本文介绍了机器学习中的一些重要概念：1. permutation-invariant与convolutional neural networks的关系；2. Region of Interest Pooling的作用，用于从不同大小的输入获取固定尺寸特征图；3. 'Ground Truth'的概念，它在不同场景中的含义和获取方式；4. 拉普拉斯矩阵的定义及其在图论中的应用。通过对这些概念的理解，有助于深入学习机器学习的基础知识。
1 .permutation-invariant：空间不相关的
local permutation-invariant aggregation:局部的无空间相关性的聚集
例子：convolution neural network 就是 permutation-related.
词汇来自 https://arxiv.org/abs/1902.06673
2. Region of Interest Pooling
Its purpose is to perform max pooling on inputs of nonuniform sizes to obtain fixed-size feature maps (e.g. 7×7).
参考：https://deepsense.ai/deep-learning-for-satellite-imagery-via-image-segmentation/
3 . Gournd True
"Ground truth" means a set of measurements that is known to be much more accurate than measurements from the system you are testing.
For example, suppose you are testing a stereo vision system to see how well it can estimate 3D
最低0.47元/天 解锁文章
扫一扫
686
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
