Journal articles: 'Policy gradients'

1

Cai, Qingpeng, Ling Pan, and Pingzhong Tang. "Deterministic Value-Policy Gradients." Proceedings of the AAAI Conference on Artificial Intelligence 34, no. 04 (April 3, 2020): 3316–23. http://dx.doi.org/10.1609/aaai.v34i04.5732.

Full text

Abstract:

Reinforcement learning algorithms such as the deep deterministic policy gradient algorithm (DDPG) has been widely used in continuous control tasks. However, the model-free DDPG algorithm suffers from high sample complexity. In this paper we consider the deterministic value gradients to improve the sample efficiency of deep reinforcement learning algorithms. Previous works consider deterministic value gradients with the finite horizon, but it is too myopic compared with infinite horizon. We firstly give a theoretical guarantee of the existence of the value gradients in this infinite setting. Based on this theoretical guarantee, we propose a class of the deterministic value gradient algorithm (DVG) with infinite horizon, and different rollout steps of the analytical gradients by the learned model trade off between the variance of the value gradients and the model bias. Furthermore, to better combine the model-based deterministic value gradient estimators with the model-free deterministic policy gradient estimator, we propose the deterministic value-policy gradient (DVPG) algorithm. We finally conduct extensive experiments comparing DVPG with state-of-the-art methods on several standard continuous control benchmarks. Results demonstrate that DVPG substantially outperforms other baselines.

APA, Harvard, Vancouver, ISO, and other styles

2

Wierstra, D., A. Forster, J. Peters, and J. Schmidhuber. "Recurrent policy gradients." Logic Journal of IGPL 18, no. 5 (September 9, 2009): 620–34. http://dx.doi.org/10.1093/jigpal/jzp049.

Full text

APA, Harvard, Vancouver, ISO, and other styles

3

Sehnke, Frank, Christian Osendorfer, Thomas Rückstieß, Alex Graves, Jan Peters, and Jürgen Schmidhuber. "Parameter-exploring policy gradients." Neural Networks 23, no. 4 (May 2010): 551–59. http://dx.doi.org/10.1016/j.neunet.2009.12.004.

Full text

APA, Harvard, Vancouver, ISO, and other styles

4

Zhao, Tingting, Hirotaka Hachiya, Voot Tangkaratt, Jun Morimoto, and Masashi Sugiyama. "Efficient Sample Reuse in Policy Gradients with Parameter-Based Exploration." Neural Computation 25, no. 6 (June 2013): 1512–47. http://dx.doi.org/10.1162/neco_a_00452.

Full text

Abstract:

The policy gradient approach is a flexible and powerful reinforcement learning method particularly for problems with continuous actions such as robot control. A common challenge is how to reduce the variance of policy gradient estimates for reliable policy updates. In this letter, we combine the following three ideas and give a highly effective policy gradient method: (1) policy gradients with parameter-based exploration, a recently proposed policy search method with low variance of gradient estimates; (2) an importance sampling technique, which allows us to reuse previously gathered data in a consistent way; and (3) an optimal baseline, which minimizes the variance of gradient estimates with their unbiasedness being maintained. For the proposed method, we give a theoretical analysis of the variance of gradient estimates and show its usefulness through extensive experiments.

APA, Harvard, Vancouver, ISO, and other styles

5

Seno, Takuma, and Michita Imai. "Policy Gradients with Memory-Augmented Critic." Transactions of the Japanese Society for Artificial Intelligence 36, no. 1 (January 1, 2021): B—K71_1–8. http://dx.doi.org/10.1527/tjsai.36-1_b-k71.

Full text

APA, Harvard, Vancouver, ISO, and other styles

6

Millidge, Beren. "Deep active inference as variational policy gradients." Journal of Mathematical Psychology 96 (June 2020): 102348. http://dx.doi.org/10.1016/j.jmp.2020.102348.

Full text

APA, Harvard, Vancouver, ISO, and other styles

7

Catling, PC, and RJ Burt. "Studies of the Ground-Dwelling Mammals of Eucalypt Forests in South-Eastern New South Wales: the Effect of Environmental Variables on Distribution and Abundance." Wildlife Research 22, no. 6 (1995): 669. http://dx.doi.org/10.1071/wr9950669.

Full text

Abstract:

The distribution and abundance of ground-dwelling mammals was examined in 13 areas within 500 000 ha of eucalypt (Eucalyptus) forest in SE New South Wales. Data are presented on the distribution and abundance of species in relation to 3 environmental gradient types involving 9 variables: 2 direct gradients (temperature, rainfall); 6 indirect gradients (aspect, steepness of slope, position on slope, landform profile around the site, altitude, season) and a resource gradient (lithology). Many species of ground-dwelling mammal of the forests of SE New South Wales were present along all gradients examined, although wide variation in abundance occurred for some species. Eight species were correlated with direct gradients and all species were correlated with at least one indirect gradient. There was wide variation and species diversity with lithology, but the variation was not related to nutrient status. Although variations in abundance occurred along environmental gradients, the composition of the ground-dwelling mammal fauna in SE New South Wales forests changed little. A fourth gradient type, the substrate gradient (biomass of plants), had the greatest effect, because in the short-term disturbances such as logging and fire play an important role. Disturbance can have a profound influence on the substrate gradient, but no influence on environmental gradients. The results are discussed in relation to the arboreal mammals and avifauna in the region and Environmental and Fauna Impact studies and forest management.

APA, Harvard, Vancouver, ISO, and other styles

8

Baxter, J., P. L. Bartlett, and L. Weaver. "Experiments with Infinite-Horizon, Policy-Gradient Estimation." Journal of Artificial Intelligence Research 15 (November 1, 2001): 351–81. http://dx.doi.org/10.1613/jair.807.

Full text

Abstract:

In this paper, we present algorithms that perform gradient ascent of the average reward in a partially observable Markov decision process (POMDP). These algorithms are based on GPOMDP, an algorithm introduced in a companion paper (Baxter & Bartlett, this volume), which computes biased estimates of the performance gradient in POMDPs. The algorithm's chief advantages are that it uses only one free parameter beta, which has a natural interpretation in terms of bias-variance trade-off, it requires no knowledge of the underlying state, and it can be applied to infinite state, control and observation spaces. We show how the gradient estimates produced by GPOMDP can be used to perform gradient ascent, both with a traditional stochastic-gradient algorithm, and with an algorithm based on conjugate-gradients that utilizes gradient information to bracket maxima in line searches. Experimental results are presented illustrating both the theoretical results of (Baxter & Bartlett, this volume) on a toy problem, and practical aspects of the algorithms on a number of more realistic problems.

APA, Harvard, Vancouver, ISO, and other styles

9

Chen, Qiulin, Karen Eggleston, Wei Zhang, Jiaying Zhao, and Sen Zhou. "The Educational Gradient in Health in China." China Quarterly 230 (May 15, 2017): 289–322. http://dx.doi.org/10.1017/s0305741017000613.

Full text

Abstract:

AbstractIt has been well established that better educated individuals enjoy better health and longevity. In theory, the educational gradients in health could be flattening if diminishing returns to improved average education levels and the influence of earlier population health interventions outweigh the gradient-steepening effects of new medical and health technologies. This paper documents how the gradients are evolving in China, a rapidly developing country, about which little is known on this topic. Based on recent mortality data and nationally representative health surveys, we find large and, in some cases, steepening educational gradients. We also find that the gradients vary by cohort, gender and region. Further, we find that the gradients can only partially be accounted for by economic factors. These patterns highlight the double disadvantage of those with low education, and suggest the importance of policy interventions that foster both aspects of human capital for them.

APA, Harvard, Vancouver, ISO, and other styles

10

Peters, Jan, and Stefan Schaal. "Reinforcement learning of motor skills with policy gradients." Neural Networks 21, no. 4 (May 2008): 682–97. http://dx.doi.org/10.1016/j.neunet.2008.02.003.

Full text

APA, Harvard, Vancouver, ISO, and other styles

11

Zhang, Chuheng, Yuanqi Li, and Jian Li. "Policy Search by Target Distribution Learning for Continuous Control." Proceedings of the AAAI Conference on Artificial Intelligence 34, no. 04 (April 3, 2020): 6770–77. http://dx.doi.org/10.1609/aaai.v34i04.6156.

Full text

Abstract:

It is known that existing policy gradient methods (such as vanilla policy gradient, PPO, A2C) may suffer from overly large gradients when the current policy is close to deterministic, leading to an unstable training process. We show that such instability can happen even in a very simple environment. To address this issue, we propose a new method, called target distribution learning (TDL), for policy improvement in reinforcement learning. TDL alternates between proposing a target distribution and training the policy network to approach the target distribution. TDL is more effective in constraining the KL divergence between updated policies, and hence leads to more stable policy improvements over iterations. Our experiments show that TDL algorithms perform comparably to (or better than) state-of-the-art algorithms for most continuous control tasks in the MuJoCo environment while being more stable in training.

APA, Harvard, Vancouver, ISO, and other styles

12

Wang, Lin, Xingang Xu, Xuhui Zhao, Baozhu Li, Ruijuan Zheng, and Qingtao Wu. "A randomized block policy gradient algorithm with differential privacy in Content Centric Networks." International Journal of Distributed Sensor Networks 17, no. 12 (December 2021): 155014772110599. http://dx.doi.org/10.1177/15501477211059934.

Full text

Abstract:

Policy gradient methods are effective means to solve the problems of mobile multimedia data transmission in Content Centric Networks. Current policy gradient algorithms impose high computational cost in processing high-dimensional data. Meanwhile, the issue of privacy disclosure has not been taken into account. However, privacy protection is important in data training. Therefore, we propose a randomized block policy gradient algorithm with differential privacy. In order to reduce computational complexity when processing high-dimensional data, we randomly select a block coordinate to update the gradients at each round. To solve the privacy protection problem, we add a differential privacy protection mechanism to the algorithm, and we prove that it preserves the [Formula: see text]-privacy level. We conduct extensive simulations in four environments, which are CartPole, Walker, HalfCheetah, and Hopper. Compared with the methods such as important-sampling momentum-based policy gradient, Hessian-Aided momentum-based policy gradient, REINFORCE, the experimental results of our algorithm show a faster convergence rate than others in the same environment.

APA, Harvard, Vancouver, ISO, and other styles

13

Persson, Bertil R. R., and Freddy Ståhlberg. "Safety Aspects of Magnetic Resonance Examinations." International Journal of Technology Assessment in Health Care 1, no. 3 (July 1985): 647–65. http://dx.doi.org/10.1017/s0266462300001549.

Full text

Abstract:

In a standard whole-body NMR-scanning machine, the static magnetic field is generated by an electric current driven through large solenoid coils. Dynamic magnetic gradient fields are generated by electric current pulses in coils located at various orientations, thus producing magnetic gradients inx, y, andzdirections. The Rf (radiofrequency) radiation is transmitted through a specially shaped coil which also serves as an antenna receiving the NMR signals.

APA, Harvard, Vancouver, ISO, and other styles

14

Montalvo, Javier, Enrique Ruiz-Labrador, Pablo Montoya-Bernabéu, and Belén Acosta-Gallo. "Rural–Urban Gradients and Human Population Dynamics." Sustainability 11, no. 11 (June 1, 2019): 3107. http://dx.doi.org/10.3390/su11113107.

Full text

Abstract:

Rural–urban gradients offer an appropriate ecological framework for understanding relevant social issues to sustainability and policy planning. We tested the hypothesis that human population growth rate at a local scale is indirectly driven by spatial and rurality gradients, which can be applied to cultural landscapes in Mediterranean Europe. The whole of local administrative/spatial units of Spain—8125 municipalities—, previously classified into five categories along a rural–urban gradient, was used as a case study. Several geospatial patterns and associations among local average per capita population growth rate, population mean age, road accessibility, and other environmental and landscape variables linked to rurality gradients were identified by means of geographic information system (GIS) and multivariate statistics. Regression analysis was used to assess the relationship between population size changes through time and other demographic and territorial variables. Population growth rate was associated with road accessibility and rurality gradient, supporting the established hypothesis. Short-term population growth or decline was directly driven by population mean age. A visual hypothesized model of local population growth rate based on empirical evidence is presented. The results are useful for decision-makers, from local land management interventions to developing strategies and policies to address the demographic challenge.

APA, Harvard, Vancouver, ISO, and other styles

15

Finch, Brian Karl. "Socioeconomic Gradients and Low Birth-Weight: Empirical and Policy Considerations." Health Services Research 38, no. 6p2 (December 18, 2003): 1819–42. http://dx.doi.org/10.1111/j.1475-6773.2003.00204.x.

Full text

APA, Harvard, Vancouver, ISO, and other styles

16

Tirado, Daniel A., Jordi Pons, Elisenda Paluzie, and Julio Martínez-Galarraga. "Trade policy and wage gradients: evidence from a protectionist turn." Cliometrica 7, no. 3 (January 10, 2013): 295–318. http://dx.doi.org/10.1007/s11698-012-0090-y.

Full text

APA, Harvard, Vancouver, ISO, and other styles

17

Rauber, Paulo, Avinash Ummadisingu, Filipe Mutz, and Jürgen Schmidhuber. "Reinforcement Learning in Sparse-Reward Environments With Hindsight Policy Gradients." Neural Computation 33, no. 6 (May 13, 2021): 1498–553. http://dx.doi.org/10.1162/neco_a_01387.

Full text

Abstract:

A reinforcement learning agent that needs to pursue different goals across episodes requires a goal-conditional policy. In addition to their potential to generalize desirable behavior to unseen goals, such policies may also enable higher-level planning based on subgoals. In sparse-reward environments, the capacity to exploit information about the degree to which an arbitrary goal has been achieved while another goal was intended appears crucial to enabling sample efficient learning. However, reinforcement learning agents have only recently been endowed with such capacity for hindsight. In this letter, we demonstrate how hindsight can be introduced to policy gradient methods, generalizing this idea to a broad class of successful algorithms. Our experiments on a diverse selection of sparse-reward environments show that hindsight leads to a remarkable increase in sample efficiency.

APA, Harvard, Vancouver, ISO, and other styles

18

Crowley, Mark. "Using Equilibrium Policy Gradients for Spatiotemporal Planning in Forest Ecosystem Management." IEEE Transactions on Computers 63, no. 1 (January 2014): 142–54. http://dx.doi.org/10.1109/tc.2013.113.

Full text

APA, Harvard, Vancouver, ISO, and other styles

19

Liu, Chujun, Andrew Lonsberry, Mark Nandor, Musa Audu, Alexander Lonsberry, and Roger Quinn. "Implementation of Deep Deterministic Policy Gradients for Controlling Dynamic Bipedal Walking." Biomimetics 4, no. 1 (March 22, 2019): 28. http://dx.doi.org/10.3390/biomimetics4010028.

Full text

Abstract:

A control system for bipedal walking in the sagittal plane was developed in simulation. The biped model was built based on anthropometric data for a 1.8 m tall male of average build. At the core of the controller is a deep deterministic policy gradient (DDPG) neural network that was trained in GAZEBO, a physics simulator, to predict the ideal foot placement to maintain stable walking despite external disturbances. The complexity of the DDPG network was decreased through carefully selected state variables and a distributed control system. Additional controllers for the hip joints during their stance phases and the ankle joint during toe-off phase help to stabilize the biped during walking. The simulated biped can walk at a steady pace of approximately 1 m/s, and during locomotion it can maintain stability with a 30 kg·m/s impulse applied forward on the torso or a 40 kg·m/s impulse applied rearward. It also maintains stable walking with a 10 kg backpack or a 25 kg front pack. The controller was trained on a 1.8 m tall model, but also stabilizes models 1.4–2.3 m tall with no changes.

APA, Harvard, Vancouver, ISO, and other styles

20

Zhang, Yifan, Qinghe Zhao, Zihao Cao, and Shengyan Ding. "Inhibiting Effects of Vegetation on the Characteristics of Runoff and Sediment Yield on Riparian Slope along the Lower Yellow River." Sustainability 11, no. 13 (July 4, 2019): 3685. http://dx.doi.org/10.3390/su11133685.

Full text

Abstract:

Riparian vegetation plays a vital role in soil and water conservation and river health maintenance. However, its inhibiting effects on water and soil loss are limited by different factors, such as slope gradient, vegetation coverage and their interaction. Therefore, this study quantified the inhibiting effect of riparian vegetation on the runoff, sediment and hydraulic characteristics of overland flow, and assessed its relative contribution to slope gradient. Specifically, we selected a riparian slope along the lower Yellow River as a case, and used a field-simulated rainfall experiment under specific rainfall intensity (90 mm/h), different vegetation coverage (0%, 15% and 30%) and slope gradients (5°, 10°, 15° and 20°). The results showed that the presence of vegetation can reduce the slope runoff rate and erosion rate. However, greater slope gradients can result in a lowering of the inhibiting effects of riparian vegetation on sediment yield. There was a critical value of vegetation coverage for inhibiting eroded sediments which was influenced by the degree of slope gradient. At 15% vegetation coverage, vegetation inhibited the slope sediment yield greatly at a slope gradient of less than 8°; while at 30% vegetation coverage, vegetation greatly inhibited the slope sediment yield at slope gradients <11°. Hydraulic characteristics were closely related to the slope gradient and vegetation coverage by the power function. Grey correlation analysis revealed that, with increasing of vegetation coverage, the effect of stream power on slope sediment yield decreased, while the effect of the friction coefficient on slope sediment yield increased. In summary, riparian vegetation can effectively inhibit slope runoff and sediment yield, but its inhibiting effect is notably affected by slope gradient.

APA, Harvard, Vancouver, ISO, and other styles

21

MARRIOTT, M. J. "Self-Cleansing Sewer Gradients." Water and Environment Journal 8, no. 4 (August 1994): 360–61. http://dx.doi.org/10.1111/j.1747-6593.1994.tb01118.x.

Full text

APA, Harvard, Vancouver, ISO, and other styles

22

Li, Kai, Yousef Emami, Wei Ni, Eduardo Tovar, and Zhu Han. "Onboard Deep Deterministic Policy Gradients for Online Flight Resource Allocation of UAVs." IEEE Networking Letters 2, no. 3 (September 2020): 106–10. http://dx.doi.org/10.1109/lnet.2020.3002341.

Full text

APA, Harvard, Vancouver, ISO, and other styles

23

Grondman, Ivo, Lucian Busoniu, Gabriel A. D. Lopes, and Robert Babuska. "A Survey of Actor-Critic Reinforcement Learning: Standard and Natural Policy Gradients." IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) 42, no. 6 (November 2012): 1291–307. http://dx.doi.org/10.1109/tsmcc.2012.2218595.

Full text

APA, Harvard, Vancouver, ISO, and other styles

24

Chu, Baeksuk, Daehie Hong, and Jooyoung Park. "Tunnel ventilation control via an actor-critic algorithm employing nonparametric policy gradients." Journal of Mechanical Science and Technology 23, no. 2 (February 2009): 311–23. http://dx.doi.org/10.1007/s12206-008-0924-5.

Full text

APA, Harvard, Vancouver, ISO, and other styles

25

Scott-Marshall, Heather. "Occupational Gradients in Work-Related Insecurity and Health: Interrogating the Links." International Journal of Health Services 49, no. 2 (March 6, 2019): 212–36. http://dx.doi.org/10.1177/0020731419832243.

Full text

Abstract:

Traditional work-related securities that constitute the career-job model of employment have been in steep decline for several decades, affecting workers across industries and occupations. Still, insecure employment remains unequally distributed across the working population according to the major axes of social stratification, namely age, gender, race, and socioeconomic class. This study investigates patterns of exposure to work-related insecurity across the occupational hierarchy and whether these contribute to occupational gradients in health outcomes. Drawing on data from a national panel survey of the Canadian workforce, a multilevel growth curve modeling approach is used to examine the relationship between work-insecurity exposures and workers’ self-rated health trajectories over 5 years. Findings show that work-related insecurity is associated with declines in self-rated health, although the type of insecurity as well as the magnitude, direction, and duration of the effect varies by occupational status-position. The application of pseudo-R2 tests confirmed this study’s central hypothesis that gradients in health outcomes across occupational hierarchies are due, in part, to differences in exposure to work-related insecurity. Going forward, the development of effective health promotion interventions that can modify work-related health gradients, must work toward mitigating the risk of exposure to adverse work circumstances that is systemic to occupational hierarchies.

APA, Harvard, Vancouver, ISO, and other styles

26

Seo, Paul Hongsuck, Piyush Sharma, Tomer Levinboim, Bohyung Han, and Radu Soricut. "Reinforcing an Image Caption Generator Using Off-Line Human Feedback." Proceedings of the AAAI Conference on Artificial Intelligence 34, no. 03 (April 3, 2020): 2693–700. http://dx.doi.org/10.1609/aaai.v34i03.5655.

Full text

Abstract:

Human ratings are currently the most accurate way to assess the quality of an image captioning model, yet most often the only used outcome of an expensive human rating evaluation is a few overall statistics over the evaluation dataset. In this paper, we show that the signal from instance-level human caption ratings can be leveraged to improve captioning models, even when the amount of caption ratings is several orders of magnitude less than the caption training data. We employ a policy gradient method to maximize the human ratings as rewards in an off-policy reinforcement learning setting, where policy gradients are estimated by samples from a distribution that focuses on the captions in a caption ratings dataset. Our empirical evidence indicates that the proposed method learns to generalize the human raters' judgments to a previously unseen set of images, as judged by a different set of human judges, and additionally on a different, multi-dimensional side-by-side human evaluation procedure.

APA, Harvard, Vancouver, ISO, and other styles

27

Cao, Yongcan, and Huixin Zhan. "Efficient Multi-objective Reinforcement Learning via Multiple-gradient Descent with Iteratively Discovered Weight-Vector Sets." Journal of Artificial Intelligence Research 70 (January 20, 2021): 319–49. http://dx.doi.org/10.1613/jair.1.12270.

Full text

Abstract:

Solving multi-objective optimization problems is important in various applications where users are interested in obtaining optimal policies subject to multiple (yet often conflicting) objectives. A typical approach to obtain the optimal policies is to first construct a loss function based on the scalarization of individual objectives and then derive optimal policies that minimize the scalarized loss function. Albeit simple and efficient, the typical approach provides no insights/mechanisms on the optimization of multiple objectives due to the lack of ability to quantify the inter-objective relationship. To address the issue, we propose to develop a new efficient gradient-based multi-objective reinforcement learning approach that seeks to iteratively uncover the quantitative inter-objective relationship via finding a minimum-norm point in the convex hull of the set of multiple policy gradients when the impact of one objective on others is unknown a priori. In particular, we first propose a new PAOLS algorithm that integrates pruning and approximate optimistic linear support algorithm to efficiently discover the weight-vector sets of multiple gradients that quantify the inter-objective relationship. Then we construct an actor and a multi-objective critic that can co-learn the policy and the multi-objective vector value function. Finally, the weight discovery process and the policy and vector value function learning process can be iteratively executed to yield stable weight-vector sets and policies. To validate the effectiveness of the proposed approach, we present a quantitative evaluation of the approach based on three case studies.

APA, Harvard, Vancouver, ISO, and other styles

28

Ohmann, Janet L., and Matthew J. Gregory. "Predictive mapping of forest composition and structure with direct gradient analysis and nearest- neighbor imputation in coastal Oregon, U.S.A." Canadian Journal of Forest Research 32, no. 4 (April 1, 2002): 725–41. http://dx.doi.org/10.1139/x02-011.

Full text

Abstract:

Spatially explicit information on the species composition and structure of forest vegetation is needed at broad spatial scales for natural resource policy analysis and ecological research. We present a method for predictive vegetation mapping that applies direct gradient analysis and nearest-neighbor imputation to ascribe detailed ground attributes of vegetation to each pixel in a digital landscape map. The gradient nearest neighbor method integrates vegetation measurements from regional grids of field plots, mapped environmental data, and Landsat Thematic Mapper (TM) imagery. In the Oregon coastal province, species gradients were most strongly associated with regional climate and geographic location, whereas variation in forest structure was best explained by Landsat TM variables. At the regional level, mapped predictions represented the range of variability in the sample data, and predicted area by vegetation type closely matched sample-based estimates. At the site level, mapped predictions maintained the covariance structure among multiple response variables. Prediction accuracy for tree species occurrence and several measures of vegetation structure and composition was good to moderate. Vegetation maps produced with the gradient nearest neighbor method are appropriately used for regional-level planning, policy analysis, and research, not to guide local management decisions.

APA, Harvard, Vancouver, ISO, and other styles

29

Chen, Yizhu, Nuanyin Xu, Qianru Yu, and Luo Guo. "Ecosystem Service Response to Human Disturbance in the Yangtze River Economic Belt: A Case of Western Hunan, China." Sustainability 12, no. 2 (January 8, 2020): 465. http://dx.doi.org/10.3390/su12020465.

Full text

Abstract:

Ecosystem conservation is one of the core elements of sustainable development. Studying the relationship between human disturbance and the ecosystem service value (ESV) change is an urgent need for the future. The Yangtze River Economic Belt is one of the key economic strategies implemented by the Chinese government and is also a demonstration zone for ecological conservation. Western Hunan is an important ecological barrier in the Yangtze basin where different ethnic groups live together and various cultures coexist. In this study, using land-use data and spatial analysis modeling, the changes in the ecosystem service value at five topographic gradients were evaluated. Human disturbance and its spatial correlation with the ecosystem service value from 1990 to 2015 were also investigated. The results demonstrated the following: (1) the proportional area of forestland and grassland increased as the topographic gradient index increased and other types of land-use gradually decreased; (2) The ecosystem service value at middle gradients increased over the study period; but ESV of the lowest topographic gradient showed a significant decline and a substantial decrease, as well as a terrain index under 0.7970; (3) The spatial analysis of human disturbance showed that more than 90% of intense human disturbance was distributed in the area of the lowest topographic gradient where topographic features were low-altitude and low-slope, and little human disturbance was scattered at other gradients; (4) There was a significant spatial aggregation distribution between the ecosystem service value and human disturbance in western Hunan, the high disturbance and low ESV aggregation was mainly distributed in Loudi City, the area east of Shaoyang City and Zhangjiajie City all belonged to the lowest topographic gradient, and the low–high and high–high aggregations were mainly distributed in Huaihua City and Xiangxi Tujia and Miao Autonomous Prefecture. Population density and gross domestic product were the main driving factors, while topography was the main ecological factor. This study could provide additional spatial information and theoretical guidance for ecosystem service management for sustainable development in western Hunan, China.

APA, Harvard, Vancouver, ISO, and other styles

30

Jiang, Zhong An, and Lan Jiang. "Gradient Analysis on Occupational Safety in Mine and Economic and Social Development." Advanced Materials Research 524-527 (May 2012): 3107–11. http://dx.doi.org/10.4028/www.scientific.net/amr.524-527.3107.

Full text

Abstract:

In order to explore the effect of macroeconomic policies on mine safety, selecting data of 36 economic and social indicators and occupational safety indicators in mining industry in recent 15 years. Then, based on the analysis of the traditional growth rate, processing dimensionless and benchmark correlation of data, the Average weights rank of economic and social indicators relative to the mining safety could finally be got. The advantage of gradient analysis is dropping the dimension interference of data, and using three progressive connective gradients to get the relationship between occupational safety indexes and economic and social indicators. Results of the analysis were also provided some references for the improvement of macroeconomic policy.

APA, Harvard, Vancouver, ISO, and other styles

31

GRAHAM, HILARY. "Tackling Inequalities in Health in England: Remedying Health Disadvantages, Narrowing Health Gaps or Reducing Health Gradients?" Journal of Social Policy 33, no. 1 (January 2004): 115–31. http://dx.doi.org/10.1017/s0047279403007220.

Full text

Abstract:

Socioeconomic inequalities in health have moved up the policy agenda of older industrial societies. The paper turns the spotlight on this development by exploring how the goal has been represented in England's national policy documents. Rather than one approach, there appears to be a range of understandings of what it means to tackle health inequalities. These understandings can be placed on a continuum, which runs from improving the health of poor groups, through closing the health gaps between those in the poorest circumstances and better-off groups, to addressing the association between socioeconomic position and health across the population. The paper points to common ground between the three approaches to tackling health inequalities, but also to important differences in the moral arguments and causal models on which they rest, and therefore in their policy goals and anticipated policy impacts.

APA, Harvard, Vancouver, ISO, and other styles

32

Tangkaratt, Voot, Syogo Mori, Tingting Zhao, Jun Morimoto, and Masashi Sugiyama. "Model-based policy gradients with parameter-based exploration by least-squares conditional density estimation." Neural Networks 57 (September 2014): 128–40. http://dx.doi.org/10.1016/j.neunet.2014.06.006.

Full text

APA, Harvard, Vancouver, ISO, and other styles

33

Schellberg, J., and L. da S. Pontes. "Plant functional traits and nutrient gradients on grassland." Grass and Forage Science 67, no. 3 (April 5, 2012): 305–19. http://dx.doi.org/10.1111/j.1365-2494.2012.00867.x.

Full text

APA, Harvard, Vancouver, ISO, and other styles

34

Yasir, Achmad Ichwan, and Gede Putra Kusuma. "Deep Deterministic Policy Gradients for Optimizing Simulated PoA Blockchain Networks Based on Healthcare Data Characteristics." Advances in Science, Technology and Engineering Systems Journal 6, no. 1 (February 2021): 757–64. http://dx.doi.org/10.25046/aj060183.

Full text

APA, Harvard, Vancouver, ISO, and other styles

35

Schulze, Sören, Johannes Leuschner, and Emily J. King. "Blind Source Separation in Polyphonic Music Recordings Using Deep Neural Networks Trained via Policy Gradients." Signals 2, no. 4 (October 7, 2021): 637–61. http://dx.doi.org/10.3390/signals2040039.

Full text

Abstract:

We propose a method for the blind separation of sounds of musical instruments in audio signals. We describe the individual tones via a parametric model, training a dictionary to capture the relative amplitudes of the harmonics. The model parameters are predicted via a U-Net, which is a type of deep neural network. The network is trained without ground truth information, based on the difference between the model prediction and the individual time frames of the short-time Fourier transform. Since some of the model parameters do not yield a useful backpropagation gradient, we model them stochastically and employ the policy gradient instead. To provide phase information and account for inaccuracies in the dictionary-based representation, we also let the network output a direct prediction, which we then use to resynthesize the audio signals for the individual instruments. Due to the flexibility of the neural network, inharmonicity can be incorporated seamlessly and no preprocessing of the input spectra is required. Our algorithm yields high-quality separation results with particularly low interference on a variety of different audio samples, both acoustic and synthetic, provided that the sample contains enough data for the training and that the spectral characteristics of the musical instruments are sufficiently stable to be approximated by the dictionary.

APA, Harvard, Vancouver, ISO, and other styles

36

Koerselman, Willem, Domien Claessens, Paul ten Den, and Erik van Winden. "Dynamic hydrochemical and vegetation gradients in fens." Wetlands Ecology and Management 1, no. 2 (March 1990): 73–84. http://dx.doi.org/10.1007/bf00177282.

Full text

APA, Harvard, Vancouver, ISO, and other styles

37

Gyuris, E. "Factors that control the emergence of green turtle hatchlings from the nest." Wildlife Research 20, no. 3 (1993): 345. http://dx.doi.org/10.1071/wr9930345.

Full text

Abstract:

Sea turtle hatchlings typically emerge at night and temperature has often been suggested as controlling the timing of emergence. Models based on threshold temperatures, above which hatchling activity is inhibited, are not fully compatible with the observed temporal distribution of emergence. This study reports on the temporal pattern of hatchling emergence and associated sand temperature profiles and tidal regimes at Heron I., in the southern Great Barrier Reef, Australia. A new model is proposed on the basis of sand temperature gradients instead of the absolute temperature of the sand surface. The excellent fit between the temperature gradient model and the observed pattern of emergence suggests that it is the temperature gradient in the top 10cm of the sand column rather than an absolute temperature per se that turtle hatchlings use to time their emergence.

APA, Harvard, Vancouver, ISO, and other styles

38

Cai, Han, Kun Ma, and Yunjian Luo. "Geographical Modeling of Spatial Interaction between Built-Up Land Sprawl and Cultivated Landscape Eco-Security under Urbanization Gradient." Sustainability 11, no. 19 (October 5, 2019): 5513. http://dx.doi.org/10.3390/su11195513.

Full text

Abstract:

Built-up land sprawl is the most intuitive manifestation and main cause of cultivated land pattern change. Understanding spatiotemporal change and its driving factors in the impact of built-up land sprawl on cultivated landscape eco-security under the urbanization gradient can reveal the nature of cultivated landscape ecological risk caused by urban sprawl and promote the scientific implementation of urban planning and cultivated land protection policies. Taking Yangzhou, a medium-sized city with rapid urbanization, as an example, we used multi-sources data (30 m resolution Landsat images in 1998, 2008, and 2018) to analyze the sprawl intensity and direction of built-up land and cultivated landscape ecological risk areas under the urbanization gradient and the spatiotemporal characteristics of the relationship between them. Further, we employed the boosted regression tree (BRT) model to quantify the effects of geographic, social, and economic factors on the impact of built-up land sprawl on cultivated landscape eco-security under different urbanization gradients. Results showed that built-up land sprawl rate increased first and then decreased from 1998 to 2018, forming the sprawl gradients of no. 1–8, no. 8–16, and no. 16–30 urban circle. Its sprawl direction is mainly 20°–90°, 190°–240°, and 305°–355°, whereas cultivated landscape ecological risk area is a continuous sprawl, and the sprawl gradient and direction of the main risk area are consistent with the built-up land. In 1998–2008, the main areas where built-up land sprawl affects cultivated landscape eco-security are located in no. 1–16 urban circle, 20°–55°. From 2008 to 2018, it was located in no. 16–26 urban circle, 305°–355°. In this process, the influence of geospatial factors decreases gradually over time but increases outwardly with the urbanization gradient. The influence of social factors decreases first with time and then increases with the urbanization gradient outward in space. Economic factors increase with time but decrease with the urbanization gradient. The influence of each factor varies greatly over time and with the urbanization gradient.

APA, Harvard, Vancouver, ISO, and other styles

39

Cascone, Valeria, Ilaria Barone, and Jacopo Boaga. "Velocity gradients choice affecting seismic site response in deep alluvial basins: Application to the Venetian Plain (Northern Italy)." Journal of Geophysics and Engineering 19, no. 1 (January 25, 2022): 1–13. http://dx.doi.org/10.1093/jge/gxab067.

Full text

Abstract:

Abstract The average shear-wave velocity of the first 30 metres of subsoil and the depth of the engineering bedrock are considered the key parameters for simplified seismic site response modelling. However, a reliable estimate of the site amplification should consider the entire shear-wave velocity profile from the ground surface down to the engineering bedrock. In deep alluvial basins, a typical geological context where the soil–bedrock interface may lie below the penetration depth of most common prospecting methods, only the shallow velocity profile can be defined in detail, while the deeper structures are commonly extrapolated with linear equations. The choice of a realistic interpolation between the shallow and deep soil still remains an open issue. We compute the 1D seismic site response of two sectors of the Venetian Plain (Northern Italy) characterised by gravelly and sandy deep formations. We model the 1D soil columns using theoretical non-linear gradients proposed in literature for deep alluvial basins. The numerical modelling results, in terms of strong motion parameters, show variations in the seismic site response up to 20%. The effect of the velocity gradients is also evaluated comparing the numerical simulations with real accelerometers recorded by a deep borehole seismometer and a seismic station located at the top of the borehole. These results demonstrate that the selection of the velocity gradient is crucial for seismic site characterisation of deep alluvial basins. In particular, the study suggests which is the most conservative gradient among the ones tested in terms of ground motion hazard estimation.

APA, Harvard, Vancouver, ISO, and other styles

40

Bose, Arnab, Aditya Ramji, Jarnail Singh, and Dhairya Dholakia. "A case study for sustainable development action using financial gradients." Energy Policy 47 (June 2012): 79–86. http://dx.doi.org/10.1016/j.enpol.2012.03.038.

Full text

APA, Harvard, Vancouver, ISO, and other styles

41

Sudhakar, G., B. Jyothi, and V. Venkateswarlu. "Role of diatoms as indicators of pollution gradients." Environmental Monitoring and Assessment 33, no. 2 (November 1994): 85–99. http://dx.doi.org/10.1007/bf00548591.

Full text

APA, Harvard, Vancouver, ISO, and other styles

42

Park, Jeiyoon, Chanhee Lee, Chanjun Park, Kuekyeng Kim, and Heuiseok Lim. "Variational Reward Estimator Bottleneck: Towards Robust Reward Estimator for Multidomain Task-Oriented Dialogue." Applied Sciences 11, no. 14 (July 19, 2021): 6624. http://dx.doi.org/10.3390/app11146624.

Full text

Abstract:

Despite its significant effectiveness in adversarial training approaches to multidomain task-oriented dialogue systems, adversarial inverse reinforcement learning of the dialogue policy frequently fails to balance the performance of the reward estimator and policy generator. During the optimization process, the reward estimator frequently overwhelms the policy generator, resulting in excessively uninformative gradients. We propose the variational reward estimator bottleneck (VRB), which is a novel and effective regularization strategy that aims to constrain unproductive information flows between inputs and the reward estimator. The VRB focuses on capturing discriminative features by exploiting information bottleneck on mutual information. Quantitative analysis on a multidomain task-oriented dialogue dataset demonstrates that the VRB significantly outperforms previous studies.

APA, Harvard, Vancouver, ISO, and other styles

43

Cantinotti, Massimiliano, Pietro Marchese, Marco Scalese, Eliana Franchi, Nadia Assanta, Martin Koestenberger, Jef Van den Eynde, Shelby Kutty, and Raffaele Giordano. "Normal Values and Patterns of Normality and Physiological Variability of Mitral and Tricuspid Inflow Pulsed Doppler in Healthy Children." Healthcare 10, no. 2 (February 11, 2022): 355. http://dx.doi.org/10.3390/healthcare10020355.

Full text

Abstract:

Background: While mitral (MV) and tricuspid valve (TV) pulsed Doppler velocities and derived gradients are commonly evaluated, data on normal pediatric values are limited. This study aimed to evaluate the normal values and physiological variability for MV and TV Doppler velocities and derived gradients in a large cohort of prospectively enrolled healthy children. Methods: The echocardiographic measurements included pulsed Doppler MV and TV E and A velocities, E deceleration times (EDT), maximal and mean gradients, and velocity time integral (VTI). Results: A total of 544 healthy subjects (median age 6.4 years, range 1 day–17.68 years) were included. MV and TV E velocity, E/A ratio, and E and A wave duration increased, while A velocity decreased with age (p < 0.001). Along with an increase in VTI, there occurred a progressive increase in maximum velocity and gradients and a decrease in mean velocities and gradients. E/A inversions were common, especially at the TV in neonates and infants. For MV, inversion in either one, two, or three consecutive beats occurred in 51.9% of neonates and 18.3% of infants, while it was rare at older ages (all p < 0.001). For TV, inversions in three consecutive beats occurred in 71.4% of neonates, while inversions in only one or two beats were more common in infants (27.3%). For TV, inversion in one or more beats, however, was not infrequent at all ages. Conclusions: We report normal values and patterns of normality and physiological variability for MV and TV inflow Doppler from a large population of healthy children.

APA, Harvard, Vancouver, ISO, and other styles

44

Siddiqi, Arjumand, Ichiro Kawachi, Lisa Berkman, S. V. Subramanian, and Clyde Hertzman. "Variation of Socioeconomic Gradients in Children's Developmental Health across Advanced Capitalist Societies: Analysis of 22 Oecd Nations." International Journal of Health Services 37, no. 1 (January 2007): 63–87. http://dx.doi.org/10.2190/ju86-457p-7656-w4w7.

Full text

APA, Harvard, Vancouver, ISO, and other styles

45

Ji, Chaonan, Uta Heiden, Tobia Lakes, and Hannes Feilhauer. "Are urban material gradients transferable between areas?" International Journal of Applied Earth Observation and Geoinformation 100 (August 2021): 102332. http://dx.doi.org/10.1016/j.jag.2021.102332.

Full text

APA, Harvard, Vancouver, ISO, and other styles

46

Sorace, Alberto, and Marco Gustin. "Distribution of generalist and specialist predators along urban gradients." Landscape and Urban Planning 90, no. 3-4 (April 2009): 111–18. http://dx.doi.org/10.1016/j.landurbplan.2008.10.019.

Full text

APA, Harvard, Vancouver, ISO, and other styles

47

Ma, Lin, Yueyao Wang, Ze Liang, Jiaqi Ding, Jiashu Shen, Feili Wei, and Shuangcheng Li. "Changing Effect of Urban Form on the Seasonal and Diurnal Variations of Surface Urban Heat Island Intensities (SUHIIs) in More Than 3000 Cities in China." Sustainability 13, no. 5 (March 7, 2021): 2877. http://dx.doi.org/10.3390/su13052877.

Full text

Abstract:

Based on the indicators of more than 3000 cities in China, this study shows that the relationship between the urban form and surface urban heat island intensity (SUHII) demonstrates seasonal and diurnal variations, and also changes along urban development and elevation gradients. SUHIIs show seasonal and diurnal change patterns along urban development and elevation gradients, but there is no obvious change trend along temperature and humidity gradients. Among them, the seasonal variation of the SUHII went up about 0.4 ℃ from the first level of urban development to the highest level, while the diurnal variation of the SUHII decreased by 0.4 °C. With urban development, the correlations between the anthropogenic heat flux (AHF), population density (POPDEN) and morphological continuity (CONTIG) with the SUHII of summer days, summer nights and winter nights continued to be enhanced, with the correlation coefficients (β) increased by about 0.3. The effect of area size (AREA) became more influential on the SUHII of summer days and nights, but its effect on the SUHII of winter nights increased first and then decreased along the urban development gradient. With the increase of elevation, the correlations of the AHF, POPDEN, AREA, CONTIG and summer day and night SUHII were gradually reduced (β decreased by about 0.4), but their impact on the SUHII of winter nights was gradually enhanced (β increased by about 0.2 to 0.3). Along temperature and humidity gradients, the positive effect of POPDEN on the summer SUHII decreased gradually (β decreased by about 0.3). However, the enhancement effects of the AHF, AREA, CONTIG and POPDEN on the SUHII of winter nights increased generally (β increased by about 0.2). According to the Random Forest model, for the SUHIIs at night, the relative importance (RI) of urban form factors was greater, while for the SUHIIs of daytime, the RIs of natural factors were greater. The contribution of the urban form to the seasonal variation of the SUHII is similar to that of natural factors, but their contribution to diurnal variation is lower. Our results suggest that it is more necessary to control the urban scale, avoid excessive urban agglomeration and reasonably control the anthropogenic heat emission in more developed and low altitude cities to reduce their summer heat exposure.

APA, Harvard, Vancouver, ISO, and other styles

48

Nahry and Noor Syiffa Fadillah. "The Empirical Study on the Impact of Road Gradient and Truck Composition on the Toll Road Traffic Performance." E3S Web of Conferences 65 (2018): 09003. http://dx.doi.org/10.1051/e3sconf/20186509003.

Full text

Abstract:

The Jakarta Outer Ring Road (JORR) is a toll road system that circles the outskirts of Jakarta, where the purpose of this road is to reduce congestion on the street network of the city centre of Jakarta. However, the high composition of trucks in JORR resulted in congestion, and moreover it is suspected that the steep gradient of this road has contributed to this situation. This study aims to investigate the impact of road gradient on the trucks and the overall traffic performance from macroscopic view point. This study is the complement to a similar study that has been done in JORR previously, to confirm the result. The previous study was based on the Vissim simulation, while this paper will be entirely focused on an empirical study. Using the data obtained from 24-hour traffic recording on selected JORR sections that represent various gradients, traffic flow models are generated, and the effect of gradient could be assessed. The speed difference between trucks and non-trucks shows 56.2%~127% gap as the result of different gradient occurred. The overall free flow speed (uf) difference of different gradient is up to 27.3%. This result will benefit to the transport authority to justify the policy concerning the existence of trucks in the traffic flow, particularly in the condition at which trucks are regarded as a culprit of traffic congestion.

APA, Harvard, Vancouver, ISO, and other styles

49

Liu, Guoqing, Li Zhao, Feidiao Yang, Jiang Bian, Tao Qin, Nenghai Yu, and Tie-Yan Liu. "Trust Region Evolution Strategies." Proceedings of the AAAI Conference on Artificial Intelligence 33 (July 17, 2019): 4352–59. http://dx.doi.org/10.1609/aaai.v33i01.33014352.

Full text

Abstract:

Evolution Strategies (ES), a class of black-box optimization algorithms, has recently been demonstrated to be a viable alternative to popular MDP-based RL techniques such as Qlearning and Policy Gradients. ES achieves fairly good performance on challenging reinforcement learning problems and is easier to scale in a distributed setting. However, standard ES algorithms perform one gradient update per data sample, which is not very efficient. In this paper, with the purpose of more efficient using of sampled data, we propose a novel iterative procedure that optimizes a surrogate objective function, enabling to reuse data sample for multiple epochs of updates. We prove monotonic improvement guarantee for such procedure. By making several approximations to the theoretically-justified procedure, we further develop a practical algorithm called Trust Region Evolution Strategies (TRES). Our experiments demonstrate the effectiveness of TRES on a range of popular MuJoCo locomotion tasks in the OpenAI Gym, achieving better performance than ES algorithm.

APA, Harvard, Vancouver, ISO, and other styles

50

Hamin, Elisabeth, Yaser Abunnasr, Max Roman Dilthey, Pamela Judge, Melissa Kenney, Paul Kirshen, Thomas Sheahan, et al. "Pathways to Coastal Resiliency: The Adaptive Gradients Framework." Sustainability 10, no. 8 (July 26, 2018): 2629. http://dx.doi.org/10.3390/su10082629.

Full text

Abstract:

Current and future climate-related coastal impacts such as catastrophic and repetitive flooding, hurricane intensity, and sea level rise necessitate a new approach to developing and managing coastal infrastructure. Traditional “hard” or “grey” engineering solutions are proving both expensive and inflexible in the face of a rapidly changing coastal environment. Hybrid solutions that incorporate natural, nature-based, structural, and non-structural features may better achieve a broad set of goals such as ecological enhancement, long-term adaptation, and social benefits, but broad consideration and uptake of these approaches has been slow. One barrier to the widespread implementation of hybrid solutions is the lack of a relatively quick but holistic evaluation framework that places these broader environmental and societal goals on equal footing with the more traditional goal of exposure reduction. To respond to this need, the Adaptive Gradients Framework was developed and pilot-tested as a qualitative, flexible, and collaborative process guide for organizations to understand, evaluate, and potentially select more diverse kinds of infrastructural responses. These responses would ideally include natural, nature-based, and regulatory/cultural approaches, as well as hybrid designs combining multiple approaches. It enables rapid expert review of project designs based on eight metrics called “gradients”, which include exposure reduction, cost efficiency, institutional capacity, ecological enhancement, adaptation over time, greenhouse gas reduction, participatory process, and social benefits. The framework was conceptualized and developed in three phases: relevant factors and barriers were collected from practitioners and experts by survey; these factors were ranked by importance and used to develop the initial framework; several case studies were iteratively evaluated using this technique; and the framework was finalized for implementation. The article presents the framework and a pilot test of its application, along with resources that would enable wider application of the framework by practitioners and theorists.

APA, Harvard, Vancouver, ISO, and other styles

Journal articles on the topic 'Policy gradients'

Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles