Александр Чесалов – Глоссариум по искусственному интеллекту: 2500 терминов. Том 2 (страница 13)

«D»

Darkforest is a computer go program, based on deep learning techniques using a convolutional neural network. Its updated version Darkforest2 combines the techniques of its predecessor with Monte Carlo tree search. The MCTS effectively takes tree search methods commonly seen in computer chess programs and randomizes them. With the update, the system is known as Darkforest3321.

Dartmouth workshop – the Dartmouth Summer Research Project on Artificial Intelligence was the name of a 1956 summer workshop now considered by many (though not all) to be the seminal event for artificial intelligence as a field322.

Data analysis is obtaining an understanding of data by considering samples, measurement, and visualization. Data analysis can be particularly useful when a dataset is first received, before one builds the first model. It is also crucial in understanding experiments and debugging problems with the system323.

Data analytics is the science of analyzing raw data to make conclusions about that information. Many of the techniques and processes of data analytics have been automated into mechanical processes and algorithms that work over raw data for human consumption324.

Data augmentation in data analysis are techniques used to increase the amount of data. It helps reduce overfitting when training a machine learning325.

Data Cleaning is the process of identifying, correcting, or removing inaccurate or corrupt data records326.

Data Curation – includes the processes related to the organization and management of data which is collected from various sources327.

Data entry – the process of converting verbal or written responses to electronic form328.

Data fusion — the process of integrating multiple data sources to produce more consistent, accurate, and useful information than that provided by any individual data source329.

Data Integration involves the combination of data residing in different resources and then the supply in a unified view to the users. Data integration is in high demand for both commercial and scientific domains in which they need to merge the data and research results from different repositories330.

Data is a collection of qualitative and quantitative variables. It contains the information that is represented numerically and needs to be analyzed.

Data Lake is a type of data repository that stores data in its natural format and relies on various schemata and structure to index the data331.

Data markup is the stage of processing structured and unstructured data, during which data (including text documents, photo and video images) are assigned identifiers that reflect the type of data (data classification), and (or) data is interpreted to solve a specific problem, in including using machine learning methods (National Strategy for the Development of Artificial Intelligence for the period up to 2030)332.

Data Mining is the process of data analysis and information extraction from large amounts of datasets with machine learning, statistical approaches. and many others333.

Data parallelism is a way of scaling training or inference that replicates an entire model onto multiple devices and then passes a subset of the input data to each device. Data parallelism can enable training and inference on very large batch sizes; however, data parallelism requires that the model be small enough to fit on all devices. See also model parallelism334.

Data Processing Unit (DPU) is a programmable specialized electronic circuit with hardware accelerated data processing for data-oriented computing335.

Data protection is the process of protecting data and involves the relationship between the collection and dissemination of data and technology, the public perception and expectation of privacy and the political and legal underpinnings surrounding that data. It aims to strike a balance between individual privacy rights while still allowing data to be used for business purposes336.

Data Refinement is used to convert an abstract data model in terms of sets for example into implementable data structures such as arrays337.

Data Science is a broad grouping of mathematics, statistics, probability, computing, data visualization to extract knowledge from a heterogeneous set of data (images, sound, text, genomic data, social network links, physical measurements, etc.). The methods and tools derived from artificial intelligence are part of this family338,339.

Data set is a set of data that has undergone preliminary preparation (processing) in accordance with the requirements of the legislation of the Russian Federation on information, information technology and information protection and is necessary for the development of software based on artificial intelligence (National strategy for the development of artificial intelligence for the period up to 2030)340.

Data Streaming Accelerator (DSA) is a device that performs a specific task, which in this case is the transfer of data in less time than the CPU would do. What makes DSA special is that it is designed for one of the characteristics that Compute Express Link brings with it over PCI Express 5.0, which is to provide consistent access to RAM for all peripherals connected to a PCI Express port, i.e., they use the same memory addresses.

Data variability describes how far apart data points lie from each other and from the center of a distribution. Along with measures of central tendency, measures of variability give you descriptive statistics that summarize your data341.

Data veracity is the degree of accuracy or truthfulness of a data set. In the context of big data, its not just the quality of the data that is important, but how trustworthy the source, the type, and processing of the data are342.

Data Warehouse is typically an offline copy of production databases and copies of files in a non-production environment343.

Database is a «container» storing data such as numbers, dates or words, which can be reprocessed by computer means to produce information; for example, numbers and names assembled and sorted to form a directory344.

DataFrame is a popular datatype for representing datasets in pandas. A DataFrame is analogous to a table. Each column of the DataFrame has a name (a header), and each row is identified by a number345.

Datalog is a declarative logic programming language that syntactically is a subset of Prolog. It is often used as a query language for deductive databases. In recent years, Datalog has found new application in data integration, information extraction, networking, program analysis, security, and cloud computing346.

Datamining – the discovery, interpretation, and communication of meaningful patterns in data347.

Dataset API (tf. data) is a high-level TensorFlow API for reading data and transforming it into a form that a machine learning algorithm requires. A tf. data. Dataset object represents a sequence of elements, in which each element contains one or more Tensors. A tf.data.Iterator object provides access to the elements of a Dataset. For details about the Dataset API, see Importing Data in the TensorFlow Programmer’s Guide348.

Debugging is the process of finding and resolving bugs (defects or problems that prevent correct operation) within computer programs, software, or systems. Debugging tactics can involve interactive debugging, control flow analysis, unit testing, integration testing, log file analysis, monitoring at the application or system level, memory dumps, and profiling. Many programming languages and software development tools also offer programs to aid in debugging, known as debuggers349.

Decentralized applications (dApps) are digital applications or programs that exist and run on a blockchain or peer-to-peer (P2P) network of computers instead of a single computer. DApps (also called «dapps») are outside the purview and control of a single authority. DApps – which are often built on the Ethereum platform – can be developed for a variety of purposes including gaming, finance, and social media350.

Decentralized control is a process in which a significant number of control actions related to a given object are generated by the object itself on the basis of self-government351.

Decision boundary – the separator between classes learned by a model in a binary class or multi-class classification problems352.

Decision boundary in the case of backpropagation-based artificial neural networks or perceptrons, the type of decision boundary that the network can learn is determined by the number of hidden layers the network has. If it has no hidden layers, then it can only learn linear problems. If it has one hidden layer, then it can learn any continuous function on compact subsets of Rn as shown by the Universal approximation theorem, thus it can have an arbitrary decision boundary.

Decision intelligence (DI) is a practical discipline used to improve the decision making process by clearly understanding and programmatically developing how decisions are made and how the outcomes are evaluated, managed and improved through feedback.