo-captions at the sentence level. However, pseudo-captions contain many words that are irrelevant to a given image. In this work, we investigate the effect of removing mismatched words from image-sentence alignment to determine how they make this task difficult. We propose a simple gating mechanism that is trained to align image features with only the most reliable words in pseudo-captions: the detected object labels. The experimental results show that our proposed method outperforms the previous methods without introducing complex sentence-level learning objectives. Combined with the sentence-level alignment method of previous work, our method further improves its performance. These results confirm the importance of careful alignment in word-level details.

📄 PDF Abstract BibTeX arXiv:2104.13872

Code (1)

ukyh/RemovingSpuriousAlignment 공식 구현 pytorch

Tasks

Image Captioningimage-sentence alignmentObjectSentence

Similar Papers 제목 키워드 기반

Unsupervised Identification and Removal of Spurious Correlations During Fine-Tuning

2026-05-26 · Ciarán M. Gilligan-Lee, Joseph Egan, Yuchen Zhu, Michael O'Riordan arxiv

Fine-tuning a pretrained language model on a curated dataset can produce spurious correlations between the fine-tuning task and unintended latent factors -- such as misaligned personas or political slant -- that the cura…

Dissecting the Failure of Invariant Learning on Graphs

2024-11-05 · Qixun Wang, Yifei Wang, Yisen Wang, Xianghua Ying

Enhancing node-level Out-Of-Distribution (OOD) generalization on graphs remains a crucial area of research. In this paper, we develop a Structural Causal Model (SCM) to theoretically dissect the performance of two promin…

Does SGD really happen in tiny subspaces?

2024-05-25 · Minhak Song, Kwangjun Ahn, Chulhee Yun

Understanding the training dynamics of deep neural networks is challenging due to their high-dimensional nature and intricate loss landscapes. Recent studies have revealed that, along the training trajectory, the gradien…

Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-tuning

2025-03-14 · YiWei Chen, Yuguang Yao, Yihua Zhang, Bingquan Shen 외

Recent vision-language models (VLMs) have made remarkable strides in generative modeling with multimodal inputs, particularly text and images. However, their susceptibility to generating harmful content when exposed to u…

Machine Unlearning

RE-SORT: Removing Spurious Correlation in Multilevel Interaction for CTR Prediction

2023-09-26 · Song-Li Wu, Liang Du, Jia-Qi Yang, Yu-Ai Wang 외

Click-through rate (CTR) prediction is a critical task in recommendation systems, serving as the ultimate filtering step to sort items for a user. Most recent cutting-edge methods primarily focus on investigating complex…

Click-Through Rate PredictionRecommendation SystemsSelection bias