基于人类先验知识的强化学习综述

强化学习通过智能体与环境互动以实现预期目标最大化,通常被用于解决关于连续决策的任务规划问题。当前任务规划主要使用规则或者运筹方法进行求解。这类方法的时间复杂度和空间复杂度随问题规模增长呈指数级增长,难以求解大规模问题;并且对于动态任务规划问题也束手无策,只能求解静态的确定性问题。强化学习中的值函数近似等近似算法解决了这类方法的"维度灾难"问题,同时强化学习在智能体与环境的交互过程中加入随机因素,使其广泛应用于解决动态与随机性问题。然而强化学习需要智能体不断与环境交互来得到最优策略,当状态空间较大时,需要更多的采样和探索对策略进行梯度更新,收敛速度较慢,难以在实际中应用。人类在学习时利用先验知识保证了学习的快速开始,并减少了问题的探索时间,因此研究人类先验知识和强化学习的意义重大,可有效减少智能体对环境的采样和探索,帮助智能体更快地找到最优策略。基于人类先验知识与智能体交互的时间节点进行分类,对整合人类先验知识到强化学习的方法进行综述,最后对其发展方向进行展望。

电力设备ChatGPT类模式与关键技术

ChatGPT(chat generative pre-trained transformer)是近些年人工智能领域发展的新技术方向,其涵盖了设备数字孪生、设备管理、平台运行等数字化功能,更具备通用性、生成式人机对话的特点。文章首先介绍了ChatGPT的发展现状,以及电力设备ChatGPT类模式和核心技术架构,说明大模型具有优异的泛化能力、逻辑推理能力、多模态数据分析生成能力等突出特点。然后,从高算力人工智能芯片、语料样本体系构建、基于Transformer的生成式预训练模型、嵌入大语言模型的多模态算法、基于人类反馈的强化学习技术这五个方面分析了电力设备ChatGPT类模式所涉及的关键技术。最后,提出了电力设备ChatGPT在电力行业开展的可行性和技术方案,总结出未来电力设备ChatGPT所面临的挑战和发展方向。

ChatGPT技术及其对军事安全影响

OpenAI发布ChatGPT以来,引发了人类社会的巨大震荡,产学研各界纷纷关注其技术实现,思考其技术局限性与应用前景。首先,追溯了GPT家族模型发展历程,分析了GPT家族模型网络结构设计与训练过程改进中的技术思想;然后,对ChatGPT的优劣势进行分析,ChatGPT呈现出自然流畅、多轮交互、泛化能力强3大优势,但依然存在一定的技术局限性;最后,评估了ChatGPT给军事安全带来的挑战,并提出应对策略,涵盖反生成式AI技术研究、网络信息安全防御体系构建、军事智能辅助决策系统建设、军事数据资源建设4个方面。

迈向可信AI:ChatGPT类生成式人工智能的治理挑战及应对

2022年以来,随着以ChatGPT、GPT-4等大型语言模型为标志的生成式人工智能的迅猛发展,人工智能迎来大模型时代。AI大模型被认为是革命性的技术进展,将给经济社会发展带来深远影响。然而,AI大模型的快速发展应用也持续引发各界对其伦理安全风险的担忧。面对生成式人工智能的伦理安全风险和治理挑战,有必要在治理上倡导并推动践行“负责任人工智能”的理念,建立敏捷灵活的AI监管和治理框架,通过多元化的治理举措支持生成式人工智能创新发展和可靠应用,推动建立负责任的、可信的、以人为本的、合乎伦理道德的生成式人工智能发展生态。

The two-way feedback and passing-way of human body

Two-way feedback of human body was published in 1992. The sensation of two-way feedback of body is a special system of human reaction, which maintains and regulates symmetry and balance of human body. The human two-way feedback reacts to human health. For human overall health and delay decrepitude, it is necessary to pay attention to the stimulations (passive acceptance and initiative interventions) and relevant influences in human body and the stimulative effect. In this paper, the experimental research of stimulation and an example of two-way feedback in human body are given. And lay a foundation of prevention, medical treatment and hygiene of human overall health.

Research Progress on Ecological Risk Assessment based on M ultifunctional Landscape

The unreasonable utilization and trade-offs of the multifunctional landscape is one of the important causes leading to ecological risk.Quantifying the relationship,heterogeneous process and driving mechanism between landscape function and human well-being can provide a scientific basis for achieving a win-win situation of regional development and ecological protection,and it is also an important issue for relevant scientific research.In this paper,we reviewed the current state of knowledge regarding landscape function and ecological risk,identified the feedback mechanisms among landscape change,landscape function,human well-being and ecological risk,constructed the ecological risk assessment framework based on landscape functions,and sorted out the key scientific issues and research trends.The interconnections between the multifunctional landscape,human well-being,and ecological risk comprise an important scientific issue in ecological risk assessment.However,there is a lack of understanding of the spatial-temporal characteristics of landscape function trade-offs,and there is also a lack of systematic and standardized methods for the identification and comprehensive evaluation of landscape functions.Exploring ecological risk management and the methodologies that integrate the two processes of nature and humanity is an inevitable trend for future research.Coupling landscape structure,function and process into ecological risk assessment,as well as focusing on the combination of human well-being,landscape function trade-off and ecological risk control policy design are important issues to be studied.