<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[snow_maple521的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/snow_maple521</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; snow_maple521]]></copyright><item><title><![CDATA[机器学习之——诈骗号码识别]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/134789960</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/134789960</guid><author>snow_maple521</author><pubDate>Mon, 04 Dec 2023 17:58:14 +0800</pubDate><description><![CDATA[主要采用sklearn和pandas对电话号码诈骗识别进行建模和对一些用户数据进行预处理]]></description><category></category></item><item><title><![CDATA[Pandas loc和iloc 以及 pivot和melt详解]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/134777913</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/134777913</guid><author>snow_maple521</author><pubDate>Mon, 04 Dec 2023 15:04:07 +0800</pubDate><description><![CDATA[pandas数据筛选 iloc loc  数据格式变换：pivot，melt]]></description><category></category></item><item><title><![CDATA[python可视化之——seaborn简单图绘制指南]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/134707718</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/134707718</guid><author>snow_maple521</author><pubDate>Fri, 01 Dec 2023 18:26:43 +0800</pubDate><description><![CDATA[该文主要介绍python下的seaborn绘图库，包含了大部分seaborn中的图形绘制。]]></description><category></category></item><item><title><![CDATA[python合并excel]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/134703195</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/134703195</guid><author>snow_maple521</author><pubDate>Thu, 30 Nov 2023 10:13:14 +0800</pubDate><description><![CDATA[摘要：python实现对多个excel合并成一个excel，1 实现简单合并，2 实现vlookup功能。]]></description><category></category></item><item><title><![CDATA[chatgpt catgpt]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/129816180</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/129816180</guid><author>snow_maple521</author><pubDate>Tue, 28 Mar 2023 16:00:11 +0800</pubDate><description><![CDATA[如何使用gpt]]></description><category></category></item><item><title><![CDATA[linux安装mysql]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/127745717</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/127745717</guid><author>snow_maple521</author><pubDate>Tue, 08 Nov 2022 10:38:43 +0800</pubDate><description><![CDATA[linux安装hive-3.1.2和安装mysql]]></description><category></category></item><item><title><![CDATA[CentOS 网络配置之ping不通]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/127728866</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/127728866</guid><author>snow_maple521</author><pubDate>Mon, 07 Nov 2022 12:44:33 +0800</pubDate><description><![CDATA[centos网络配置]]></description><category></category></item><item><title><![CDATA[python读取json文件报错]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/123178420</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/123178420</guid><author>snow_maple521</author><pubDate>Mon, 28 Feb 2022 11:25:09 +0800</pubDate><description><![CDATA[在跑目标检测任务时，python读取annotation里instance_train2014报如下错：

ValueError: Unterminated string starting at: line 1 column 18194403 (char 18194402)

ValueError: Expecting object: line 1 column 49012736 (char 49012735)

ValueError: Expecting object: line 1 column 11.]]></description><category></category></item><item><title><![CDATA[python对比两个excel表 【超简单】]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/117478781</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/117478781</guid><author>snow_maple521</author><pubDate>Wed, 02 Jun 2021 18:48:43 +0800</pubDate><description><![CDATA[注：由于我们不是专门做数据处理与分析的，因此懂的也少，所以编写的python程序是简单易懂的，如果存在bug或者可以有更好的方法，欢迎大佬指出!!!

主要思路：
就是将excel表中每条数据变成一个字典，然后添加到列表中，组成一个元素为字典类型的列表（由于数据不可公开，所以做了马赛克）。如下：
excel数据：表1
​
excel数据：表2

可以看到每个excel表数据很多，都几千条，如果一个个看很费时。
import openpyxl
# 加载第一个Excel
data_raw= openpyxl..]]></description><category></category></item><item><title><![CDATA[知识蒸馏之手写体识别]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/116007538</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/116007538</guid><author>snow_maple521</author><pubDate>Thu, 22 Apr 2021 11:56:45 +0800</pubDate><description><![CDATA[知识蒸馏

本文主要是根据该网站视频（https://www.bilibili.com/video/BV1s7411h7K2?t=906）进行总结，如有理解误差，望批评指点

1. 首次提出
首次提出：https://arxiv.org/pdf/1503.02531.pdf
作者的动机是想找到一个方法，把多个模型的知识提炼给单个模型。
虽然现在很多分类模型都采用交叉熵衡量预测值与真实值，然而真实值采用的one-hot向量所能提共的信息没有概率分布多。
原理：概率分布比onehot更能提供信息-暗知识。
lo]]></description><category></category></item><item><title><![CDATA[RuntimeError: CUDA out of memory. Tried to allocate 22.00 MiB (GPU 0； 10.76 GiB total capacity； 9.61]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/115427827</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/115427827</guid><author>snow_maple521</author><pubDate>Sun, 04 Apr 2021 13:30:29 +0800</pubDate><description><![CDATA[RuntimeError: CUDA out of memory. Tried to allocate 22.00 MiB (GPU 0; 10.76 GiB total capacity; 9.61 GiB already allocated; 4.50 MiB free; 332.12 MiB cached)
有时遇到这种情况，怎么改batchsize都还是会报内存溢出错误，nVidia-smi时发现gpu很多空闲，这时出现的原因极可能就是中间变量数据导致的，这时可以检查loss，acc，pred等中间]]></description><category></category></item><item><title><![CDATA[【论文阅读笔记】SCR: Self-Critical Reasoning for Robust Visual Question Answering]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/115113737</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/115113737</guid><author>snow_maple521</author><pubDate>Tue, 23 Mar 2021 17:13:19 +0800</pubDate><description><![CDATA[论文地址：https://arxiv.org/pdf/1905.09998v3.pdf
项目地址：https://github.com/jialinwu17/Self_Critical_VQA
摘要

Visual Question Answering (VQA) deep-learning systems tend to capture superfi-cial statistical correlations in the training data because of strong language]]></description><category></category></item><item><title><![CDATA[关于pytorch复现模型的一些报错总结]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/114920107</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/114920107</guid><author>snow_maple521</author><pubDate>Wed, 17 Mar 2021 20:13:44 +0800</pubDate><description><![CDATA[1.RuntimeError: One of the differentiated Tensors does not require grad
关于这个报错的意思是：有一个参数不需要计算导数
此torch.autograd.grad函数参数如下：outputs，inputs,函数功能是求outputs关于inputs的导数，此处的inputs需要加requires_grad_()定义：
v = Variable(v).to(device).requires_grad_()

2. RuntimeError:]]></description><category></category></item><item><title><![CDATA[pytorch安装进度为0解决方法]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/114591485</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/114591485</guid><author>snow_maple521</author><pubDate>Tue, 09 Mar 2021 17:35:18 +0800</pubDate><description><![CDATA[在终端添加以下源

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anac]]></description><category></category></item><item><title><![CDATA[视觉问答VQA论文近期总结]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/111564037</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/111564037</guid><author>snow_maple521</author><pubDate>Tue, 22 Dec 2020 19:02:31 +0800</pubDate><description><![CDATA[声明：全是结合论文加自己简单理解，可能存在理解错误，望见谅，不足很多没写，欢迎补充。

论文笔记总结
1. 写在前面
模型的实验准确度对比如下：





Y/N
Num
Other
All
test-std All
年份




SAN
79.3
36.6
46.1
58.7
58.9
2016


H-COA
79.7
38.7
51.7
61.8
62.1
2017


MUTAN
85.14
39.81
58.52
67.42
67.36
2017


BAN
85.42
54.04
60.52
.]]></description><category></category></item><item><title><![CDATA[Detecting Visual Relationships with Deep Relational Networks]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/111127921</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/111127921</guid><author>snow_maple521</author><pubDate>Sun, 13 Dec 2020 18:51:00 +0800</pubDate><description><![CDATA[用深度关系网络检测视觉关系
物体之间的关系在图像理解中起着重要的作用，以前的研究方法将“关系”作为一个分类问题，每一种关系类型（如：‘骑’）或每个不同的视觉短语（如‘人-骑-马’）作为一个类别。但这种存在缺陷，例如视觉外观多样，视觉短语多样。针对此，本文建立一个综合框架来解决这个问题。框架的核心就是 深度关系网络
1. 引言
早期将视觉关系视为一个分类任务，考虑将对象和关系谓词的不同组合作为一个类，但是会遇到很多不平衡类，例如视觉基因组中有超过75k不同的视觉短语，每个短语的样本数量从少量到超过10k不等，]]></description><category></category></item><item><title><![CDATA[图卷积网络在文本和视频的关系对齐中的应用]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/111088810</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/111088810</guid><author>snow_maple521</author><pubDate>Sun, 13 Dec 2020 11:34:08 +0800</pubDate><description><![CDATA[1.写在前面
原文来自知网《基于视觉-文本关系对齐的跨模态视频片段检索》
这是一篇关于利用文本进行视频片段检索的论文。
2.论文要点
论文总结了前人研究工作的缺点，并借助图结构在表达关系时的强大能力与图卷积网络的近年发展，，提出了跨模态关系对齐的图卷积框架 CrossGraphAlign，
具体就是CrossGraphAlign首先为查询文本和待检索视频分别生成文本关系图和视觉关系图。接着CrossGraphAlign中的视觉-文本关系对齐的图卷积网络试图匹配一段时间内的文本关系图和视觉关系图。最后基于匹配]]></description><category></category></item><item><title><![CDATA[从linux服务器配置到远程访问]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/109801762</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/109801762</guid><author>snow_maple521</author><pubDate>Tue, 08 Dec 2020 18:08:53 +0800</pubDate><description><![CDATA[1.下载NVIDIA所需驱动

首先输入nvidia-smi便知道nvidia的驱动是否安装，图中提示没有安装。可按如下步骤安装

按它给的命令下载相应驱动


图中的sudo apt install nvidia-340
图中的sudo apt install nvidia-utils-390


按ubuntu推荐的驱动直接安装


查看显卡型号：输入：ubuntu-drivers devices
自动安装：sudo ubuntu-drivers autoinstall

执行好后就直接reboot.重]]></description><category></category></item><item><title><![CDATA[【结果分析】之murel项目结果分析]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/110001560</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/110001560</guid><author>snow_maple521</author><pubDate>Mon, 23 Nov 2020 18:09:19 +0800</pubDate><description><![CDATA[1.murel:视觉问答VQA中的多模态关系推理

项目介绍
本次结果是第20个epoch,采用的是验证集的results文件，由于中途断了，导致没有执行测试集test.

2.结果分析

我们评估采用的使官网提供的eval,

2.1 结果
Overall Accuracy is: 64.50
Per Question Type Accuracy is the following:
none of the above : 58.44
what are the : 50.41
what is : 43.08]]></description><category></category></item><item><title><![CDATA[【结果分析】之block项目]]></title><link>https://blog.youkuaiyun.com/snow_maple521/article/details/109700919</link><guid>https://blog.youkuaiyun.com/snow_maple521/article/details/109700919</guid><author>snow_maple521</author><pubDate>Mon, 23 Nov 2020 16:57:25 +0800</pubDate><description><![CDATA[block

vqa项目介绍

1.vqaEval
# This code is based on the code written by Tsung-Yi Lin for MSCOCO Python API available at the following link: 
# (https://github.com/tylin/coco-caption/blob/master/pycocoevalcap/eval.py).
import sys
import re

class VQAEval:
	de]]></description><category></category></item></channel></rss>