AI Blog

AI Blog

by Michele Laurelli

AI Glossary

A comprehensive dictionary of artificial intelligence terms and concepts

160 terms

A

Accuracy

Metric

Proportion of correct predictions out of total predictions made.

Activation Function

Concept
/ˌæktɪˈveɪʃən ˈfʌŋkʃən/

A mathematical function applied to a neuron's output to introduce non-linearity into the network.

Activation Layer

Architecture
/ˌæktɪˈveɪʃən ˈleɪər/

A layer that applies a non-linear activation function element-wise to its input.

Adagrad

Algorithm

Adaptive learning rate optimizer that adapts rates per parameter based on historical gradients.

Adam Optimizer

Algorithm
/ˈædəm/

An adaptive learning rate optimization algorithm combining momentum and RMSprop.

Adversarial Training

Technique

Training technique improving model robustness by including adversarial examples.

AlexNet

Architecture

Deep CNN that won ImageNet 2012, pioneering deep learning in computer vision.

Algorithm

Concept

A step-by-step procedure or formula for solving a problem or performing a task.

Artificial Intelligence (AI)

Concept

Field of computer science focused on creating systems capable of performing tasks requiring human intelligence.

Attention Mechanism

Technique
/əˈtɛnʃən ˈmɛkənɪzəm/

A technique allowing models to focus on specific parts of the input when producing output.

Attention Score

Concept

Computed similarity between query and key vectors before softmax normalization in attention.

Attention Weight

Concept
/əˈtɛnʃən weɪt/

A scalar value indicating how much focus to place on a specific part of the input when producing output.

AUC-ROC

Metric

Area under ROC curve, measuring binary classifier quality across all thresholds.

Autoencoder

Architecture
/ˈɔːtoʊɪnˌkoʊdər/

A neural network trained to reconstruct its input, learning compressed representations in the process.

Average Pooling

Technique
/ˈævərɪdʒ ˈpuːlɪŋ/

A pooling operation that computes the average value from each window of the feature map.

C

Causal Masking

Technique

Masking technique preventing attention to future positions in autoregressive models.

Chain-of-Thought Prompting

Technique

Prompting technique encouraging LLMs to show intermediate reasoning steps before answering.

Classification

Task
/ˌklæsɪfɪˈkeɪʃən/

A supervised learning task where the goal is to predict discrete class labels for input data.

CLIP (Contrastive Language-Image Pre-training)

Model
/klɪp/

A multimodal model trained to understand relationships between images and text.

Clustering

Task
/ˈklʌstərɪŋ/

An unsupervised learning task that groups similar data points together based on their features.

CNN (Convolutional Neural Network)

Architecture
/siː ɛn ɛn/

A deep learning architecture specialized for processing grid-like data such as images, using convolutional layers.

COCO (Common Objects in Context)

Dataset

Large-scale dataset for object detection, segmentation, and captioning with 330k images.

Computer Vision

Field
/kəmˈpjuːtər ˈvɪʒən/

A field of AI enabling computers to derive meaningful information from visual inputs like images and videos.

Confusion Matrix

Concept
/kənˈfjuːʒən ˈmeɪtrɪks/

A table used to evaluate classification model performance by showing true vs predicted classes.

Contrastive Learning

Paradigm

Self-supervised learning contrasting positive pairs against negative pairs.

Convolution

Concept
/ˌkɒnvəˈluːʃən/

A mathematical operation that slides a filter/kernel over input data to extract features.

Convolutional Layer

Architecture
/ˌkɒnvəˈluːʃənəl ˈleɪər/

A layer in CNNs that applies convolution operations to extract spatial features from input data.

Cross-Validation

Technique

Resampling technique evaluating model performance by splitting data into multiple train-test folds.