---
id: 20260724-T0-05
title: "原生多维次二次算子：输入依赖的长卷积追踪全局关系"
title_en: "Native Multi-Dimensional Subquadratic Operators Use Input-Dependent Long Convolutions"
url: https://ai.daily.yangsir.net/daily/20260724-T0-05
issue_date: 2026-07-24
publish_date: 2026-07-23T04:00:00.000Z
category: research
source_name: "arXiv cs.LG (ML)"
source_url: https://arxiv.org/abs/2607.19378
---

# 原生多维次二次算子：输入依赖的长卷积追踪全局关系

arXiv 新研究提出一种基于输入依赖长卷积的原生多维次二次算子，在保持计算效率的同时实现全局感受野和输入依赖性。传统方法在处理图像、体积等多元数据时需要展平或使用固定卷积核，无法兼顾全局与灵活。该方法可直接操作多维数据，为高效视觉和3D模型提供了新的架构思路。

## English Version

**Native Multi-Dimensional Subquadratic Operators Use Input-Dependent Long Convolutions**

A new arXiv paper introduces native multi-dimensional subquadratic operators using input-dependent long convolutions, achieving global receptive fields and input dependency without flattening data. Unlike standard convolutions or recurrent models, it works directly on multi-dimensional tensors like images and volumes. This opens new architectural directions for efficient vision and 3D models.

---

**来源**：[arXiv cs.LG (ML)](https://arxiv.org/abs/2607.19378)

**详情页**：https://ai.daily.yangsir.net/daily/20260724-T0-05

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*