Skip to main content
Contact

AI glossary Models & Architecture

What is deep learning?

← All glossary terms
In short

A form of machine learning that uses neural networks with many layers to learn patterns directly from raw data such as text, images or sound.

Explained

Deep learning is a branch of machine learning based on artificial neural networks with many layers. Each layer transforms its input into a slightly more abstract representation, so the network learns useful features by itself instead of relying on features designed by hand. Training adjusts millions or billions of parameters using large datasets and the backpropagation algorithm. Deep learning underpins computer vision, speech recognition, language models and most of today's state-of-the-art AI.

How it works

A neural network is made of layers of simple units connected by weights. Data goes in at one end, passes through the layers and produces an output, such as a label or the next word.

During training the output is compared with the right answer, and backpropagation adjusts the weights to reduce the error. Repeated over many examples, the early layers learn simple patterns (edges, sounds, word pieces) and later layers combine them into more complex ones.

Main architectures

  • Convolutional neural networks (CNNs): images and video.
  • Recurrent networks and LSTMs: sequences, now largely replaced by transformers.
  • Transformers: language, and increasingly images, audio and multimodal data.
  • Diffusion models: image, audio and video generation.
  • Autoencoders and embedding models: compact representations of data.

When it makes sense

  • The input is unstructured: text, images, audio, video.
  • There is a lot of data, or a pre-trained model you can adapt.
  • Accuracy matters more than a fully transparent model.

For small tabular datasets, simpler methods such as gradient-boosted trees are often as accurate, cheaper and easier to explain.

Frequently asked questions

What is the difference between deep learning and machine learning?

Deep learning is a subset of machine learning. Classic machine learning often relies on features engineered by people; deep learning uses many-layered neural networks that learn those features from raw data. Deep learning usually needs more data and compute in exchange.

Why is it called deep learning?

"Deep" refers to the number of layers in the neural network. A shallow network has one or two hidden layers; a deep network has many, which lets it build more abstract representations step by step.

Is ChatGPT deep learning?

Yes. ChatGPT runs on large language models, which are deep neural networks based on the transformer architecture and trained on large text corpora.

What is deep learning used for?

Image and video recognition, speech recognition and synthesis, translation, language models and chat assistants, recommendation systems, and scientific uses such as protein structure prediction. Most recent advances in AI rely on it.

Does deep learning need a lot of data?

Training from scratch does. In practice most projects start from a pre-trained model and adapt it with a much smaller amount of their own data, through fine-tuning or by combining it with retrieval.

Sources

Updated: 2026-10-21

Want to see how this applies to your organisation?

Talk to an AI Expert