Many digital platforms have employed free-content promotion strategies to deal with the high uncertainty levels regarding digital content products.However,the diversity of digital content products and user heterogenei...Many digital platforms have employed free-content promotion strategies to deal with the high uncertainty levels regarding digital content products.However,the diversity of digital content products and user heterogeneity in content preference may blur the impact of platform promotions across users and products.Therefore,free-content promotion strategies should be adapted to allocate marketing resources optimally and increase revenue.This study develops personal-ized free-content promotion strategies based on individual-level heterogeneous treatment effects and explores the causes of their heterogeneity,focusing on the moderating effect of user engagement-related variables.To this end,we utilize ran-dom field experimental data provided by a top Chinese e-book platform.We employ a framework that combines machine learning with econometric causal inference methods to estimate individual treatment effects and analyze their potential mechanisms.The analysis shows that,on average,free-content promotions lead to a significant increase in consumer pay-ments.However,the higher the level of user engagement,the lower the payment lift caused by promotions,as more-engaged users are more strongly affected by the cannibalization effect of free-content promotion.This study introduces a novel causal research design to help platforms improve their marketing strategies.展开更多
Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices...Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.展开更多
The time-dependent behaviors of coal and rocks were easily ignored. Besides, “three-stage” triaxial loading and unloading mechanics tests of sandstone were conducted based on the idea of the initial high in-situ str...The time-dependent behaviors of coal and rocks were easily ignored. Besides, “three-stage” triaxial loading and unloading mechanics tests of sandstone were conducted based on the idea of the initial high in-situ stress state recovery according to the full-life cycle evolution characteristics of surrounding rocks in deep mines(pre-excavation,excavation and post-excavation). The time-dependent stress-strain curves of sandstone were obtained. Meanwhile, the deformation and strength fitting relationships with time of sandstone were also built. Furthermore, the dilatancy and volumetric recovery mechanical mechanisms of sandstone were revealed. The results showed that: 1) There were significant time-dependent evolution characteristics on the deformation and strength of sandstone;2) There were significant correlations among the internal friction angle, cohesion and the simulated depths;3) Volumetric recovery phenomenon of sandstone was observed for the first time, which mainly occurred at the simulated depth of 2000 m. The above research conclusions could provide a certain theoretical basis for the stability control of surrounding rocks in deep mines.展开更多
日益频繁的鸟类活动给输电线路的安全运行带来了极大威胁,而现有拟声驱鸟装置由于缺乏智能性,无法长期有效驱鸟.为了解决上述问题,本文提出基于改进Q⁃learning算法的拟声驱鸟策略.首先,为了评价各音频的驱鸟效果,结合模糊理论,将鸟类听...日益频繁的鸟类活动给输电线路的安全运行带来了极大威胁,而现有拟声驱鸟装置由于缺乏智能性,无法长期有效驱鸟.为了解决上述问题,本文提出基于改进Q⁃learning算法的拟声驱鸟策略.首先,为了评价各音频的驱鸟效果,结合模糊理论,将鸟类听到音频后的动作行为量化为不同鸟类反应类型.然后,设计单一音频驱鸟实验,统计各音频驱鸟效果数据,得到各音频的初始权重值,为拟声驱鸟装置的音频选择提供实验依据.为了使计算所得的音频权重值更符合实际实验情况,对CRITIC(Criteria Impor⁃tance Though Intercrieria Correlation)方法的权重计算公式进行了优化.最后,使用实验所得音频权重值对Q⁃learning算法进行改进,并设计与其他拟声驱鸟策略的对比实验,实验数据显示改进Q⁃learning算法的拟声驱鸟策略驱鸟效果优于其他三种驱鸟策略,收敛速度快,驱鸟效果稳定,能够降低鸟类的适应性.展开更多
Brain functional networks model the brain's ability to exchange information across different regions,aiding in the understanding of the cognitive process of human visual attention during target searching,thereby c...Brain functional networks model the brain's ability to exchange information across different regions,aiding in the understanding of the cognitive process of human visual attention during target searching,thereby contributing to the advancement of camouflage evaluation.In this study,images with various camouflage effects were presented to observers to generate electroencephalography(EEG)signals,which were then used to construct a brain functional network.The topological parameters of the network were subsequently extracted and input into a machine learning model for training.The results indicate that most of the classifiers achieved accuracy rates exceeding 70%.Specifically,the Logistic algorithm achieved an accuracy of 81.67%.Therefore,it is possible to predict target camouflage effectiveness with high accuracy without the need to calculate discovery probability.The proposed method fully considers the aspects of human visual and cognitive processes,overcomes the subjectivity of human interpretation,and achieves stable and reliable accuracy.展开更多
基金supported by the Anhui Postdoctoral Scientific Research Program Foundation(2022B579).
文摘Many digital platforms have employed free-content promotion strategies to deal with the high uncertainty levels regarding digital content products.However,the diversity of digital content products and user heterogeneity in content preference may blur the impact of platform promotions across users and products.Therefore,free-content promotion strategies should be adapted to allocate marketing resources optimally and increase revenue.This study develops personal-ized free-content promotion strategies based on individual-level heterogeneous treatment effects and explores the causes of their heterogeneity,focusing on the moderating effect of user engagement-related variables.To this end,we utilize ran-dom field experimental data provided by a top Chinese e-book platform.We employ a framework that combines machine learning with econometric causal inference methods to estimate individual treatment effects and analyze their potential mechanisms.The analysis shows that,on average,free-content promotions lead to a significant increase in consumer pay-ments.However,the higher the level of user engagement,the lower the payment lift caused by promotions,as more-engaged users are more strongly affected by the cannibalization effect of free-content promotion.This study introduces a novel causal research design to help platforms improve their marketing strategies.
基金supported by the National Natural Science Foundation of China(62171088,U19A2052,62020106011)the Medico-Engineering Cooperation Funds from University of Electronic Science and Technology of China(ZYGX2021YGLH215,ZYGX2022YGRH005)。
文摘Deep neural networks(DNNs)have achieved great success in many data processing applications.However,high computational complexity and storage cost make deep learning difficult to be used on resource-constrained devices,and it is not environmental-friendly with much power cost.In this paper,we focus on low-rank optimization for efficient deep learning techniques.In the space domain,DNNs are compressed by low rank approximation of the network parameters,which directly reduces the storage requirement with a smaller number of network parameters.In the time domain,the network parameters can be trained in a few subspaces,which enables efficient training for fast convergence.The model compression in the spatial domain is summarized into three categories as pre-train,pre-set,and compression-aware methods,respectively.With a series of integrable techniques discussed,such as sparse pruning,quantization,and entropy coding,we can ensemble them in an integration framework with lower computational complexity and storage.In addition to summary of recent technical advances,we have two findings for motivating future works.One is that the effective rank,derived from the Shannon entropy of the normalized singular values,outperforms other conventional sparse measures such as the?_1 norm for network compression.The other is a spatial and temporal balance for tensorized neural networks.For accelerating the training of tensorized neural networks,it is crucial to leverage redundancy for both model compression and subspace training.
基金Supported by the State Key Program of National Natural Science of China（70931001）, the Science Fund for Creative Research Group of National Natural Science Foundation of China （60821063）, National Science and Technology Support Plan of China （2006BAH02A09）, the Science Fund for Youth Scholars of Ministry of Education of China （200801451053）, and the Research Committee and the Department of Industrial and Systems Engineering of Hong Kong Polytechnic University Research Grants （G-U323）
基金Projects(52034009, 51974319) supported by the National Natural Science Foundation of ChinaProject(2020JCB01) supported by the Yue Qi Distinguished Scholar Project of China。
文摘The time-dependent behaviors of coal and rocks were easily ignored. Besides, “three-stage” triaxial loading and unloading mechanics tests of sandstone were conducted based on the idea of the initial high in-situ stress state recovery according to the full-life cycle evolution characteristics of surrounding rocks in deep mines(pre-excavation,excavation and post-excavation). The time-dependent stress-strain curves of sandstone were obtained. Meanwhile, the deformation and strength fitting relationships with time of sandstone were also built. Furthermore, the dilatancy and volumetric recovery mechanical mechanisms of sandstone were revealed. The results showed that: 1) There were significant time-dependent evolution characteristics on the deformation and strength of sandstone;2) There were significant correlations among the internal friction angle, cohesion and the simulated depths;3) Volumetric recovery phenomenon of sandstone was observed for the first time, which mainly occurred at the simulated depth of 2000 m. The above research conclusions could provide a certain theoretical basis for the stability control of surrounding rocks in deep mines.
文摘日益频繁的鸟类活动给输电线路的安全运行带来了极大威胁,而现有拟声驱鸟装置由于缺乏智能性,无法长期有效驱鸟.为了解决上述问题,本文提出基于改进Q⁃learning算法的拟声驱鸟策略.首先,为了评价各音频的驱鸟效果,结合模糊理论,将鸟类听到音频后的动作行为量化为不同鸟类反应类型.然后,设计单一音频驱鸟实验,统计各音频驱鸟效果数据,得到各音频的初始权重值,为拟声驱鸟装置的音频选择提供实验依据.为了使计算所得的音频权重值更符合实际实验情况,对CRITIC(Criteria Impor⁃tance Though Intercrieria Correlation)方法的权重计算公式进行了优化.最后,使用实验所得音频权重值对Q⁃learning算法进行改进,并设计与其他拟声驱鸟策略的对比实验,实验数据显示改进Q⁃learning算法的拟声驱鸟策略驱鸟效果优于其他三种驱鸟策略,收敛速度快,驱鸟效果稳定,能够降低鸟类的适应性.
基金sponsored by the National Defense Science and Technology Key Laboratory Fund(Grant No.61422062205)the Equipment Pre-Research Fund(Grant No.JCKYS2022LD9)。
文摘Brain functional networks model the brain's ability to exchange information across different regions,aiding in the understanding of the cognitive process of human visual attention during target searching,thereby contributing to the advancement of camouflage evaluation.In this study,images with various camouflage effects were presented to observers to generate electroencephalography(EEG)signals,which were then used to construct a brain functional network.The topological parameters of the network were subsequently extracted and input into a machine learning model for training.The results indicate that most of the classifiers achieved accuracy rates exceeding 70%.Specifically,the Logistic algorithm achieved an accuracy of 81.67%.Therefore,it is possible to predict target camouflage effectiveness with high accuracy without the need to calculate discovery probability.The proposed method fully considers the aspects of human visual and cognitive processes,overcomes the subjectivity of human interpretation,and achieves stable and reliable accuracy.
