Deep Learning
Machine learning using many-layered neural networks to learn representations from raw data.
Overview
Deep learning trains neural networks with many layers so that the network discovers its own features instead of relying on hand-designed ones. It solves the problem that feature engineering does not scale: for images, audio and text, useful features are hard to specify but easy to learn. It exists because more layers plus more data plus more compute turned out to beat carefully engineered features on almost every perceptual task.
Technology Graph
How Deep Learning connects to the rest of the stack.
Uses and depends on
- Machine Learning concept Algorithms Deep Learning based on Machine Learning High ✓ source
Applied to
- Dropout technique Techniques Dropout used for Deep Learning High ✓ source
- GELU concept Algorithms GELU used for Deep Learning High ✓ source
- GPU hardware Hardware GPU used for Deep Learning High ✓ source
- JAX framework Frameworks JAX used for Deep Learning Medium ✓ source
- Layer Normalization concept Algorithms Layer Normalization used for Deep Learning High ✓ source
- PyTorch framework Frameworks PyTorch used for Deep Learning High ✓ source
- Residual Connection concept Algorithms Residual Connection used for Deep Learning High ✓ source
- TPU hardware Hardware TPU used for Deep Learning High ✓ source
- TensorFlow framework Frameworks TensorFlow used for Deep Learning High ✓ source
- Triton Inference Server tool Infrastructure Triton Inference Server used for Deep Learning High ✓ source