Deep learning is the most hot topic in this decade, Chatgpt, Sora, and Midjourny , all about deep learning, the question is what is deep learning
To answer this question we need to go back to the 1950's. The British mathematician Alan Turing proposed the Turing test, to test machine intelligence compared to human intelligence. In order for a machine to be called intelligent it should be able to has the following capabilities :
- Natural language processing
- Knowledge representation
- Automated reasoning
- Machine learning
- Computer Vision
- Robotics
In 1950 Alan Turing set the first design for General Artificial Intelligence (GAI), until today researchers and engineers work hard to build a machine that has an indistinguishable intelligence, a system that act and think like a human !!. Other researchers believe that building a such complex machine will require a huge amount of computation power to match the human brain complexity, so they define AI as systems that have a rational behavior and not necessarily like human, so they start to break down the AGI to small systems.
Deep learning is one of the sub fields of machine learning, based on artificial neuron network (ANN), which is a mathematical abstraction of a biological neuron. The human brain contains billions of neurons that has trillions of connection, the biological neuron receive signals from other neuron through the dendrite , and send signal through the axon as illustrates by the following picture
 |
| Analogy of biological neuron |
In contrary, the mathematical neuron is a function which get activated (fired🔥) when a linear combination of its inputs exceed a threshold value. The mathematical model of a neuron can be described by the following equation :
The output of a neuron is equals to the summation of the input x multiplied by the weight W of the neuron plus the neuron bias b, F is the activation function. Don't worry I will explain more about the ANN in further posts. These mathematical neurons, also known as nodes, can be interconnected
to perform complex tasks. Neurons directly connected to the input are
called the input layer, while those connected to the output are called
the output layer. Between the input layer and the output layer there is the middle layer, if the ANN model has more than one middle layer we call it as deep neuron network .
 |
| Deep Artificial Neuron Network |