Agent

An AI system that perceives its environment and takes action to achieve specific goals. Agents can be simple rule-based systems or complex learning systems that adapt to their environment.

Large Language Model (LLM)

A class of AI models trained on massive amounts of text data to understand and generate human language. LLMs often use transformer-based architectures, which excel at capturing the contextual relationships among words and phrases. Because of their size and complexity, LLMs can be adapted for a variety of Natural Language Processing tasks such as text generation, summarization, translation, and question-answering. Examples include GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers).

Algorithm

A specific set of rules or instructions that define how an AI system processes information and makes decisions. Algorithms form the foundation of all AI systems and determine their behavior.

Artificial Intelligence (AI)

The field of computer science focuses on creating intelligent systems capable of performing tasks that typically require human intelligence—such as learning, reasoning, problem-solving, perception, and language understanding.

Neural Network (Artificial Neural Network)

A computing architecture inspired by biological neural networks consisting of interconnected nodes (neurons) that process and transmit information. Neural networks underpin many modern AI systems, especially in deep learning.

Bias (in AI)

Systematic errors in AI systems result from imbalanced or unrepresentative training data or design flaws, potentially leading to unfair or discriminatory outcomes. Addressing bias is crucial for responsible AI development.

Computer Vision

The field of AI focuses on enabling computers to understand and process visual information (images, videos). Applications include facial recognition, autonomous vehicles, and medical imaging analysis.

Dataset

A collection of data used to train, validate, and test AI models. High-quality, diverse datasets are essential for developing effective AI systems.

Deep Learning

A subset of machine learning using neural networks with multiple layers, enabling systems to learn complex patterns and representations directly from data.

Explainable AI (XAI)

Approaches and technologies that make AI systems' decisions transparent and interpretable to humans. Especially important in regulated industries like healthcare and finance.

Feature

An individual measurable property or characteristic is used as input in a machine-learning model. Features are the basic components that models use to make predictions or decisions.

Generative AI

Systems capable of creating new content—such as text, images, music, or code—based on patterns learned from training data. This includes large language models and image-generation technologies.

Machine Learning (ML)

A subset of AI that enables systems to learn and improve from experience without explicit programming. ML algorithms identify patterns in data to make predictions or decisions.

Model

A mathematical representation of a real-world process created through training on data. Models serve as the decision-making engine in AI systems.

Natural Language Processing (NLP)

Technology that enables computers to understand, interpret, and generate human language. NLP powers applications such as machine translation, chatbots, and text analysis.

Optimization

The process of adjusting an AI model's parameters to improve its performance on specific tasks—often by minimizing a loss function or maximizing a reward.

Reinforcement Learning (RL)

A type of machine learning where an agent learns optimal behaviors through trial-and-error interactions with an environment, receiving rewards or penalties for its actions.

Supervised Learning

A machine learning approach where models are trained on labeled data, learning to map inputs to known outputs. Commonly used for tasks like classification and regression.

Training (Model Training)

The process of teaching an AI model to perform specific tasks by exposing it to examples and adjusting its parameters (weights) based on performance.

Transfer Learning

A technique where knowledge learned in one task is applied to a different but related task, improving efficiency and reducing the need for large training datasets.

Unsupervised Learning

A machine learning approach where models learn patterns and structures from unlabeled data without predefined outputs. Useful for clustering, dimensionality reduction, and anomaly detection.

Overfitting

When a model learns patterns that are too specific to the training data, resulting in poor performance on new, unseen data.

Underfitting

When a model is too simple or hasn't been trained long enough and fails to capture the underlying trends in the data, leading to poor performance on both training and test sets.

Hyperparameters

Configuration variables set before training (e.g., learning rate, number of layers, batch size). These are not learned from the training data and must be tuned for optimal performance.

Activation Function

A function applied to the output of a neuron in a neural network to introduce non-linearity (e.g., ReLU, Sigmoid, Tanh).

Loss Function (Cost Function)

A measure of how far a model's predictions are from the correct values. Training typically aims to minimize this function to improve model accuracy.

Gradient Descent

An optimization algorithm is used to minimize the loss function by iteratively adjusting model weights in the opposite direction of the gradient of the loss.

Backpropagation

An algorithm that calculates the gradient of the loss function with respect to each weight in a neural network by propagating errors backward, enabling weight updates.

Epoch

One complete pass through the entire training dataset. Multiple epochs are used to refine model parameters iteratively.

Batch (Mini-Batch)

A subset of the training data is used in one iteration of model updates. Training can use full batches (batch gradient descent) or smaller mini-batches (mini-batch gradient descent).

Convolutional Neural Network (CNN)

A class of deep neural networks primarily used for analyzing visual imagery, using convolution operations to extract spatial features.

Recurrent Neural Network (RNN)

A class of neural networks designed for sequence data (e.g., time series, text). RNNs maintain an internal hidden state that captures information about previous inputs.

Long Short-Term Memory (LSTM)

A specialized type of RNN that addresses vanishing/exploding gradients with gating mechanisms, allowing models to capture long-range dependencies in sequential data.

Generative Adversarial Network (GAN)

A framework consisting of two models—a generator and a discriminator—that compete against each other. GANs are widely used for tasks like realistic image generation.

Big Data

Extremely large, complex datasets that require advanced tools for storage, processing, and analysis. Big data and AI often intersect as vast datasets can significantly enhance AI capabilities.

Edge Computing

Performing data processing closer to the data source (on local devices or servers) rather than relying solely on centralized cloud servers. In AI, this reduces latency and bandwidth usage and enables real-time inference.

Turing Test

Proposed by Alan Turing, it tests a machine's ability to exhibit intelligence indistinguishable from a human. If an evaluator cannot reliably tell a machine from a human, the machine is said to have passed the Turing Test.