偷懒了,哈哈

Problem G

Time Limit : 6000/2000ms (Java/Other)   Memory Limit : 65535/16384K (Java/Other)
Total Submission(s) : 56   Accepted Submission(s) : 19
Font: Times New Roman | Verdana | Georgia

Problem Description

“…黑黑的天空低垂/亮亮的繁星相随/虫儿飞虫儿飞/你在思念谁…”

I don’t know whether you could recall this soft song in your memory, but every time I listened to it, I thought about you. Shall it be several years since last time we met each other? I had given up every chance that I could tell you what I wanted you know. I think I was absolutely ridiculous.

We had lost contact with each other since that summer, I cannot remember how many times I wanted to make you a call and at last turned down the idea, I was such a coward that I even didn’t know how to start a conversation. I did not even know where you were! How can I be such a timid person?

Give thanks to god. Few days ago I finally heard about your recentness from one of my old classmates, and contact with you again! Though we’re thousands of miles away, I can be no more satisfied to chat with you. However, I know I may never have the courage to tell you what I am thinking, I hope I won’t disturb you too much. If only I am a firefly, I could fly near where you are. Even when your entire world has turned black, I always will be the light in the dark night.

Now here’s the problem, how tough will it be for a little firefly to cross thousands of miles to reach the destination? What is the possibility to survive? To make the problem simple, let’s consider the world as an undirected graph. The cities are vertexes. Little firefly can only fly from one city to another. However, a firefly is too small and too weak; it could easily be hurt on the way between two cities. Now I have the number of cities, the possibility being attacked between each pair of cities, the start city and the destination city, I want to know what the maximum possibility for the little bug to survive is. You know, one attack is deadly enough.

Input

The input file consists several test cases, each cases start with an integer N (2<=N<=300), which means the count of cities; the second line of the input is two integer S, D (1<=S, D<=N), S is the number of start city; D is the number of destination city. The cities are indexed from 1 to N. The next N lines, each line contains N non-negative real number that not larger than 1.The i-th input of the j+2-th line means the possibility being attacked on the way from City j to City i. End of file indicates the end of the input.

Output

For each test case, output one line with a real number which represents the maximum possibility. The answer should be rounded to forth digit after the decimal point.

Sample Input

2
1 2
0.0 0.01
0.01 0.0

Sample Output

0.9900



HINT
Huge input, use scanf to read data.
我本想用Dijkstra的,但是看到又2000MS呢,数据规模又小于是就碰碰运气,用了Floyd,没想到过了!!汗了
#include<iostream>
using namespace std;
double map[301][301];
int n,s,e;
void solve()
{
  for(int k=1;k<=n;k++)
   for(int i=1;i<=n;i++)
    for(int j=1;j<=n;j++)
    {
     if(map[i][j]<map[i][k]*map[k][j])
        map[i][j]=map[i][k]*map[k][j]; 
     }
   printf("%.4lf/n",map[s][e]);   
   
}
int main()
{
  double t;
  while(scanf("%d",&n)!=EOF)
   {
     scanf("%d %d",&s,&e);
     for(int i=1;i<=n;i++)
      for(int j=1;j<=n;j++)
      {  
       scanf("%lf",&t);
       map[i][j]=1-t;
      }
      solve();                       
   }  
   return 0;  
}
训练数据保存为deep_convnet_params.pkl,UI使用wxPython编写。卷积神经网络(CNN)是一种专门针对图像、视频等结构化数据设计的深度学习模型,在计算机视觉、语音识别、自然语言处理等多个领域有广泛应用。其核心设计理念源于对生物视觉系统的模拟,主要特点包括局部感知、权重共享、多层级抽象以及空间不变性。 **1. 局部感知与卷积操作** 卷积层是CNN的基本构建块,使用一组可学习的滤波器对输入图像进行扫描。每个滤波器在图像上滑动,以局部区域内的像素值与滤波器权重进行逐元素乘法后求和,生成输出值。这一过程能够捕获图像中的边缘、纹理等局部特征。 **2. 权重共享** 同一滤波器在整个输入图像上保持相同的权重。这显著减少了模型参数数量,增强了泛化能力,并体现了对图像平移不变性的内在假设。 **3. 池化操作** 池化层通常紧随卷积层之后,用于降低数据维度并引入空间不变性。常见方法有最大池化和平均池化,它们可以减少模型对微小位置变化的敏感度,同时保留重要特征。 **4. 多层级抽象** CNN通常包含多个卷积和池化层堆叠在一起。随着网络深度增加,每一层逐渐提取更复杂、更抽象的特征,从底层识别边缘、角点,到高层识别整个对象或场景,使得CNN能够从原始像素数据中自动学习到丰富的表示。 **5. 激活函数与正则化** CNN中使用非线性激活函数来引入非线性表达能力。为防止过拟合,常采用正则化技术,如L2正则化和Dropout,以增强模型的泛化性能。 **6. 应用场景** CNN在诸多领域展现出强大应用价值,包括图像分类、目标检测、语义分割、人脸识别、图像生成、医学影像分析以及自然语言处理等任务。 **7. 发展与演变** CNN的概念起源于20世纪80年代,其影响力在硬件加速和大规模数据集出现后真正显现。经典模型如LeNet-5用于手写数字识别,而AlexNet、VGG、GoogLeNet、ResNet等现代架构推动了CNN技术的快速发展。如今,CNN已成为深度学习图像处理领域的基石,并持续创新。 资源来源于网络分享,仅用于学习交流使用,请勿用于商业,如有侵权请联系我删除!
评论
成就一亿技术人!
拼手气红包6.0元
还能输入1000个字符
 
红包 添加红包
表情包 插入表情
 条评论被折叠 查看
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值