作者: Christof Löding , P. Madhusudan , Daniel Neider
DOI: 10.1007/978-3-662-49674-9_10
关键词: Sample space 、 Semantics (computer science) 、 Convergent synthesis 、 occam 、 Variety (cybernetics) 、 Theoretical computer science 、 Counterexample 、 Space (commercial competition) 、 Artificial intelligence 、 Mathematics 、 Iterative learning control
摘要: We develop abstract learning frameworks for synthesis that embody the principles of CEGIS counterexample-guided inductive algorithms in current literature. Our framework is based on iterative from a hypothesis space captures synthesized objects, using counterexamples an sample space, and concept abstractly defines semantics synthesis. show variety literature can be embedded this general framework. also exhibit three recipes convergent synthesis: first two finite spaces Occam learners generalize all techniques convergence used existing engines, while third, involving well-founded quasi-orderings, new, we instantiate it to concrete problems.