Unverified50% confidenceFactExact time

Research indicates that appropriately applying data augmentation techniques such as RandAugment, MixUp, and CutMix on Tiny ImageNet can typically yield accuracy improvements of 3–8 percentage points.

1

Sources

50%

Confidence

Long-term

Relevance

9/1/2026

First Seen

Sources

51% Accuracy with Just 595K Parameters: A Lightweight and Interpretable Prototype Network in Practice

Related Entities

Tiny ImageNetconcept MixUpconcept CutMixconcept

Related Claims

UnverifiedCutMix 由 2019 年 Yun 等人提出，从一张图像裁剪矩形区域粘贴到另一张图像，标签按面积比例混合，在数据量有限的微调场景下通常能带来 1-3% 的精度提升68% similar Unverified现实的技术趋势是分层混合使用三种数据：用海量第一视角数据学习视觉表征，用UMI数据获得动作层面对齐，用少量高保真遥操作数据做策略微调67% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/837581

API

curl https://kongchang.com/api/v1/knowledge/claims/837581

MCP

get_claim(id=837581)