All you should know about translation equivariance/invariance in CNN

First published on Medium

Translation invariance and translation equivariance are two important concepts in convolutional neural networks (CNNs) related to the network’s ability to recognise objects regardless of their position within an image.

Create a custom dataset

To create a customized dataset in PyTorch, you should create a subclass of the torch.utils.data.Dataset class and implement the following methods:

article: true title: 算法: Binary Search icon: fa6-solid:repeat

• binary search，也称作二分法（half-interval search），每次划分一半进行下一步搜索，所以时间复杂度无非就是while循环的次数！
• 
  • maximum comparisons are dependent on the height of a tree logn，极端情况下我们需要查找所有n个elements得到target.
    • 第一次折半查找的为区间n，比较次数 c=1（中间值可能的个数为：2^c - 1 = 1， index: [8]）
    • 第二次折半查找的为区间n/2，比较次数 c=2（中间值可能的个数为：2^c - 1 = 3，index: [8, 4, 12]）
      • if target <= mid point: (left point, mid point-1) 即( 1, 7)
      • or target >= mid point: (mid point+1, right point) 即( 9, 15)
      • 一共可能的区间个数为2 ， 即( 1, 7) or (9, 15)
    • 第三次次折半查找的为区间n/4，比较次数 c=3（中间值可能的个数为：2^c - 1 = 7，index: [8, 4, 12, 2, 6, 10, 14]）
      • if target <= mid point: (left point, mid point-1)
      • or target >= mid point: (mid point+1, right point)
      • 一共可能的区间个数为4 ， 即( 1, 3) or (5, 7) or ( 9, 11) or (13, 15)
    • 第四次次折半查找的为区间n/8，比较次数 c=4（中间值可能的个数为：2^c - 1 = 15，index: [8, 4, 12, 2, 6, 10, 14, 1, 3, 5, 7, 9, 11, 13, 15]）
      • if target <= mid point: (left point, mid point-1)
      • or target >= mid point: (mid point+1, right point)
      • 一共可能的区间个数为8 ， 即 (1,1) or (3,3) or (5,5) or (7,7) or (9,9) or (11,11) or (13,13) or (15,15)
      • 数组只余一个元素无法再分，计算结束。
      • 此时中间值可能的个数=元素总个数len(nums), 算法经历了所有可能值
  • 总共有n个元素，每次查找的区间大小为n，n/2，n/4，…，n/2^c 且2^c_max - 1 = len(nums) 可得 c=log2(n+1), n+1为常数所以可以把它看作c=log2(n),（ c为比较的次数or循环的次数or二分树的高度），log2(n)中2可以省略所以时间复杂度为 O(logn)
    • minimum time of binary search: 1次比较得到target, O(1)
    • maximum time of binary search: log(n)次比较得到target, O(logn)

🖼️ 自制的流程图

1. ReAct Prompt

2. Pytorch DataLoader

Tools

AMiner 🌟

AMiner (aminer.org) aims to provide comprehensive search and mining services for researcher social networks. In this system, we focus on:

(1) creating a semantic-based profile for each researcher by extracting information from the distributed Web;

(2) integrating academic data (e.g., the bibliographic data and the researcher profiles) from multiple sources;

(3) accurately searching the heterogeneous network;

(4) analyzing and discovering interesting patterns from the built researcher social network. The main search and analysis functions in AMiner include:

You need to use some HTML in your Markdown. You could also use some text expander app to type it more quickly.

If you prefer video, here is my Youtube video! Video Link

Here is my Macro for Keyboard Maestro which automate applications or web sites, text or images, simple or complex, on command or scheduled. You can also use other software such as TextExpander, ProKeys chrome extension, Alfred Snippets etc.

Keep you updated with state of the art technology is really important in the Machine learning and deep learning field.

让您了解最先进的技术在机器学习和深度学习领域非常重要。

There are some tips from Andre NG on how to read research papers. https://www.youtube.com/watch?v=733m6qBH-jI

Wanna play around with the code? Link

    import os
    import numpy as np
    import argparse
    import cv2
    import matplotlib.pyplot as plt

Tools: 工具：

@raindrop_io : bookmark manager/read later/highlighting@raindrop_io ： 书签管理器/稍后阅读/高亮显示 @IFTTT : automation&workflow@IFTTT ： 自动化&工作流 You can customize the automation on ifttt. I normally choose the collection folder and tags.您可以在 ifttt 上自定义自动化。我通常会选择收藏文件夹和标签。