Welcome, as we begin to look at modern-day AI! 🤗
Deep Learning is a branch of Machine Learning that trains a computer model to learn by imitating how the human brain learns. This uses a model called Neural Networks.
Deep Learning was conceived even before 1980, but it got the right boost in the 2010s with new technologies that are perfect for it to train faster like the GPUs by NVIDIA, and more robust techniques, you can read more on the history of Deep Learning here.
How does deep learning work, compared to the human brain?
In the human brain, neurons are the fundamental units that process and transmit information through electrical and chemical signals. Each neuron receives input from other neurons through its dendrites, processes this information in the cell body, and then sends an output signal through its axon to other neurons.
Deep Learning models, mimic this biological process. In deep learning, nodes (similar to neurons) are organized into layers using neural networks. Each node receives input from nodes in the previous layer, processes the input (through a mathematical function), and then passes the output to nodes in the next layer.
We'll explain this in more detail in the next chapter, don't fret 🙂.
Deep learning models can be trained using all machine learning types, from supervised, unsupervised, semi-supervised, and re-enforcement learning.
Deep Learning models can have hundreds of layers, which earns it the name "Deep" learning. Compared to the human brain the most sophisticated Deep Learning model is still far off.
The brain contains approximately 86 billion neurons, while Deep Learning artificial neural networks may contain millions to billions of nodes(neurons), depending on the network’s architecture. The human brain has trillions of synaptic connections, whereas neural networks have connections determined by the number of layers and nodes, often numbering in the millions.
Deep Learning powers most of the current AI technology, from Large Language Models(LLMs) used in Generative AI products like ChatGPT to image and voice recognition technologies. Even before this, we have seen Deep Learning's prowess when Google's AlphaGo defeated a Go (Go is a very complex board game) professional in 2015, you can read more here.
I'll not be far off if I say that modern-day AI is based solely on Deep Learning.
The chapter would be short and sweet since Deep Learning is based on Neural Networks. In our next chapter, we'll learn in-depth about Neural Networks.
On our journey to this point, we've learned, trained, and deployed over 20 ML models, this has given us the perfect background to take on Deep Learning head-on 🦾.
Get your groove on for the next one: Neural Networks.
See ya! 👽