11 terms
Showing all terms starting with D
Techniques to artificially increase training dataset size by applying transformations like rotation, flipping, or paraphrasing to existing samples.
A subset of machine learning using neural networks with many layers (deep) to learn hierarchical representations from raw data.
A generative AI model that creates images by learning to reverse a gradual noising process, used in tools like Stable Diffusion and DALL-E.
A regularisation technique that randomly deactivates neurons during training to prevent overfitting and improve generalisation.
An automated sequence of steps that ingests, transforms, and loads data for use in training, fine-tuning, or serving AI models.
A transformer architecture that generates text autoregressively, used in GPT-style models for text generation and chat.
A retrieval method that uses embedding vectors to find semantically similar documents, as opposed to sparse keyword-based search.
Techniques like PCA or UMAP that compress high-dimensional data into fewer dimensions while preserving meaningful structure for visualisation or modelling.
AI systems that extract structured information from unstructured documents such as invoices, contracts, and forms using OCR and NLP.
The process of adjusting a model trained on general data to perform well on a specific domain with different data distributions.
A model compression technique that quantises weights at inference time, reducing memory usage and speeding up models without retraining.
