Common ML Terms

Machine learning introduces many new terms, and it's normal not to remember every definition the first time you encounter it. This glossary provides quick explanations of common machine learning and artificial intelligence terminology used throughout aicompute.dev. Use it as a reference whenever you come across an unfamiliar concept.

Accuracy

Accuracy measures how often a machine learning model makes the correct prediction. It is one of the most common evaluation metrics for classification problems.

Algorithm

An algorithm is the mathematical method or procedure a machine learning model uses to learn patterns from data. Different algorithms are suited to different types of problems.

Batch

A batch is a small group of training examples processed together during one step of model training. Using batches makes training faster and more efficient.

Classification

Classification is a type of machine learning task where the goal is to predict categories or labels, such as whether an email is spam or not spam.

Data Drift

Data drift occurs when real-world data changes over time and becomes different from the data used to train the model. This can reduce prediction accuracy and may require retraining.

Dataset

A dataset is a collection of data used to train, validate, or test a machine learning model. High-quality datasets are essential for building reliable AI systems.

Deep Learning

Deep learning is a branch of machine learning that uses neural networks with many layers to learn complex patterns. It is widely used for computer vision, speech recognition, natural language processing, and generative AI.

Deployment

Deployment is the process of making a trained machine learning model available for real-world use through applications, websites, APIs, or cloud services.

Embeddings

Embeddings are numerical vector representations that capture the meaning or relationships of data. They are commonly used in search, recommendation systems, and large language models.

Epoch

An epoch is one complete pass through the entire training dataset. Most models require multiple epochs before training is complete.

Feature

A feature is an individual piece of information used by a machine learning model to make predictions. Examples include age, temperature, word frequency, or pixel values.

Feature Engineering

Feature engineering is the process of creating, selecting, or transforming features so they are more useful for machine learning models.

Fine-Tuning

Fine-tuning is the process of taking a pretrained model and training it further on a smaller, specialized dataset for a specific task.

Hyperparameters

Hyperparameters are settings chosen before training begins that influence how a model learns. Examples include learning rate, batch size, tree depth, and the number of training epochs.

Inference

Inference is the process of using a trained model to make predictions on new data. Training teaches the model, while inference is when the model performs its task.

Large Language Model (LLM)

A Large Language Model (LLM) is a neural network trained on massive amounts of text data. LLMs can generate text, answer questions, summarize information, write code, and perform many other language-related tasks.

Loss Function

A loss function measures how far a model's predictions are from the correct answers during training. The model tries to reduce this value as learning progresses.

Machine Learning Stack

A machine learning stack is the complete collection of tools, frameworks, workflows, and infrastructure used to build, train, deploy, monitor, and maintain machine learning systems.

Model

A model is the machine learning system that learns patterns from data and makes predictions or decisions based on what it has learned.

Monitoring

Monitoring is the process of tracking deployed machine learning models to ensure they continue performing accurately and reliably over time.

Neural Network

A neural network is a machine learning model made up of interconnected layers that learn complex patterns from data. Neural networks are the foundation of deep learning.

Overfitting

Overfitting happens when a model memorizes its training data instead of learning general patterns. It performs well during training but poorly on new unseen data.

Regression

Regression is a type of machine learning task where the goal is to predict continuous numerical values, such as house prices or temperatures.

Supervised Learning

Supervised learning trains models using labeled examples where the correct answers are already known. It is commonly used for prediction and classification tasks.

Training

Training is the process of teaching a model by providing examples, measuring prediction errors, and adjusting internal parameters to improve performance over time.

Underfitting

Underfitting occurs when a model is too simple to learn meaningful patterns from the data, resulting in poor performance on both training and test data.

Unsupervised Learning

Unsupervised learning trains models using unlabeled data. Instead of predicting known answers, the model discovers hidden patterns, groups, or relationships within the data.

Validation Set

A validation set is a portion of the dataset used during development to evaluate and tune a model before testing it on completely unseen data.