期刊文献+
共找到4篇文章
< 1 >
每页显示 20 50 100
预测资源分配::马尔可夫决策过程的无监督学习
1
作者 吴佳骏 赵剑羽 +1 位作者 孙乘坚 杨晨阳 《中国科学:信息科学》 CSCD 北大核心 2024年第8期1983-2000,共18页
当已知未来的移动轨迹等信息时,面向视频点播业务的预测资源分配可以在满足用户体验的前提下降低基站能耗或提高网络吞吐量传统的预测资源分配方法采用先预测用户轨迹等信息再优化功率等资源分配的方法,在预测窗较长时预测误差大,导致... 当已知未来的移动轨迹等信息时,面向视频点播业务的预测资源分配可以在满足用户体验的前提下降低基站能耗或提高网络吞吐量传统的预测资源分配方法采用先预测用户轨迹等信息再优化功率等资源分配的方法,在预测窗较长时预测误差大,导致预测所带来的增益降低.为了解决这个问题,近期已有文献把预测资源分配建模为马尔可夫决策过程,采用深度强化学习进行在线决策.然而,对于这类适于采用强化学习的马尔可夫决策过程,现有文献往往以试错的方式对状态进行设计.此外,对于有约束的优化问题,现有利用强化学习解决无线问题的方法大多通过在奖励函数上加入包含需要手动调节超参数的惩罚项满足约束.本文以移动用户视频播放不卡顿约束下使基站发射能耗最小的问题为例,提出在线求解预测资源分配的无监督深度学习方法对信息预测和资源分配进行联合优化,并建立这种方法与深度强化学习的联系.所提出的方法可以通过在线端到端无监督深度学习提高预测资源分配的性能,能以系统化而非试错式的方式设计状态,可以自动而非通过引入超参来满足复杂的约束.仿真结果表明,所提出的在线无监督深度学习与深度强化学习所达到的发射能耗相近,但能够简化状态的设计,验证了理论分析结果. 展开更多
关键词 预测资源分配 马尔可夫决策过程 无监督深度学习 深度强化学习 状态设计 复杂约束
原文传递
CsRAXs negatively regulate leaf size and fruiting ability through auxin glycosylation in cucumber
2
作者 Jiacai Chen Liu Liu +8 位作者 Guangxin Chen Shaoyun Wang Ye Liu Zeqin Zhang Hongfei Li Liming Wang zhaoyang Zhou jianyu zhao Xiaolan Zhang 《Journal of Integrative Plant Biology》 SCIE CAS CSCD 2024年第5期1024-1037,共14页
Leaves are the main photosynthesis organ that directly determines crop yield and biomass.Dissecting the regulatory mechanism of leaf development is crucial for food security and ecosystem turn-over.Here,we identified ... Leaves are the main photosynthesis organ that directly determines crop yield and biomass.Dissecting the regulatory mechanism of leaf development is crucial for food security and ecosystem turn-over.Here,we identified the novel function of R2R3-MYB transcription factors CsRAXs in regulating cucumber leaf size and fruiting ability.Csrax5 single mutant exhibited enlarged leaf size and stem diameter,and Csrax1/2/5 triple mutant displayed further enlargement phenotype.Overexpression of CsRAX1 or CsRAX5 gave rise to smaller leaf and thinner stem.The fruiting ability of Csrax1/2/5 plants was significantly enhanced,while that of CsRAX5 overexpression lines was greatly weakened.Similarly,cell number and free auxin level were elevated in mutant plants while decreased in overexpression lines.Biochemical data indicated that CsRAX1/5 directly promoted the expression of auxin glucosyltransferase gene CsUGT74E2.Therefore,our data suggested that CsRAXs function as repressors for leaf size development by promoting auxin glycosylation to decrease free auxin level and cell division in cucumber.Our findings provide new gene targets for cucumber breeding with increased leaf size and crop yield. 展开更多
关键词 auxin glycosylation CsRAXs CsUGT74E2 fruiting ability leaf size
原文传递
居家办公会影响上市公司与投资者沟通吗?——基于投资者互动平台的文本分析
3
作者 赵健宇 侯一凡 林东杰 《中国会计评论》 2023年第4期659-684,共26页
越来越多的企业实行居家办公制度,该制度对上市公司与投资者沟通的效率会产生何种影响是理论界与实务界都十分关注的话题。本文采用中国A股上市公司在上交所“e互动”和深交所“互动易”平台上的投资者问答数据,借助新冠疫情暴发以来各... 越来越多的企业实行居家办公制度,该制度对上市公司与投资者沟通的效率会产生何种影响是理论界与实务界都十分关注的话题。本文采用中国A股上市公司在上交所“e互动”和深交所“互动易”平台上的投资者问答数据,借助新冠疫情暴发以来各地疫情防控响应级别调整这一外生事件,检验居家办公模式如何影响上市公司董事会秘书与投资者的沟通效率。实证结果表明,在居家办公期间,董事会秘书与投资者在网络平台上的沟通效率更高,具体体现为董事会秘书回复投资者问题的及时性和质量更高。机制检验发现,对于通勤时间长的董事会秘书、女性董事会秘书、所在公司数字化转型程度高的董事会秘书和任期长的董事会秘书,居家办公制度对其与投资者沟通效率的提升作用更显著。进一步地,我们还发现董事会秘书居家办公期间不回复投资者提问的情况有所减少,且在投资者互动平台上做出的回应具有较高的信息含量。本文为居家办公制度对上市公司与投资者沟通效率的影响提供了经验证据,也为企业如何实现居家办公、发展居家办公模式提供了重要见解。 展开更多
关键词 居家办公 董事会秘书 投资者沟通 投资者互动平台
原文传递
EVALUATING QUINOA LODGING RISK AND YIELD UNDER DIFFERENT IRRIGATION THRESHOLDS, NITROGEN RATES AND PLANTING DENSITIES IN NORTH-WESTERN CHINA
4
作者 Ning WANG Fengxin WANG +4 位作者 Clinton C.SHOCK Lei GAO Chaobiao MENG Zejun HUANG jianyu zhao 《Frontiers of Agricultural Science and Engineering》 2022年第4期614-626,共13页
Lodging is a major yield-limiting factor of quinoa production.In 2018 and 2019,the orthogonal field experiments were conducted to investigate the responses of quinoa lodging risk and yield to irrigation threshold(soil... Lodging is a major yield-limiting factor of quinoa production.In 2018 and 2019,the orthogonal field experiments were conducted to investigate the responses of quinoa lodging risk and yield to irrigation threshold(soil matric potential of−15,−25 and−55 kPa),nitrogen rate(80,160 and 240 kg·ha^(−1))and planting density(20,30 and 40 plants m^(−2)).Results showed that high irrigation thresholds and nitrogen rates significantly(P<0.05)increased plant height and fresh weight per plant,and high planting densities reduced stem diameter and strength,all of those led to significantly(P<0.05)high lodging risks.The−15 and−55 kPa treatments gave the lowest actual yield(P<0.05)in 2018 and 2019,respectively.Higher lodging rate with a nitrogen rate of 240 kg·ha^(−1) resulted in a lower actual yield than 80 and 160 kg·ha^(−1) in both years.Planting density of 30 plants m^(−2) gave a significantly(P<0.05)greater estimated yield than 20 plants m^(−2) and had a lower lodging rate than 40 plants m^(−2),resulting in the maximum actual yield among planting densities.In conclusion,a moderate irrigation threshold of−25 kPa,a nitrogen rate of 80−160 kg·ha^(−1) and an intermediate planting density of 30 plants m−2 were determined to be best for quinoa cultivation in North-western China.In addition,the lower-stem lodging index(quarter plant height)could evaluate lodging risk more accurately than middle-stem(half plant height)or upper-stem(three quarters plant height)lodging indexes. 展开更多
关键词 lodging index orthogonal design soil matric potential stem strength
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部