How to build a Simple Neural Network in Python (One Layer) Part I

Опубликовано: 06 Ноябрь 2021
на канале: Algovibes

11,579

337

Example is from the book "Machine Learning for Finance" by Jannes Klaas.
In this video we are building a simple one layer Neural Network from scratch in Python. In specific we are setting up the input layer and initialize random weights and feed this data to the activation function (sigmoid).
The output is then compared with the actual output (y) and measured with the binary cross entropy loss.
If you found this interesting I will continue with optimizing the network using Gradient Descent and Parameter update using Backpropagation and provide details on how to proceed.

Link to Book (Not affiliated btw)
https://www.amazon.de/Machine-Learnin...

Mentioned articles:
Bias and weights
https://towardsdatascience.com/whats-...

Cross Entropy Loss:
https://towardsdatascience.com/cross-...

00:00 - 01:36 Introduction / Resources
01:36 - 02:31 Input Layer and output (y)
02:31 - 04:20 Initialize random weights and bias
04:20 - 04:50 Getting z (Summation + bias)
04:50 - 05:53 Sigmoid function and getting A (Output of layer)
05:53 - 06:16 Comparing layer output with y
06:16 - 10:20 Binary Cross Entropy Loss
10:20 - 10:45 What needs to be done / outlook

#Python #NeuralNetwork