Neural Network is a Ridiculous Name.

Introduction

Artificial neural networks (ANNs) are often labeled as such because they are inspired by the human brain. However, this comparison is somewhat misleading and akin to saying a child's drawing closely resembles an actual person—a point not meant to demean the skills of third graders. The reality is that the neurons embedded in these artificial networks are drastically simplified versions of biological neurons.

Let's break down the mechanics behind these artificial neurons. At its core, each neuron receives a list of numbers as input and produces a single output value. The process is fairly straightforward: each input number is multiplied by a corresponding weight. These weights are adjusted during the training phase of the network and remain fixed during actual operation.

For instance, let’s consider a scenario where the inputs to a neuron are 1, 2, and 3, and the associated weights are -1, 0, and 1. The computation starts by multiplying each input by its corresponding weight, resulting in outputs of -1, 0, and 3, respectively. Once these values have been calculated, they are aggregated by summing them up.

Following this, we introduce an additional value known as the bias, which is added to the total. If the resulting sum is negative, it is adjusted to zero; otherwise, it remains unchanged. Thus, in this example, the final output of the neuron would be a value of 3.

The astonishing aspect of this architecture is that when around 100 billion of these artificial neurons are interconnected, we can create models like ChatGPT. Despite their simplified nature, these interconnections and computations form the basis for advanced artificial intelligence systems.

Keywords

• Artificial neural networks (ANNs)
• Neurons
• Weights
• Training phase
• Bias
• Output value
• Interconnections
• Advanced artificial intelligence

FAQ

What are artificial neural networks (ANNs)?
Artificial neural networks are computational models inspired by the human brain, designed to process information through interconnected nodes (neurons).

How do artificial neurons function?
Artificial neurons accept a list of numbers as input, multiply each by specific weights, sum the results, add a bias, and then produce a single output value.

What are weights in neural networks?
Weights are values that determine the importance of each input; they are modified during training to improve the network's performance.

What is the role of bias in a neuron?
The bias is an additional value added to the output, aiding in the adjustment of the final neuron value, which can help in shifting the activation function.

How is ChatGPT related to artificial neurons?
ChatGPT utilizes a vast number of interconnected artificial neurons—approximately 100 billion—forming a complex neural network capable of advanced language understanding and generation.