AI glossary · Fundamentals
Deep learning
In one sentence
Deep learning is a type of machine learning that uses neural networks with many layers. Each layer picks out more abstract features of the data, which lets it handle text, speech and images without hand-written rules.
Also known as: deep neural networks · DL · deep structured learning
What is deep learning?
It’s the technique that made AI genuinely usable with text, speech and images. Before it, getting a system to recognise an invoice or a spoken word meant telling it which features to look for. With deep learning, the network works those features out for itself from lots of examples.
Put it to work
AI agents that look after your customers
From €49/mo + VAT
Voice, WhatsApp and email with the phone line included: agent, numbers and CRM in one fee.
See plans and pricingThe ideas go back decades, but they took off around 2012, when three things came together: large datasets, graphics cards that could handle the maths and better training methods. Since then it has driven almost all machine learning on complex data.
Related to this term
UTIO agents
- Available 24/7 in 26 languages
- Set it up yourself for €0: built from your website
- 14-day free trial, no card needed
From €49/mo + VAT
See plans and pricingHow it works
The building block is the neural network: layers of connected units, with a numeric weight on each connection.
- Input. Data goes in as numbers: the pixels of an image, the audio of a call or the tokens of a text.
- Hidden layers. Each layer transforms what it gets from the one before. Early layers pick up simple features (edges, sounds); later ones combine them into richer concepts (letters, words, intent).
- Output. The final layer gives the result: a label, a transcript or the next token of a reply.
During training, the outputs are compared with the right answers and the weights are nudged. Those weights are the model’s parameters. Today’s language models are deep networks with billions of them.
Example
A drinks wholesaler in Glasgow takes orders from pubs as WhatsApp voice notes and photos of handwritten order sheets. Deep learning sits behind each step: one model turns the voice note into text, another reads the photographed sheet, and a language model pulls out product, quantity and delivery date.
Nobody wrote rules for every customer’s accent or every sheet layout. The networks learned them from large numbers of speech and document examples.
What it means for your business
You don’t need to understand the layers to benefit, but you should know the consequences. A deep learning model works on learned patterns: it’s right a lot of the time, but it can’t explain its reasons and may stumble on cases very unlike the ones it saw. That’s why a good agent knows how to say “I don’t know” and pass the conversation to a person.
At UTIO, deep learning sits behind every part of the voice agent: understanding the caller, deciding on a reply and speaking in a natural voice. In the AI API, transcription, text to speech and OCR are deep models too, all available with a single key.
Frequently asked questions
Why is it called 'deep' learning?
Because of the number of layers in the neural network. A network with many layers between input and output is called deep. It describes the architecture, not a deeper kind of understanding.
What's the difference between deep learning and machine learning?
Deep learning is a subset of machine learning. Classic methods often need a person to decide which features of the data to look at; deep learning learns those features itself. The price is much more data and computing power.
What is deep learning used for?
Speech recognition, machine translation, reading scanned documents, image recognition and the language models behind AI chat assistants and agents.
Related terms
- Neural network
An artificial neural network is a mathematical model made of layers of connected units, called neurons, that turn a numeric input into an output. It learns by adjusting the weight of each connection based on examples.
- Machine learning
Machine learning is the branch of AI in which a system learns to do a task from examples, rather than following hand-written rules, and gets better as it sees more data.
- Transformer
A transformer is a neural network architecture that processes every word in a text at once and uses an attention mechanism to work out which parts relate to each other. It's the foundation of almost every language model in use today.
- LLM
An LLM (large language model) is an AI model trained on vast amounts of text to predict the next word. That single skill lets it draft, summarise, translate, classify and hold a conversation.
- Model parameters
A model's parameters are its internal numeric values, mainly the weights of a neural network, which are adjusted during training and store what it has learned. Their count is used as a rough measure of model size.
Updated on