理学 学习理论 释 PAC learning PAC学习 PAC (Probably Approximately Correct)学习，中文翻译为概率近似正确学习，是计算学习理论中最重要的研究内容之一，由图灵奖得主Valiant [1]于1984年首先提出的学习框架。PAC学习给出了一种数学形式化刻画机器学习的框架，基于此框架，很多学习问题得以从理论上进一步探讨，如什么样的学习任务是可以学习的。
人工智能 机器学习 释文 probably approximately correct learning 概率近似正确学习
- multi-storey housing
- multistoried plant community
- multi-stories dwelling/apartment/house/building
- multi-stream service
- multisulfur system
- multisystem
- multi-tape Turing machine
- multitape Turing machine
- multi-target drug
- multi target model
- multi-target model
- multi-target or multi-hit model
- multitasking
- multi-task learning
- multitask management
- multitask principal-agent theory
- Multi-Telescope Radio-Linked Interferometer
- multi-temperature Monte Carlo simulated annealing
- multi-temporal analysis
- multi-tenancy
- multi-terminal channel
- multi-terminal information theory
- multi-terminal network
- multitest
- multitester