Zhaohua Zhang

Chenchen Zhao, Jianhuan Zhuo, Muxi Chen et al.

Composed image retrieval (CIR) requires multi-modal models to jointly reason over visual content and semantic modifications presented in text-image input pairs. While current CIR models achieve strong performance on common benchmark cases, their accuracies often degrades in more challenging scenarios where negative candidates are semantically aligned with the query image or text. In this paper, we attribute this degradation to focus imbalances, where models disproportionately attend to one modality while neglecting the other. To validate this claim, we propose FBCIR, a multi-modal focus interpretation method that identifies the most crucial visual and textual input components to a model's retrieval decisions. Using FBCIR, we report that focus imbalances are prevalent in existing CIR models, especially under hard negative settings. Building on the analyses, we further propose a CIR data augmentation workflow that facilitates existing CIR datasets with curated hard negatives designed to encourage balanced cross-modal reasoning. Extensive experiments across multiple CIR models demonstrate that the proposed augmentation consistently improves performance in challenging cases, while maintaining their capabilities on standard benchmarks. Together, our interpretation method and data augmentation workflow provide a new perspective on CIR model diagnosis and robustness improvements.

Zhaohua Zhang, Jianhuan Zhuo, Muxi Chen et al.

The CLIP model has become a cornerstone of large-scale retrieval systems by aligning text and image data in a unified embedding space. Despite its simplicity and efficiency, CLIP struggles when applied to tasks whose input distributions diverge from its training corpus, such as queries with multilingual, long-form, or multimodal differences. To avoid costly retraining, existing methods mainly adopt query-rewriting strategies with large language models (LLMs), aiming to mitigate distribution gaps at the query level. However, due to the lack of supervision signals, LLMs fail to generate the optimal one that fits the training distribution. We address this challenge with GRAPE (Grouped Ranking-Aware Policy Optimization Enhancement), a plug-and-play enhancement approach that incorporates ranking signals into retrieval-guided query rewriting with LLMs. Intuitively, GRAPE proposes to leverage GRPO to bridge distributional differences -- including length, multilingual, and modality shifts -- by transforming queries into forms better aligned with the retriever's training distribution. However, our preliminary experiment finds that naively finetuning LLM with similarity scores can lead to score inflation, where nearly all candidates are assigned unexpectedly high scores regardless of their true relevance. To address score inflation, we propose a corpus-relative ranking-based reward, which explicitly aligns optimization with ranking metrics while suppressing spurious score inflation. Extensive experiments demonstrate that GRAPE consistently improves retrieval performance under distributional shifts -- including multilingual differences (Flickr30k-CN, CVLUE, XM3600), length differences (Wikipedia), and multimodal differences (CIRR) -- achieving an average improvement of 4.9\% in Recall\@10. The code is available at https://github.com/Chinese0123456/GRAPE.git

Muxi Chen, Zhaohua Zhang, Chenchen Zhao et al.

Static benchmarks have provided a valuable foundation for comparing Text-to-Image (T2I) models. However, their passive design offers limited diagnostic power, struggling to uncover the full landscape of systematic failures or isolate their root causes. We argue for a complementary paradigm: active exploration. We introduce FailureAtlas, the first framework designed to autonomously explore and map the vast failure landscape of T2I models at scale. FailureAtlas frames error discovery as a structured search for minimal, failure-inducing concepts. While it is a computationally explosive problem, we make it tractable with novel acceleration techniques. When applied to Stable Diffusion models, our method uncovers hundreds of thousands of previously unknown error slices (over 247,000 in SD1.5 alone) and provides the first large-scale evidence linking these failures to data scarcity in the training set. By providing a principled and scalable engine for deep model auditing, FailureAtlas establishes a new, diagnostic-first methodology to guide the development of more robust generative AI. The code is available at https://github.com/cure-lab/FailureAtlas

Ju Yang, Jiancong Tong, Rebecca J. Stones et al.

When two terms occur together in a document, the probability of a close relationship between them and the document itself is greater if they are in nearby positions. However, ranking functions including term proximity (TP) require larger indexes than traditional document-level indexing, which slows down query processing. Previous studies also show that this technique is not effective for all types of queries. Here we propose a document ranking model which decides for which queries it would be beneficial to use a proximity-based ranking, based on a collection of features of the query. We use a machine learning approach in determining whether utilizing TP will be beneficial. Experiments show that the proposed model returns improved rankings while also reducing the overhead incurred as a result of using TP statistics.