• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Elisa: A new system for AI-assisted logico-mathematical scientific discovery incorporating novel techniques in infinite model finding.
【24h】

Elisa: A new system for AI-assisted logico-mathematical scientific discovery incorporating novel techniques in infinite model finding.

机译：Elisa：一种用于AI辅助逻辑数学科学发现的新系统，该系统将新颖的技术结合到了无限的模型发现中。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Contemporary artificial intelligence (AI) started in 1956, on the strength of a passionate resurgence in the desire to understand the principles of human reasoning, not only in order to provide a secure, rigorous foundation on which scientific theories stand, but to mechanize such reasoning---to build computational agents capable of human-level reasoning, including, ideally, those that can produce their own discoveries, and verify them according to scientific principles. The first such agents included Logic Theorist, Program I, and Program II. In parallel efforts, continuing a debate much older than AI itself, many contended (and some still do) that no such mechanization is achievable: that human reasoning and discovery are capabilities that fall outside of what is computational.;That sophisticated (and even naive) computational agents have much to offer in scientific endeavors, however, is incontestable, and as progress is made in understanding the nature of human reasoning, more sophisticated agents can be built to assist in the practice of scientific discovery.;In this dissertation, we present such a system: ELISA, an intelligent assistant for scientific discovery. The system, a suite of advanced reasoning tools integrated in a highly visual workspace, provides an interactive environment for a human reasoner to explore, hypothesize, and reason over scientific and mathematical domains. In this first realization of the ELISA architecture, particular attention is paid to complete or partial mechanization of particular aspects of the process of scientific discovery, specifically concept and conjecture generation, argumentation, and verification.;As a full-blown and general implementation of the ambitious architecture described herein would require an order of magnitude more effort than would fit in one dissertation, the introduced implementation is restricted to logico-mathematical scientific domains---domains which can be represented in a formal logical system. This restriction allows for existing powerful automated reasoning systems to be leveraged and incorporated into the ELISA system.;After setting the historical context, the presentation of ELISA begins with the architecture, inspired by a high-level theory of the process of scientific discovery. Next, the first implementation of the ELISA architecture, called ELISA1 and realized as an interactive visual workspace for discovery, is described. Finally, we zero in on a particular functionality of the ELISA 1 system, where a new technique in automated model finding is introduced. This technique hurdles a certain deficiency of existing model-finding technologies: that such systems are capable only of producing finite models. Our approach springboards from such existing work and expands the range of models that can be found to include some well-behaved, infinite domains.
机译：当代的人工智能（AI）始于热情复兴的力量，其目的是理解人类推理的原理，这不仅是为了为科学理论提供可靠，严格的基础，而且是使这种推理机械化的基础。 ---建立具有人类水平推理能力的计算代理，理想情况下包括能够产生自己发现的代理，并根据科学原理对其进行验证。最早的此类代理包括逻辑理论家，计划I和计划II。在并行的努力下，继续进行比AI本身更久的辩论，许多人争辩（并且有些仍在这样做），无法实现这样的机械化：人类的推理和发现是超出计算能力之外的能力;那么复杂（甚至天真） ）计算代理人在科学工作中可以提供很多东西，但这是无可争议的，并且随着人们对人类推理本质的理解的不断进步，可以构建更复杂的代理人来协助科学发现的实践。提出这样的系统：ELISA，一种科学发现的智能助手。该系统是集成在高度可视化工作区中的一组高级推理工具，它为人类推理人员提供了一个交互式环境，可以在科学和数学领域内进行探索，假设和推理。在ELISA架构的第一个实现中，要特别注意完全或部分机械化科学发现过程的特定方面，特别是概念和猜想的产生，论证和验证。本文描述的宏伟架构将比一个论文需要更多的工作量，引入的实现仅限于逻辑数学科学领域-可以在形式逻辑系统中表示的领域。这种限制使得现有强大的自动化推理系统得以利用，并整合到ELISA系统中。在设定历史背景之后，ELISA的介绍从体系结构开始，受到科学发现过程的高级理论的启发。接下来，描述了称为ELISA1的ELISA体系结构的第一个实现，该实现被实现为用于发现的交互式可视工作区。最后，我们将重点介绍ELISA 1系统的特定功能，其中介绍了一种自动模型查找的新技术。该技术克服了现有模型查找技术的某些缺陷：此类系统仅能够生成有限模型。我们的方法是从现有工作中跳板，并扩展了模型的范围，可以发现这些模型包括行为良好的无限域。

著录项

  • 作者

    Shilliday, Andrew Edward.;

  • 作者单位

    Rensselaer Polytechnic Institute.;

  • 授予单位 Rensselaer Polytechnic Institute.;
  • 学科 Artificial Intelligence.;Computer Science.
  • 学位 Ph.D.
  • 年度 2009
  • 页码 202 p.
  • 总页数 202
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 人工智能理论;自动化技术、计算机技术;
  • 关键词

  • 入库时间 2022-08-17 11:37:54

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号