Neural Networks

Published May 31, 2023

Artificial neural networks (ANN) are a method in artificial intelligence that teaches computers to process data in a way similar to the human brain.

What Is An Artificial Neural Network (ANN)

An ANN a type of machine learning process, called deep learning, that uses interconnected nodes or neurons in a layered structure that resembles the human brain.
An ANN is one of the main tools used for machine learning. It is a brain-inspired system used to replicate the way that humans learn.

The Process

Neural networks are trained on specific data which is broken down into the input layer (where information is fed into), then to the hidden layer (where weight and bias are calculated from the input layer), and lastly the output layer (where information is output after algorithms are processed from the hidden and input layers). This is done via perceptrons (similar to human neurons but for machines to process, store, calculate, and move data forward).

Relation To AI

AI

AI works on a collection of technologies that allow the computer to sense, learn, reason & act.
AI works because of machine learning which is equipped with algorithms and vast amounts of datasets trained to recognize patterns and solve complex problems.

ANN

Neural networks are the core of deep learning algorithms. A neural network is trained on specific datasets from which it extracts the information and passes it through different layers searching for patterns.
Once a neural network is trained on a dataset, it can start to encounter new scenarios and start making predictions based on that previous dataset.

Neural Networks

Activation Function: An activation function is the function used by a node in a neural network to take the summed weighted input to the node and transform it into the output value.
Backpropagation: Backpropagation is a crucial algorithm in the field of machine learning, specifically in the training of artificial neural networks (ANNs).
Binary Step Activation Function: The simplest threshold based activation function which works by either firing the node in case the threshold value is surpassed or doing nothing at all.
Convolutional Neural Networks: Convolutional Neural Networks are a type of neural network that are primarily used for computer vision tasks, such as image classification, object detection, and semantic segmentation.
Gaussian Activation Function: The Gaussian activation function takes the input and transforms it into a Gaussian or Normal Distribution curve, with the output values varying depending on the specific implementation and parameters.
Linear Activation Function: The linear activation function is an activation function where the activation is proportional to the input.
Long Short-Term Memory Networks: A type of recurrent neural network (RNN) architecture designed to capture long-term dependencies in sequential and time-series data.
Recurrent Neural Networks: Recurrent Neural Networks are a type of neural network distinguished by storing and re-using the output from previous steps as an additional input in the current step
RMSprop: RMSprop is an optimization algorithm that dynamically adjusts the learning rate during the training of neural networks.
Sigmoid Activation Function: A sigmoid activation function is a specific type of sigmoid function commonly used in machine learning and various fields of artificial intelligence.