I recently wrote a perceptron that determines whether a point is above or below a line. I'm trying to write a multi-layered ANN so I can use it to solve more complicated problems than "is this point above/below this line?"
Here's my source code: [URL] ....
Considering that a A XOR B is simply (NOT A AND OR (A AND NOT , and both of those are linearly separable...
I can't seem to find the logic error in my perceptron. The perceptron's feedback function pushes back a bias of 1.0 into the input vector, and calculates the dot product of the input vector and it's weight vector, and returns 1 if the dot product is > 0, and -1 otherwise. This will determine whether the inputted point is above or below the hyperplane given by the functor.
The training function instantiates a trainer object , which feeds it a vector random numbers, along with the correct answer (in this case, whether the point is above or below the hyperplane given by a functor). It then calls the feedback function, and subtracts its answer from the correct answer, and updates the weights by adding the input vector to the weight vector, and multiplying the weight vector by the error, and by a correction factor, in this case 0.01.
Typedef.hpp
Code:
#ifndef TYPEDEF_HPP
#define TYPEDEF_HPP
#include <vector>
#include <random>
typedef std::vector<float> FloatV;
typedef std::random_device RNG;
typedef std::uniform_int_distribution<> UID;
[Code] ....
When I use my perceptron, it classifies everything appropriately, except it always says points that are exactly on the line are above it.
View 9 Replies View RelatedSuppose multiple threads are are trying to insert into a STL Map, both are trying to insert with same key. As per my understanding it will not cause any issue till you invalidate the iterator. Here as per me it will not invalidate the iterator.
View 2 Replies View RelatedWhen i am entering single character instead of 'miz' then it is working great. But when i am using multi character like 'miz' it is showing error output. i did this program with c++ it is working but not working in c.When i am running this program it is showing {warning] multi character constant.
Code:
#include<stdio.h>
#include<string.h>
main()
{
char a;
printf("Please Enter your First Name: ");
scanf("%s",&a);
[Code]....
I have pseudo code for the minimax algorithm but i was wondering what i would need to add to make it work with many players (n players)?
score minimax(state s, int n)
if (n = 0 OR s is a terminal node) {
if win position for program
return +inf;
if loss position for prg
[Code] ....
i want to make a two-dimensional array that is consisted by objects but i dont want to initialize the objects when i make the array.. i just want to make an array that each of his "cells" has the size of an object.For example,
class Human{
int hm=5;
...
};
and now i want to make a two dimensional array that each "cell" has the size of an Human but not a Human itself.
So basically, I started out with each wizard == 1 winform, but then I found another way to do it by making the content of each wizard step a user control, then say, on initial deployment, it load usercontrol1, then when i click next, the panel hide usercontrol1 for usercontrol2 and so forth. Would it be feasible to create all usercontrols (all the wizard step) and add them in an array, then i can load them by index?
View 7 Replies View RelatedI am new to Path Finding so I just know BFS, DFS, Djisktra, and Basic A*
I've searched for a few multi agent pathfind paper here and there. But I can't seem to understand them..
I am path finding in a grid map, 4 ways neighbor movement
My map isn't so big it's 300 x 300 up to 400 x 400
Most of the map is open space, I would say that at worst case 4/5 of the map is open space.
And there is probably around 100 agents.
I've tried running BFS 100 times and it takes around 200ms on average
Changing BFS to A* which I am thinking right now will definitely cost lesser time
But are there better ways in doing this ? and also Does running Single Agent A* multiple times doesn't cause problem for other agent ? perhaps causes deadlocks ? ( Considering that the path that have been determine for agents before the current searched is considered blocked )
#include<iostream.h>
#include<conio.h>
main()
[Code]....
I tried to make the output look like this --> ENTER YOUR NAME:Victor Collin ENTER YOUR COURSE:BSComSci
but everytime I entered my name the course went to the next line.How to make the output in the same?
I made a C++ Wrapper on top of ODBC.
I got a Connection, and a Statement class.
The Connection class holds the Database Connection ODBC Handle.
The Statement class holds the Statement Handle, there can be multiple per Connection. It is made and destroyed each time a query must be done on the DB.
Now I'm just wondering.
I have a multithreaded application.
Is keeping one Connection for one DB enough for the whole application? Where the different threads will make Statements based on that one Connection.Or is it better to keep a Connection per thread?
Here's the line of code with the warning
Code: if((addressbook[3][num][4]=='-')&&(strlen(addressbook[3][num])=='10'))
Or what might be the possible causes of such a warning?
I want to have a multi user chat over TCP/IP. I already have the code for both the server and the client and so far I am able to send messages from the client to the server. Now I want to make it a multi user chat. I am executing the codes using Visual studio's Command prompt(not Windows Command Prompt). I have read somewhere that we have to use select() function.
View 2 Replies View Relatedthis last few days I've been coding this one particular file
Hero.cpp
namespace Hero {
namespace {
// all data here
}
}
then the code grew a bit to about 700 line
Then now I want to implement hero skill system It needs the access to data inside the unnamed namespace where I put just about everything.. but it's unnamed namespace, it's only valid within one file and I should hide all that data from Hero interface in Hero.h How should I do this ?
I have to delete an object at a time that I don't know it's held by another thread(s) or not. For example
static void foo(CLASS* ref){
pthread_mutex_lock(&l);//[POINT-A]
//using ref
//do something thread safe
pthread_mutex_unlock(&l);
delete ref;// how can I find out there is no any thread held at [POINT-A]
}
I need to create subfunctions to do basic Matrix mathematics (addition, subtraction, etc.) I need to be able to pass bot of my Matrices to subfunctions. I started with the addition sub function and I cant get the code to run. I keep getting expected primary-expression before ']' token error for line 75.
#include <iostream>
#include <stdio.h>
using namespace std;
[Code].....
I have always written like a>='0'&&a<='9'&&a>='a'&&a<='z' in loops etc, but no more. Basically add whatever you want to condition, and if you want point a to point b just separate them with a '-' sign. Simply
while(!isCharInside(x,"|a-zA-Z0-9.*/=|-|")){/*code*/}
bool isCharInside(char check,string condition="a-zA-Z*/+=|-|0-9"){
for(unsigned int x=0,z=getLenght(condition);x<z;++x){
if(condition[x+1]=='-'&&(condition[x]!='|'&&condition[x+2]!='|')){
if(condition[x]>condition[x+2])swap(condition[x],condition[x+2]);
for(;condition[x]<=condition[x+2];++condition[x]){if(condition[x]==check)return true;};x+=2;}
else if(check==condition[x])return true;
}
return false;}
I am working on the game of fifteen.Yes this is home work. I was wondering if it is possible to save the indexes of an element in a two dim array. I need to find the position of the space in the two dim array save it, find the position of the tile which is passed in as an int and swap the two if they are adjacent. Here is my code
Edit: I can do this by saving i and j in separate int variables but is there a more elegant way of doing this
int board[MAX][MAX];
void init();
void swap(int* lhs, int* rhs);
[Code].....
I'm trying to union eleven tables to call out data. Parent table is 'Events', child tables are 'SR1Laptimes', 'SR2Laptimes' and so on (there are ten SR... tables). Primary key in all tables is EventName. Parent/Child relationship is Events.EventName/SR1Laptimes.EventName etc All tables that start with SR have the same Schema. I'm trying to call out MIN(Q1) across all table but first need to Union them I believe. Here is my code.
myCommand.CommandText = "SELECT MIN(Q1), MIN(Q2), MIN(Q3), MIN(Q4), MIN(LaptimesMinutes), MIN(LaptimesSeconds) FROM (SELECT * FROM Events UNION ALL SELECT * FROM SR1Laptimes UNION ALL SELECT * FROM SR2Laptimes) WHERE (Events.Track = @track) AND (Events.Number = @number) AND (Events.Rider = @rider)";
myCommand.Parameters.AddWithValue("@track", analysisTrackComboBox.Text);
myCommand.Parameters.AddWithValue("@number", analysisNumberComboBox.Text);
myCommand.Parameters.AddWithValue("@rider", analysisRiderComboBox.Text);
I delete rows in datagridview using chcekbox column. I write the following code:
private void btndelpay_Click(object sender, EventArgs e) {
DialogResult dr = MessageBox.Show("delete row", "do you want to delete",
MessageBoxButtons.YesNo, MessageBoxIcon.Question, MessageBoxDefaultButton.Button2);
if (dr == DialogResult.Yes)
[Code] ......
my problem is /how to umcheck from rows in datagridview in chechboxes column when the user select dialogResult.NO
Here is the code,
Code:
void foo() {
Acquiring lock
do something...
Func();
Releasing Lock
}
If the function Func throws an exception, there is potential deadlock issue. Then I handle exception like this,
Code:
void foo() {
Acquiring lock
do something...
try{
Func();
[Code] ....
Is this a good practice? I wonder how I can apply RAII in handling exception here.
I have a compressor that takes a file and first compresses it to an intermediate file "temp.lz", before compressing it to the final format. Immediately after that the file be removed by calling c's remove(char*). The problem I am afraid of is if calling this compressor from different processes is safe because of the intermediate file created which has the same name for all. (say temp.lz). Will I have problems when callinga sytem call from different process something like: system ("compress -i test.txt -o test.z") ?
I was thinking of rewriting the compressor to avoid the use of the temporary file but that is a bit awkward. I just need to use a temporary stream (file) different for each process to avoid race conditions.
If I wanted an int connected to a float connected to a string connected to a POD, would all I have to do is :
Code:
template<class T, class X>
struct node {
T data;
node<X> *next;
};
Or would I be forced into using polymorphism?
In our Qt application, we have to load several files on application start-up starting from a root directory.
We are recursively scanning the directories and loading the files.
The total time it takes is about 12 seconds. Can we reduce this time if we use multi-threading?
I have heard that multi-threading does not work everywhere and it increases code complexity and debugging issues.
Would multi-threading solve the above problem? We want it to be platform independent (Mac, Linux, Windows).
I have this multi-dimensional array. There are 12 rows, each represents the months, and 2 columns, the 1st column represents the lowest temperature and the 2nd column represents the highest temperature.
I need to find the lowest temperature in only the first column and the highest in only the 2nd column.
This takes in the temperatures. LOHI = 2, MONTHS = 12, LOW = 0, HIGH = 1
void PopulateTemp(int temp[][LOHI])
{
for(int month = 0; month < MONTHS; month++)
{
cout << "Month " << month+1 << endl;
[Code]......
My professor said we only needed 1 for loop as you only need to read the data from one column in each?
I'll just tell you in short how my Problem looks like: I should implement a New Class in a SourceCode i didnt write myself. The source code is extremely sized (i think approx >100.000 Lines of Code), so i dont want to change too much in it in order to get my Implementation done.
MY problem looks simplicified like that: Starting from 3 classes and my new class the pseudo-code looks like that:
Class1(){
float* m_CalibX, m_CalibY;
.. }
Class2(){
char* m_ImageData[];
[Code] .....
So, i need Parameters from 3 different classes to insert in my NewClass. The 3 Classes dont know anyting about each other. So, i need to implement a Class-Instance from Type NewClass which is known by the other 3 Classes. I did solve it in this way:
//ClassInstance.h
#include "NewClass.h"
static NewClass ClassInstance
I just wrote a headerfile with a class-instance which is getting included by the other 3 Classes. So they all know the same Instance and writing their Parameters into it. Is this a decent solution or could it happen to get bugs/ logical mistakes with it?