U

Definition: A convolutional network architecture for fast and precise segmentation of images. Originally designed for medical image segmentation, the U-Net architecture is built upon a fully convolutional network and modified to work with fewer training images and yield more precise segmentations.

Definition: Occurs when a statistical model or machine learning algorithm cannot capture the underlying trend of the data. Underfitting would occur, for instance, when fitting a linear model to non-linear data. Such a model would have poor predictive performance.

Definition: In probability theory and statistics, a uniform distribution means that every event has equal chances of occurring. In the context of AI, random variables with uniform distribution are often used in simulations and algorithms requiring a random state with equal probability.

Definition: A type of search algorithm that operates without any domain-specific information. Such algorithms are also known as blind search and include methods like breadth-first search and depth-first search, which systematically explore the space without using any strategy about whether one state of the problem is better than another.

Definition: In software development and AI, unit testing involves testing individual units/components of a software or a program. The purpose is to validate that each unit of the software performs as designed.

Definition: Describing or designed to analyse only one variable. In statistics and machine learning, univariate data consist of only one variable per observation, which is analysed without reference to any other variable.

Definition: A type of machine learning algorithm used to draw inferences from datasets consisting of input data without labeled responses. Common unsupervised learning methods include clustering and association algorithms.

Definition: In data processing and machine learning, up-sampling is the process of increasing the number of samples in a dataset, used to balance class distribution or increase the resolution of images.

Definition: In machine learning algorithms, an update rule modifies the parameters (weights) of a model in response to the error of the output. In neural networks, this rule is often based on the backpropagation algorithm.

Definition: A function in economics and game theory that quantifies the happiness or satisfaction obtained by consuming a good or outcome. In AI, utility functions are used to model preferences over sets of choices or to evaluate the performance of agents.

Definition: In decision making, utility theory refers to the study of rational decision making under uncertainty. AI systems often use utility theory principles to make decisions that maximise the expected utility.

Definition: A 128-bit number used to uniquely identify information in computer systems. In AI applications, UUIDs may be used to uniquely identify data sets, models, or other entities.

Definition: An architecture and framework for building systems that analyse large volumes of unstructured information to discover knowledge and drive decision-making.

Definition: The process of creating products that provide meaningful and relevant experiences to users. This involves the design of the entire process of acquiring and integrating the product, including aspects of branding, design, usability, and function.

Definition: The means by which a user and a computer system interact, particularly the use of input devices and software. In AI, the design of user interfaces can incorporate intelligent automation, natural language processing, and adaptive learning systems to enhance user interactions.