人工智能框架技术趋势研究

作为人工智能开发环节中的基础工具,人工智能框架承担着AI技术生态中操作系统的角色,是AI学术创新与产业商业化的重要载体。随着其重要性的不断突显,人工智能框架已经成为人工智能产业创新的焦点之一,引起了学术界、产业界的高度重视。在此背景下,本文从人工智能框架的概念内涵、价值意义入手,梳理人工智能框架演进历程,总结当前人工智能框架技术体系,最后研判得出人工智能框架技术发展的“四泛”趋势,即泛开发、泛设备、泛场景、泛工程。     

文档智能: 数据集、模型和应用

文档智能是指通过计算机进行自动阅读、理解以及分析商业文档的过程,是自然语言处理和计算机视觉交叉领域的一个重要研究方向。近年来,深度学习技术的普及极大地推动了文档智能领域的发展,以文档版面分析、文档信息抽取、文档视觉问答以及文档图像分类等为代表的文档智能任务均有显著的性能提升。该文对于早期基于启发式规则的文档分析技术、基于统计机器学习的算法以及近年来基于深度学习和预训练的方法进行简要介绍,并展望了文档智能技术的未来发展方向。     

结合AI规划和工作流的动态服务组合框架研究

动态服务组合是构建面向服务、松耦合、集成化的应用系统的主要途径.现行的动态服务组合方法中,工业界倾向于采用BPEL建模服务流程,而语义网研究团体则吸取AI规划的思想以实现Web服务组合,以上两种方法都不同程度地存在各自的不足.在对上述方法进行对比分析的基础上,提出了一种结合AI规划和工作流技术的动态服务组合方法.实践表明,这种方法能有效克服采用单个方法所造成的不足,并可实现从需求到部署的全过程.     

From ethics to standards – A path via responsible AI to cyber-physical production systems

The central claim of the paper is that the development and control of Cyber-Physical Production Systems (CPPS) requires a systematic approach to handle and include explicit ethical considerations. Since the contribution of artificial intelligence (AI) technologies, and of agent-based models in particular, was instrumental in the evolution of CPPSs, approaches of ethical AI should be endorsed in CPPS development by design. The paper discusses recent advances for ethical AI and suggests a pathway from ethical norms towards standards. As it is argued, taking the responsible AI approach is promising when tackling the main ethic-related challenges of Cyber-Physical Production Systems. We expose a number of dilemmas to be resolved so that AI systems incorporated in CPPS cause no damages either in humans, equipment or in the environment and increase the trust in the users of current and future AI technologies.     

A comprehensive survey and future trend of simulation study on FMS scheduling

Since the late 1970s when the first collection of papers on scheduling of flexible manufacturing systems (FMSs) has been published, it has been one of the most popular topics for researchers. A number of approaches have been delivered to schedule FMSs including simulation techniques and analytical methods, whereas the former is the most widely used tool for modeling FMSs. The objective of this paper is to review scheduling study on FMSs and analyse future trend that employed simulation techniques as the analyzing tool. Scheduling methodologies are categorized into, namely traditional simulation techniques with single criterion scheduling approaches, traditional simulation techniques with multi-criteria scheduling approaches, and artificial intelligence (AI) approaches in FMSs. It is concluded that AI approaches will be dominating in future study.     

AI and machine learning techniques for managing complexity,changes and uncertainties in manufacturing

The application of pattern recognition techniques, expert systems, artificial neural networks, fuzzy systems and nowadays hybrid artificial intelligence (AI) techniques in manufacturing can be regarded as consecutive elements of a process started two decades ago. The paper outlines the most important steps of this process and introduces some new results with special emphasis on hybrid AI and multistrategy machine learning approaches. Agent-based (holonic) systems are highlighted as promising tools for managing complexity, changes and disturbances in production systems. Further integration of approaches is predicted.     

VIS+AI: integrating visualization with artificial intelligence for efficient data analysis

Visualization and artificial intelligence (AI) are well-applied approaches to data analysis. On one hand, visualization can facilitate humans in data understanding through intuitive visual representation and interactive exploration. On the other hand, AI is able to learn from data and implement bulky tasks for humans. In complex data analysis scenarios, like epidemic traceability and city planning, humans need to understand large-scale data and make decisions, which requires complementing the strengths of both visualization and AI. Existing studies have introduced AI-assisted visualization as AI4VIS and visualization-assisted AI as VIS4AI. However, how can AI and visualization complement each other and be integrated into data analysis processes are still missing. In this paper, we define three integration levels of visualization and AI. The highest integration level is described as the framework of VIS+AI, which allows AI to learn human intelligence from interactions and communicate with humans through visual interfaces. We also summarize future directions of VIS+AI to inspire related studies.     

用AI观点探讨变结构控制

本文从人工智能（ＡＩ）观点出发分析变结构控制的实质及其优缺点并探讨其改进的途径。     

A survey of modern knowledge modeling techniques

A major characteristic regarding developments in the broad field of artificial intelligence (AI) during the 1990s has been an increasing integration of AI with other disciplines. A number of other computer science fields and technologies have been used in developing intelligent systems, starting from traditional information systems and databases, to modern distributed systems and the Internet. This paper surveys the knowledge modeling techniques that have received most attention in recent years among developers of intelligent systems, AI practitioners and researchers. The techniques are described from two perspectives, theoretical and practical. Hence the first part of the paper presents major theoretical and architectural concepts, design approaches, and research issues. The second part deals with several practical systems, applications, and ongoing projects that use and implement the techniques described in the first part.     

Pragmatism and purism in artificial intelligence and legal reasoning

The paper identifies and assesses the implications of two approaches to the field of artificial intelligence and legal reasoning. The first — pragmatism — concentrates on the development of working systems to the exclusion of theoretical problems. The second — purism — focuses on the nature of the law and of intelligence with no regard for the delivery of commercially viable systems. Past work in AI and law is classified in terms of this division. By reference to The Latent Damage System, an operational system, the paper articulates and responds to conceivable purist (jurisprudential and AI) objections to such a program. The methods of the pragmatist are also called into question and refined. The author concludes that pragmatism within a purist framework is the only sound approach to developing reliable AI systems in law.     

Principles of intelligent learning systems design

One of the possible system approaches to the construction of artificial intelligence (AI) systems is described in this paper. The approach integrates three well-known AI trends: heuristic programming, structural modelling and simulated evolution. The structure of an experimental learning system ELS Y is described. The system is designed in accordance with the proposed principles. The main feature of the system is a dynamic generation of AIS architectures on the basis of means-ends analysis method. An AIS architecture can be subjected to mutation in order to obtain a more intelligent one. Every AI system is capable of self-organization on two levels: the first one formed with an associative computing memory, and the second one with arrangement of knowledge structures.     

Optimization of Intelligent Approach for Low-Cost INS/GPS Navigation System

Due to the inherent highly nonlinear vehicle state error dynamics obtained from low-cost inertial navigation system (INS) and Global Positioning System (GPS) along with the unknown statistical properties of these sensors, the optimality/accuracy of the classical Kalman filter for sensor fusion is not guaranteed. Therefore, in this paper, low-cost INS/GPS measurement integration is optimized based on different artificial intelligence (AI) techniques: Neural Networks (NN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) architectures. The proposed approaches are aimed at achieving high-accuracy vehicle state estimates. The architectures utilize overlapping windows for delayed input signals. Both the NN approaches and the ANFIS approaches are used once with overlapping position windows as the input and once with overlapping position and velocity windows as the input. Experimental tests are conducted to evaluate the performance of the proposed AI approaches. The achieved accuracy is presented and discussed. The study finds that using ANFIS, with both position and velocity as input, provides the best estimates of position and velocity in the navigation system. Therefore, the dynamic input delayed ANFIS approach is further analyzed at the end of the paper. The effect of the input window size on the accuracy of state estimation is also discussed.     

The artificial intelligence approach to pattern recognition—a perspective and an overview

The present paper reviews the techniques for automated extraction of information from signals. The techniques may be classified broadly into two categories—the conventional pattern recognition approach and the artificial intelligence (AI) based approach. The conventional approach comprises two methodologies—statistical and structural. The paper reviews salient issues in the application of conventional techniques for extraction of information. The systems that use the artificial intelligence approach are characterized with respect to three key properties. The basic differences between the approaches and the computational aspects are reviewed. Current trends in the use of the AI approach are indicated. Some key ideas in current literature are reviewed.     

Towards the design and implementation of computer-modeled opponents for use in entertainment and educational software: Some initial considerations

While advances in computer technology have led to major improvements in the presentation and appearance of modern computer games, there has been no equivalent improvement in the artificial intelligence (AI) performance of these games. Traditional AI designs have had surprisingly little impact on the development of game-based intelligent systems. New approaches to AI, particularly those using autonomous software agents encapsulated in software robots (softbots), have the potential to make an enormous impact on game-based AI systems.Softbot-based autonomous software agents provide a framework capable of supporting psychologically based models of human behavior in a game environment. Used in this manner, softbots provide a new set of tools to explore the synthesis of human-like reasoning in complex and dynamic environments.This paper explores some initial attempts to develop an exploratory, game-based, experimental environment capable of hosting softbot-based, psychological models of human behavior. Some preliminary work in the design and implementation of a simple, extensible, softbot-based, computer-modeled opponent is also discussed.     

A situation-theoretic account of nonmonotonic reasoning

Situation theory in the sense of Barwise and Perry (1983) and nonmonotonic reasoning have been relatively disparate research programs in AI, the former focusing on computational approaches to natural language processing while the latter has been extensively used as a basic architecture for rational agents. The aim of the present paper is to suggest one way that these two approaches might fit together. Specifically, a situation semantics is given for a system of defeasible reasoning.     

Collective intelligence: it's all in the numbers

AI has been an exporter of ideas to computing in general (neural networks, agents, though robotics is more complex). But AI is now embracing ideas from elsewhere that were initially scorned because they were thought to have nothing to do with modeling intelligence and, especially, human intelligence. These are the statistical and probabilistic approaches to information capture and use that have become particularly prominent in machine learning but have spread all over AI in the last two decades. Pattern recognition was accepted in particular areas, like machine vision, as a kind of technological fix. But statistical and probabilistic approaches are now mainstream.     

关于图象分割性能评估的评述

概述了图象分割性能评估发的发展,总结了分割性能评估的基本理论框架：确定图象分割性能评估指标,构造评估测试图象集,评估模型与实验分析,以及分割性能评估的常用方法：统计法,基于ＡＩ的方法和混合法。对评估模型的设计作一些尝试性的探讨。     

Daily peak ozone forecast in Istanbul

In this study, we attempt to develop an ozone forecast model using two different approaches. The first approach is to use a multiple linear regression method and the second is to use a feed-forward artificial neural network. Models are developed for the ozone period of April through to September of the years 2002 and 2003 and verified for May to August 2004. In both models, 19 predictors are used. Calculated agreement indices (AI) for the model development period are 0.82 for the linear regression model and 0.88 for the artificial neural network model. On the other hand, AI values decrease to 0.53 and 0.64 for the validation period. Poor performance of the models in the validation phase might be due to the different maximum daily ozone averages of these two periods. While the average of maximum ozone values is 61.1 μg m^?3 in the model development phase, it is 42.2 μg m^?3 in the model validation phase.     

基于权值图的基因芯片数据差异双聚类挖掘算法*

研究了从基因芯片中挖掘差异双聚类的算法。差异双聚类中的基因在不同类别的数据中表达水准不同,这样的差异双聚类可以有效地找出影响基因表达水平的关键实验因素以及对实验条件敏感的基因。传统的双聚类方法采取分别在两类基因数据中找出聚类,再进行比较以得到最终的差异双聚类,该策略的时间效率不高。为了快速地找出差异双聚类,提出一个全新的基于权值图的差异双聚类方法,该方法的主要创新之处在于直接在由两类数据构成的权值图上挖掘双聚类,避免了分别挖掘再比较的步骤。实验结果证实该算法具有较高的运行效率。     

Formal systems in Artificial Intelligence: an illustration using semigroup,automata and language theory

As the field of Artificial Intelligence (AI) matures, methods relying less on ad hoc procedures and more on well defined operations in the context of formal structures are emerging. Evidence for this can be found, for example, in the increasingly sophisticated approaches to inexact reasoning in knowledge-based systems and in the mathematical framework developed for semantic net representation. This tutorial and brief survey presents a facet of formal language theory which provides a formalism for naturally representing and operating on strings of symbols and which also admits obvious means of handling issues such as specialization and generalization, pervasive in AI. The broad subject area covered is part of semigroup theory which has a natural relationship to both finite state automata theory and regular languages. This paper aims to introduce some of the underlying theory relevant to representation issues in AI in this context.