发现视觉嵌入可沿线性方向保留年龄、美学等排序属性，仅需少量样本即可恢复排序轴。
On the rankability of visual embeddings
- 通过线性投影检验嵌入是否保持属性顺序，定义为可排序性。
- 7个主流编码器在9个数据集上表现良好，仅2个极端样本即可恢复有效排序轴。
- 适用于向量数据库图像排序，推动无监督排序嵌入研究。
我们研究视觉嵌入模型是否能在线性方向上捕捉连续的序数属性，称其为“排序轴”。若将嵌入投影到此类轴上仍保持属性顺序，则称该模型对该属性具有可排序性。在7个流行编码器和9个包含年龄、人群密度、头部姿态、美学评分、时间新旧等属性的数据集上，我们发现许多嵌入本身具备可排序性。令人惊讶的是，只需极少数样本，甚至仅两个极端样本，便足以恢复有意义的排序轴，而无需全量标注监督。这些发现为向量数据库中的图像排序开辟了新应用场景，并激励对可排序嵌入的结构与学习机制进一步研究。代码已开源：https://github.com/aktsonthalia/rankable-vision-embeddings。
原文摘要 · Abstract (English)
We study whether visual embedding models capture continuous, ordinal attributes along linear directions, which we term _rank axes_. We define a model as _rankable_ for an attribute if projecting embeddings onto such an axis preserves the attribute's order. Across 7 popular encoders and 9 datasets with attributes like age, crowd count, head pose, aesthetics, and recency, we find that many embeddings are inherently rankable. Surprisingly, a small number of samples, or even just two extreme examples, often suffice to recover meaningful rank axes, without full-scale supervision. These findings open up new use cases for image ranking in vector databases and motivate further study into the structure and learning of rankable embeddings. Our code is available at https://github.com/aktsonthalia/rankable-vision-embeddings.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。