I

Definition: A computer vision technique that allows machines to identify objects, places, people, writing, and actions in images. It uses machine learning models to detect and classify multiple objects in a single image.

Definition: Occurs when the number of observations in one class is significantly lower than those in other classes. This can pose a challenge in machine learning as it may lead to models that are biased towards the majority class.

Definition: The process of using a trained machine learning model to make predictions based on new data. Inference is about applying the learned patterns to unknown data to make decisions.

Definition: The science of searching for information in documents, searching documents themselves, searching for metadata which describe documents, or searching within databases, whether relational stand-alone databases or hypertextually linked networks like the internet.

Definition: A model of machine learning where generalisations are not explicitly formed; instead, the training instances are used to predict the output for new instances based on a similarity measure.

Definition: A system that perceives its environment and takes actions that maximise its chances of successfully achieving its goals. These are often used in scenarios involving complex dynamics or information-gathering.

Definition: A network of physical objects (‘things’) embedded with sensors, software, and other technologies for the purpose of connecting and exchanging data with other devices and systems over the internet.

Definition: A process for calculating a desired result by means of a repeated cycle of operations. In AI, iterative processes are used in algorithms that require refinement and improvement over multiple steps or iterations.

Definition: A branch of applied mathematics and electrical engineering involving the quantification of information. Historically, it has been used in data compression and telecommunications, but is also fundamental in modern machine learning and AI.

Definition: Indirect measures of user preference in a system, such as browsing time or mouse movements, as opposed to explicit feedback like ratings or reviews.

Definition: The process of replacing missing data with substituted values in a dataset. Data imputation helps improve the quality of data and thereby the accuracy of machine learning models.

Definition: A type of learning where the model makes generalisations from specific examples. This is opposite to deductive learning, where the reasoning starts with a general statement and moves towards a specific instance.

Definition: A popular architecture for convolutional neural networks, originally developed for the GoogLeNet architecture, designed to help the network learn more complex patterns.

Definition: A method in machine learning where the model is continuously updated with new data without needing to retrain from scratch. It is useful in situations where data is continuously generated over time.

Definition: The process of creating an index for all the features in the data, allowing for quicker retrieval. In AI, indexing can be applied to vast datasets to enhance the speed and performance of algorithms.

Definition: The set of assumptions that a learning algorithm uses to predict outputs given inputs that it hasn’t encountered before. Inductive bias helps guide the learning process and can significantly impact the performance of the model.

Definition: A graphical representation of a decision-making framework, showing a summary of the information, variables, decisions, and objectives involved.

Definition: A data structure used to create a searchable database of the contents of a dataset or collection of information. It is widely used in document retrieval systems like search engines.

Definition: A nonlinear dimensionality reduction method based on the spectral theory that seeks to preserve the geodesic distances in the reduced space. It is often used in manifold learning.

Definition: A type of image segmentation task in computer vision where each instance of a class in the image is uniquely identified and segmented with a high-resolution boundary.

Definition: A technique for attributing the prediction of a neural network to its inputs, used in the field of explainable AI to highlight how much each input feature contributes to the final decision.

Definition: An algorithm for anomaly detection that works on the principle of isolating anomalies instead of profiling normal data points. It is efficient and useful for identifying outliers in data.

Definition: A type of logic that differs from classical Boolean logic by handling the concept of negation differently, often used in programming languages and to model human reasoning more closely.