Название: Centrality and Diversity in Search: Roles in A.I., Machine Learning, Social Networks, and Pattern Recognition Автор: M.N. Murty, Anirban Biswas
Издательство: Springer
Год: 2019
Страниц: 100
Язык: английский
Формат: pdf (true), epub
Размер: 10.1 MB
Понятия центральности и разнообразия очень важны в алгоритмах поиска и играют центральную роль в приложениях искусственного интеллекта (ИИ), машинного обучения (МЛ), социальных сетей и распознавания образов. Эта работа исследует значение центральности и разнообразия в представлении, регрессии, ранжировании, кластеризации, оптимизации и классификации. Текст предназначен для широкого круга читателей. Требуя только базовых знаний в математике на уровне бакалавриата, работа подходит для старших студентов и аспирантов, а также исследователей, работающих в области машинного обучения, интеллектуального анализа данных, социальных сетей и распознавания образов.
The concepts of centrality and diversity are highly important in search algorithms, and play central roles in applications of artificial intelligence (AI), machine learning (ML), social networks, and pattern recognition. This work examines the significance of centrality and diversity in representation, regression, ranking, clustering, optimization, and classification.
Centrality and diversity have different roles in different tasks associated with AI and ML. For example, search may be generically viewed as playing an important role in:
• AI problem solving. Here, we represent a problem configuration as a state and we reach the goal state or final state by using appropriate search scheme.
• Representation of a problem configuration in AI , representation of a data point, class, or cluster.
• Optimization which itself involves the search for an appropriate solution.
• Selecting a model for classification, clustering, or regression.
• Search engines where the search is the most natural operation.
Representation itself is an important task in a variety of tasks. Popularly representation deals with every task in AI and ML . Optimization is controlled through some regularizer to reduce the diversity in the solution space.
Clustering is an important data abstraction task that is popular in ML , data mining, and pattern recognition. Classification and regression have some common characteristics and bias–variance trade-off unifies them. Ranking is important in a variety of tasks including information retrieval.
Centrality and diversity play different roles in different tasks. In classification and regression, they show up in the form of variance and bias. In clustering, centroids represent clusters and diversity is essential in arriving at a meaningful partition. Diversity is essential in ranking search results, recommendations, and summarization of documents.
The text is designed to be accessible to a broad readership. Requiring only a basic background in undergraduate-level mathematics, the work is suitable for senior undergraduate and graduate students, as well as researchers working in machine learning, data mining, social networks, and pattern recognition.
Скачать Centrality and Diversity in Search: Roles in A.I., Machine Learning, Social Networks, and Pattern Recognition