Artificial Intelligence Abbreviations
103 abbreviations · part of Technology & Computer Abbreviations
Abbreviations from artificial intelligence and machine learning: core fields (AI, ML, NLP, computer vision), model architectures (CNN, RNN, GAN and transformer models such as GPT and BERT), training methods (RLHF, LoRA, SFT), evaluation metrics and the tools used to build and deploy models. Several abbreviations, such as CV, LDA and ANN, have two different meanings within machine learning itself, and each entry explains how to tell them apart.
| Abbreviation | Stands for | Description |
|---|---|---|
| Adam | Adaptive moment estimation | The most widely used optimizer for training neural networks, which adapts the learning rate for each parameter using running averages of recent gradients and their… |
| AGI | Artificial general intelligence | A hypothetical AI system able to understand, learn and perform any intellectual task a human can, across domains, rather than excelling only at narrow tasks. |
| AI | Artificial intelligence | The field of computer science that builds systems able to perform tasks normally requiring human intelligence, such as understanding language, recognizing images… |
| AIGC | AI-generated content | Text, images, audio, video or code produced by generative AI rather than written or made directly by a person. |
| ANI | Artificial narrow intelligence | AI designed to perform a single task or a limited range of tasks, such as playing chess, recognizing faces or recommending videos, without general understanding. |
| ANN | Approximate nearest neighbor (search) | Algorithms that quickly find the items most similar to a query vector among millions or billions of vectors by accepting slightly imperfect results. |
| ASI | Artificial superintelligence | A hypothetical AI that would greatly exceed the best human minds in virtually every field, including scientific creativity, strategy and social skills. |
| ASL | AI Safety Level | A tiered set of safety and security standards in Anthropic's Responsible Scaling Policy, under which more capable AI models must meet stricter safeguards before they… |
| ASR | Automatic speech recognition | The technology and research field of converting spoken language into text by computer, the engine behind voice assistants, dictation software, call-center analytics and… |
| AUC | Area under the curve | Total drug exposure over time, measured as the area under a graph of blood concentration against time. |
| AutoML | Automated machine learning | Tools and techniques that automate steps of building machine learning models, such as feature engineering, algorithm selection and hyperparameter tuning, so non-experts… |
| BERT | Bidirectional Encoder Representations from Transformers | A transformer-based language model from Google that reads text in both directions at once to understand context, trained by predicting masked words. |
| BLEU | Bilingual evaluation understudy | An automatic metric for machine translation that scores how many words and short phrases in a system's output match one or more human reference translations, reported… |
| BoW | Bag of words | A simple text representation that counts how often each word appears in a document while ignoring grammar and word order. |
| BPE | Byte-pair encoding | A method that builds a vocabulary of subword units by repeatedly merging the most frequent pairs of characters or bytes, so a model can represent any word with a… |
| CLIP | Contrastive Language-Image Pre-training | A model from OpenAI that learns a shared space for images and text by training on hundreds of millions of image-caption pairs, so it can match pictures to descriptions… |
| CNN | Convolutional neural network | A neural network that slides small learned filters across an image to detect patterns such as edges, textures and shapes, building up to whole objects. |
| CoT | Chain of thought | A prompting technique and model behavior in which a language model writes out intermediate reasoning steps before its final answer, improving accuracy on math, logic… |
| CUDA | Compute Unified Device Architecture | Nvidia's parallel-computing platform and programming model that lets developers run general-purpose code on Nvidia GPUs. |
| CV | Computer vision | The field of AI that enables computers to interpret images and video, including recognizing objects and faces, reading text, tracking motion and reconstructing 3D scenes. |
| DL | Deep learning | A subset of machine learning that uses neural networks with many layers to learn features directly from raw data such as pixels, audio and text. |
| DNN | Deep neural network | Any neural network with multiple hidden layers between input and output, able to learn hierarchical features, such as edges, then shapes, then whole objects in images. |
| DPO | Direct preference optimization | A method for aligning language models with human preferences that trains directly on pairs of preferred and rejected answers, without a separate reward model or… |
| DQN | Deep Q-network | A reinforcement learning method that uses a deep neural network to estimate the value of each action in each state, learning to play video games directly from raw… |
| F1 | F1 score | A classification metric equal to the harmonic mean of precision and recall, ranging from 0 to 1. |
| FM | Foundation model | A large AI model trained on broad data at scale that can be adapted to many downstream tasks, such as GPT-4, Claude or Stable Diffusion. |
| GA | Genetic algorithm | An optimization method inspired by natural selection that evolves a population of candidate solutions through selection, crossover and mutation, keeping the fittest… |
| GAN | Generative adversarial network | A generative model made of two neural networks trained against each other: a generator creates fake samples, such as images, and a discriminator tries to tell them from… |
| GBM | Gradient boosting machine | An ensemble method that builds decision trees one after another, each correcting the errors of the previous ones, producing highly accurate models for tabular data. |
| GenAI | Generative artificial intelligence | AI systems that create new content, such as text, images, audio, video and code, in response to prompts, rather than only classifying or predicting. |
| GNN | Graph neural network | Operates on graph-structured data, such as social networks, molecules, road maps or fraud rings, by passing messages between connected nodes to learn representations of… |
| GPT | Generative pre-trained transformer | A family of large language models developed by OpenAI, and more broadly the design they share: a transformer network pre-trained on huge text corpora to generate text… |
| GRU | Gated recurrent unit | A simplified recurrent neural network cell with two gates, update and reset, that captures long-range dependencies like an LSTM with fewer parameters and faster training. |
| HITL | Human in the loop | A design in which people review, correct or approve an AI system's outputs or training data rather than letting it act fully autonomously. |
| HMM | Hidden Markov model | A statistical model of a system that moves between hidden states, each producing observable outputs with certain probabilities. |
| ICL | In-context learning | The ability of large language models to perform a new task from instructions or a few examples placed in the prompt, without any change to the model's weights. |
| IoU | Intersection over union | A metric for object detection and segmentation that divides the overlap between a predicted region and the true region by the area of their union, from 0 (no overlap)… |
| KG | Knowledge graph | A database that represents facts as a network of entities, such as people, places and products, connected by labeled relationships. |
| KL | Kullback-Leibler (divergence) | A measure of how one probability distribution differs from a reference distribution, used throughout machine learning in loss functions, in variational autoencoders and… |
| KNN | K-nearest neighbors | A simple machine learning algorithm that classifies a new data point by looking at the k most similar examples in the training data and taking a majority vote, or their… |
| LDA | Linear discriminant analysis | A classification and dimensionality-reduction method that finds linear combinations of features that best separate two or more classes. |
| Llama | Large Language Model Meta AI | A family of large language models released by Meta, notable for making model weights available for download so developers can run and fine-tune them on their own… |
| LLM | Large language model | Built on the transformer architecture and trained on vast amounts of text to predict the next token, an LLM can generate and summarize text, answer questions, translate… |
| LMM | Large multimodal model | A large model that handles more than one type of input or output, such as text, images, audio and video, within a single system, extending the large language model idea… |
| LoRA | Low-rank adaptation | A parameter-efficient fine-tuning method that freezes a model's original weights and trains small low-rank matrices added to some layers, cutting the number of… |
| LSTM | Long short-term memory | A type of recurrent neural network cell with gates that decide what information to keep, update or forget, allowing it to learn dependencies across long sequences. |
| MAE | Mean absolute error | The average of the absolute differences between predictions and actual values, expressed in the outcome's units. |
| mAP | Mean average precision | The standard accuracy metric for object detection and information retrieval, averaging precision across recall levels and across all classes or queries. |
| MCP | Model Context Protocol | An open standard for connecting AI assistants to external tools and data sources, such as file systems, databases, calendars and business apps, through a common… |
| MDP | Markov decision process | A mathematical framework for sequential decision-making, defined by states, actions, transition probabilities and rewards, in which the next state depends only on the… |
| ML | Machine learning | The branch of AI in which computers learn patterns from data instead of following explicitly programmed rules, improving at a task as they see more examples. |
| MLE | Maximum likelihood estimation | A method of estimating model parameters by choosing the values that make the observed data most probable. |
| MLOps | Machine learning operations | The practices and tools for deploying, monitoring, versioning and retraining machine learning models reliably in production, applying DevOps ideas to the machine… |
| MLP | Multilayer perceptron | The basic fully connected feed-forward neural network, made of an input layer, one or more hidden layers and an output layer, with each neuron linked to every neuron in… |
| MMLU | Massive Multitask Language Understanding | A benchmark of about 16,000 multiple-choice questions across 57 subjects, from elementary math to law and medicine, used to compare the general knowledge and reasoning… |
| MoE | Mixture of experts | A neural network design in which a router sends each input token to a small subset of specialized subnetworks, called experts, so a model can have a huge total… |
| MSE | Mean squared error | The average of squared differences between predictions and actual values, a standard loss function in regression and machine learning. |
| MT | Machine translation | Automatic translation of text or speech from one language to another by computer. |
| NER | Named-entity recognition | An NLP task that finds and labels names of real-world things in text, such as people, organizations, places, dates and amounts. |
| NeRF | Neural radiance field | A technique that trains a neural network to represent a 3D scene from a set of 2D photos, so new views can be rendered from any angle with realistic lighting and… |
| NLG | Natural language generation | The part of natural language processing that produces human-readable text from data or internal representations, such as weather reports, financial summaries, product… |
| NLP | Natural language processing | The field of AI concerned with enabling computers to process, understand and generate human language, covering translation, sentiment analysis, summarization, speech… |
| NLTK | Natural Language Toolkit | A Python library for teaching and research in natural language processing, offering tokenizers, stemmers, part-of-speech taggers, parsers and dozens of text corpora. |
| NLU | Natural language understanding | The part of natural language processing focused on extracting meaning from text or speech, such as a user's intent, the entities mentioned and sentiment. |
| NMT | Neural machine translation | Machine translation that uses neural networks trained on large parallel corpora to translate whole sentences at once, producing far more fluent output than earlier… |
| NN | Neural network | A computing model made of layers of interconnected nodes, or artificial neurons, whose connection weights are adjusted during training so the network learns to map… |
| OCR | Optical character recognition | Technology that turns images of text, such as scanned documents, photos of receipts or screenshots, into machine-readable text. |
| ONNX | Open Neural Network Exchange | An open format for representing machine learning models so they can move between frameworks, such as from PyTorch to an optimized runtime on phones, servers or browsers. |
| OOD | Out-of-distribution | Describes data that differs from what a model saw in training, such as photos taken under new lighting or text from a new domain. |
| PCA | Principal component analysis | A technique that transforms correlated variables into a smaller set of uncorrelated components ordered by how much variance they explain. |
| PEFT | Parameter-efficient fine-tuning | A family of techniques that adapt large pretrained models to new tasks by training only a small fraction of parameters, such as adapters, prefixes or low-rank matrices… |
| POS | Part of speech | Refers to the grammatical category of a word, such as noun, verb, adjective or adverb. |
| PPO | Proximal policy optimization | A reinforcement learning algorithm that improves an agent's policy in small, clipped steps so training stays stable. |
| PR | Precision-recall (curve) | A plot of a classifier's precision, the share of predicted positives that are correct, against its recall, the share of actual positives found, across decision… |
| QA | Question answering | An NLP task in which a system answers questions posed in natural language, either by extracting a span from a given passage or by generating an answer from its… |
| RAG | Retrieval-augmented generation | A technique in which an AI system first searches a knowledge source, such as company documents or a vector database, and then gives the most relevant passages to a… |
| ReLU | Rectified linear unit | An activation function that outputs the input if it is positive and zero otherwise, written f(x) = max(0, x). |
| RF | Random forest | An ensemble machine learning method that trains many decision trees on random samples of the data and features, then combines their votes or averages. |
| RL | Reinforcement learning | A branch of machine learning in which an agent learns by trial and error, taking actions in an environment and receiving rewards or penalties, gradually discovering a… |
| RLAIF | Reinforcement learning from AI feedback | A variant of RLHF in which an AI model, guided by written principles or instructions, generates the preference judgments used to train a reward model, reducing the need… |
| RLHF | Reinforcement learning from human feedback | A training method in which people compare or rate a model's outputs, a reward model learns those preferences, and the language model is then optimized with… |
| RMSE | Root mean square error | The square root of the mean squared error, giving typical prediction error in the same units as the outcome. |
| RNN | Recurrent neural network | A neural network that processes sequences one step at a time while carrying a hidden state forward, giving it a memory of earlier inputs. |
| ROC | Receiver operating characteristic | A curve plotting the true positive rate against the false positive rate of a classifier or diagnostic test across every possible threshold. |
| RPA | Robotic process automation | Software "bots" that mimic human actions in user interfaces, such as clicking, copying data between applications and filling in forms, to automate repetitive rule-based… |
| SD | Stable Diffusion | An open-weight text-to-image model that generates and edits pictures from written prompts using a latent diffusion process. |
| SFT | Supervised fine-tuning | Continued training of a pretrained model on curated examples of inputs paired with desired outputs, such as instructions and ideal answers, to teach it a task, format… |
| SGD | Stochastic gradient descent | An optimization method that trains models by repeatedly nudging parameters in the direction that reduces error, estimating the gradient from a small random batch of… |
| SHAP | SHapley Additive exPlanations | A method that explains an individual prediction by assigning each input feature a contribution value, based on Shapley values from cooperative game theory. |
| SLAM | Simultaneous localization and mapping | The problem of building a map of an unknown environment while tracking the device's own position within it, solved with cameras, lidar and motion sensors. |
| SLM | Small language model | A language model with relatively few parameters, typically from a few hundred million to several billion, designed to run cheaply or on local devices such as phones and… |
| SOTA | State of the art | In AI research, the best published result on a benchmark or task at a given time. |
| STT | Speech-to-text | Technology that transcribes spoken audio into written text for dictation, captions, meeting transcripts and voice commands. |
| SVM | Support vector machine | A supervised learning algorithm that classifies data by finding the boundary, or hyperplane, that separates classes with the widest possible margin. |
| t-SNE | t-distributed stochastic neighbor embedding | A nonlinear technique for visualizing high-dimensional data in two or three dimensions, placing similar points close together so clusters become visible. |
| TF | TensorFlow | Google's open-source machine learning framework for building and training neural networks, with tools for deploying models on servers, phones and browsers through… |
| TF-IDF | Term frequency-inverse document frequency | A statistic that scores how important a word is to a document within a collection: words frequent in one document but rare across the collection score highest. |
| TTS | Text-to-speech | Technology that converts written text into spoken audio, used in screen readers, voice assistants, navigation apps, audiobooks and accessibility tools. |
| VAE | Variational autoencoder | A generative neural network that compresses data into a probabilistic latent space and learns to decode samples from that space back into realistic data. |
| ViT | Vision Transformer | An image recognition model that splits a picture into small patches and processes them as a sequence with a transformer, the architecture originally built for language… |
| VLM | Vision-language model | A model that processes both images and text, so it can describe photos, answer questions about charts or screenshots, read documents and follow visual instructions. |
| XAI | Explainable artificial intelligence | Methods and research aimed at making AI models' decisions understandable to people, for example by showing which input features drove a prediction or which image… |
| YOLO | You Only Look Once | A family of real-time object detection models that locate and classify every object in an image in a single pass of a neural network, fast enough for video. |
More technology & computer abbreviations
- Cloud Computing Abbreviations (63)
- Computer Abbreviations (92)
- Computer Hardware Abbreviations (151)
- Consumer Electronics Abbreviations (104)
- Cryptocurrency & Blockchain Abbreviations (74)
- Cybersecurity Abbreviations (119)
- Data & Database Abbreviations (53)
- File Format Abbreviations (125)
- Internet & Web Abbreviations (76)
- Networking Abbreviations (113)
- Programming Abbreviations (154)
- Software Abbreviations (70)
- Telecom Abbreviations (114)