Exploitable results ILP2: enhancing machine learning At the crossover between inductive machine learning and logic programming, inductive logic programming (ILP) is now the most promising subfield of machine learning. Using a combination of logical analysis, background knowledge and observational data, computer programs look for underlying regularities or rules in large bodies of data, which can be used to form predictive models. These rules are used to advance more general hypotheses, which then contribute to understanding the problem. Bringing together ILP experts and specialists in particular domains, ILP 2 extends the fundamental research results of the ILP project into four key areas: natural language processing, data mining and discovery, design and configuration, and database design. ILP is different from other fields of machine learning, such as neural networks and classical attribute value learning (eg decision-tree generators). It performs concept-learning using an induction process, within the representations offered by computational logic, and handles background knowledge, relational and structural representations. Efficient computer programs are used to derive general rules or relations from a combination of data (observations, examples, etc), background knowledge (stored in a knowledge base), and meta-knowledge about the kind of regularities likely to be found (bias). ILP then outputs general descriptions in the form of logic programs that can be used to make predictions about unseen observations. Searching for OpenAIRE data... There was an error trying to search data from OpenAIRE No results available
