Glossary

Deep learning

Machine learning using neural networks with many layers, able to learn complex patterns directly from raw data such as images or text.

Deep learning is machine learning that uses neural network models with many stacked layers, "deep" referring to that depth rather than to any measure of sophistication. Each layer learns to represent the data at a different level of abstraction, for example edges, then shapes, then objects in an image, so the network builds its own useful representations rather than relying on manually designed ones.

This differs from classical machine learning, where practitioners hand-design input variables through feature engineering before a simpler model is trained on them; deep networks instead learn features directly from raw pixels, audio waveforms or text tokens, given enough training data and compute. That shift is what made deep learning the dominant approach for computer vision, speech and natural language tasks, including the large language models behind modern generative AI.

Deep learning typically requires far more data, compute and training time than shallower models, and its decisions are harder to explain, which matters in regulated settings. It also tends to need careful architecture choices and regularization to avoid overfitting, and results can be sensitive to how the training data was collected, including biases baked into that data that the model will otherwise reproduce.

Last reviewed September 22, 2026

In the index now

Related terms

Neural network A model made of layers of connected nodes that learns to map inputs to outputs by adjusting connection weights during training. Training data The dataset with known outcomes that a machine learning model learns patterns from before being used to predict new cases. Computer vision A field of AI that trains models to interpret and extract information from images and video, such as detecting objects or faces. Large language model (LLM) A neural network trained on vast amounts of text to predict and generate language, the technology behind most modern AI assistants. Feature engineering Creating, transforming or selecting the input variables a model is trained on, to make the underlying pattern easier for it to learn.

Related tools

GluonTS Open-source Python toolkit from AWS for building and evaluating probabilistic deep-learning forecasting models. Kaggle Learn Free micro-courses from Kaggle teaching Python, SQL, pandas and machine learning through short interactive notebook exercises. MATLAB Matrix-based numerical computing environment and language for algorithm development, simulation, and applied statistical/engineering analysis. PyTorch Forecasting Open-source deep-learning forecasting library built on PyTorch and PyTorch Lightning for panel/multi-series data.

Related guides

Image and video analytics, explained How pictures and video become structured data — computer vision, the labeled training data it depends on, and how streaming teams measure watch quality.