Project Neural Network #0
Difficulty: easy
Time: 1h
Languages: any, use a NN framework
======================
Train a Neural Network to perform a modulo operation, to get used to Neural Networks.

#input #output #math #decimal #NeuralNetwork

Solutions to @linuxer4fun



https://t.me/prograchallenges

Project Neural Network #1
Difficulty: easy
Time: 10 minutes
Languages: any (pick one now, you will stay with it for the entire project)
===============================
Program two functions.
-) A sigmoid
-) A sigmoid's derivative

#input #output #math #NeuralNetwork #function

Solutions to @Linuxer4Fun
Documentation: https://en.wikipedia.org/wiki/Sigmoid_function



https://t.me/prograchallenges

Project Neural Network #2
Difficulty: medium
Time: 1-2h
Languages: The one you picked
===========================
Program a single perceptron that can take a quantity of inputs (n) and has a single output.
Use the sigmoid function you programmed earlier to squash the outputs/inputs.

#input #output #NeuralNetwork #MachineLearning #MathematicalModels #LinearFunctions

Solutions to @Linuxer4Fun
Documentation: https://en.wikipedia.org/wiki/Perceptron


https://t.me/prograchallenges

Project Neural Network #3
Difficulty: medium
Time: 2-3h
Languages: The one you picked
=======================
Use the perceptrons of the last Challenge and arrange them into layers.

The first layer is the input layer, the next layers are hidden layers and the last one is the output layer.

Every Perceptron gets as input the SUM of all the outputs of the last layer, each multiplied by some value (called "weights"). In later challenges, we will learn how to improve those weights in order to make out Neural Network be a more and more accurate mathematical model.

#input #output #NeuralNetwork #MachineLearning #Arrays #Structure

Solutions to @Linuxer4Fun
Documentation: https://en.wikipedia.org/wiki/Artificial_neural_network

Project Neural Network #4
Difficulty: hard
Time: 7h
Languages: the one you picked
========================
With this basic Neural Network structure, improve the weights of each single perceptron. For this, use backpropagation. Intuitively it means: "how much does the error value change, when I change this weight?".

#input #output #NeuralNetwork #MathematicalModels #Maths #Arrays #algorythm #backpropagation

Solutions to @Linuxer4Fun
Documentation: https://de.wikipedia.org/wiki/Backpropagation


++++++++++++++++++++++++++


This marks the end of the first project. Did you like it? Do you want more? Do you have suggestions? Discuss those in the challenge-solvers group: https://t.me/joinchat/EXSmZwy0PqOMNWPKxI3shw

Do you want to share the programming fun with your friends? Then, give them this link: https://t.me/prograchallenges