Glossary

Search results

Machine Learning(also: ML)

A branch of artificial intelligence in which computer systems learn patterns from data to make predictions or decisions without being explicitly programmed for each scenario. In accessibility contexts, machine learning is used for a wide range of applications: predicting…

Machine Teaching(also: Interactive Machine Teaching)

A paradigm in human-computer interaction where non-expert users guide the training of machine learning models through interactive feedback, such as providing examples, labels, or corrections. Unlike traditional machine learning where data scientists prepare datasets and tune…

Markov Decision Process(also: MDP)

A mathematical framework for modelling decision-making in situations where outcomes are partly random and partly under the control of a decision-maker. In accessibility and assistive technology, Markov decision processes and their extension, partially observable Markov decision…

Markov Logic Networks(also: MLN, MLNs)

A machine learning framework that combines first-order logic with probabilistic graphical models to handle uncertainty in rule-based reasoning. In assistive technology contexts, MLNs enable context-aware systems to make intelligent decisions by weighing multiple factors—such as…

MediaPipe

An open-source framework by Google for building multimodal machine learning pipelines, commonly used for real-time face, hand, and body tracking. In accessibility applications, MediaPipe Holistic extracts 3D landmarks from the user's body and hands via webcam, while MediaPipe…

Meta-learning(also: Learning to Learn)

A branch of machine learning where models are trained to learn new tasks from very few examples by leveraging knowledge gained from previous tasks. In accessibility applications, meta-learning enables technologies like teachable object recognizers that can quickly adapt to…

Minimum Viable Description(also: MVD)

Minimum viable description (MVD) is an emerging framework for audio description that establishes the foundational level of visual information needed to provide equal access to video content without introducing bias or cognitive overload. Rather than attempting to describe…

Misgendering

The act of referring to someone using language that does not reflect their gender identity, such as incorrect pronouns, titles, or gendered terms. In digital accessibility and AI contexts, misgendering occurs when automated systems incorrectly classify a person's gender based on…

Misinterpretation(also: AI Misinterpretation)

An AI error where the model incorrectly identifies or describes something that is actually present in the input. In image descriptions, misinterpretation includes errors like mistaking one product for another (shampoo for cleaning product), reading numbers incorrectly ("6mg"…

Mixed-Initiative Interaction(also: Mixed-Initiative Systems, Human-Agent Collaboration)

An interaction paradigm in which both the human user and the computer system can take initiative in directing the task, rather than one party being entirely in control. In accessibility contexts, mixed-initiative interaction is particularly important for AI-powered assistive…

Mixture of Experts(also: MoE)

Mixture of experts is a neural network architecture that routes each input through a small subset of specialist subnetworks ('experts') rather than activating the whole model. A gating network decides which experts handle a given token or query, letting the overall model be much…

Model Reliability(also: AI Reliability, Model Trustworthiness)

The degree to which an AI model consistently produces accurate, truthful, and complete outputs across different inputs and contexts. In the context of visual access technology for BLV users, model reliability encompasses factual accuracy (not fabricating content), interpretive…

Multi-Model Comparison(also: Cross-Model Comparison, Ensemble Verification)

The practice of generating responses from multiple AI models for the same input and comparing their outputs to assess reliability, identify errors, and provide a more comprehensive understanding of the content. In accessibility contexts, multi-model comparison is used to help…

Multimodal AI(also: Multimodal Generative AI)

Artificial intelligence systems capable of processing and generating content across multiple modalities such as text, images, audio, and video. In accessibility contexts, multimodal AI is significant because a single tool can address diverse access needs — for example,…

Multimodal Large Language Model(also: MLLM, Vision-Language Model, VLM)

A deep learning model that can process and generate content across multiple types of input including text, images, audio, and video. In accessibility contexts, MLLMs like GPT-4o, Gemini, and Claude have become transformative tools for blind and low vision users, enabling…

15 results.