scholarly journals AIBPO: Combine the Intrinsic Reward and Auxiliary Task for 3D Strategy Game

Complexity ◽  
2021 ◽  
Vol 2021 ◽  
pp. 1-9
Author(s):  
Huale Li ◽  
Rui Cao ◽  
Xuan Wang ◽  
Xiaohan Hou ◽  
Tao Qian ◽  
...  

In recent years, deep reinforcement learning (DRL) achieves great success in many fields, especially in the field of games, such as AlphaGo, AlphaZero, and AlphaStar. However, due to the reward sparsity problem, the traditional DRL-based method shows limited performance in 3D games, which contain much higher dimension of state space. To solve this problem, in this paper, we propose an intrinsic-based policy optimization (IBPO) algorithm for reward sparsity. In the IBPO, a novel intrinsic reward is integrated into the value network, which provides an additional reward in the environment with sparse reward, so as to accelerate the training. Besides, to deal with the problem of value estimation bias, we further design three types of auxiliary tasks, which can evaluate the state value and the action more accurately in 3D scenes. Finally, a framework of auxiliary intrinsic-based policy optimization (AIBPO) is proposed, which improves the performance of the IBPO. The experimental results show that the method is able to deal with the reward sparsity problem effectively. Therefore, the proposed method may be applied to real-world scenarios, such as 3-dimensional navigation and automatic driving, which can improve the sample utilization to reduce the cost of interactive sample collected by the real equipment.

2014 ◽  
Vol 644-650 ◽  
pp. 381-384
Author(s):  
Xin Zhang ◽  
Hao Zhou ◽  
Guo Song Liu

In order to improve the efficiency of auto parts distribution logistics, to lower the cost of auto production in transportation logistics, and to reduce accidents, in this paper it is designed that an automatic guided vehicle control system to replace the manned tractors in the distribution sites. The system is equipped with an infrared homing device that can ensure the automated guided vehicle (AGV) along a predetermined route automatic driving at a given distribution information, without the needs to manually guided. Test results show that the circuit performance of AGV control system is stable to ensure the accuracy of the tracking in the practical application, and the mean absolute error of the tracking is less than 0.04m.


Geophysics ◽  
2015 ◽  
Vol 80 (2) ◽  
pp. WA99-WA111 ◽  
Author(s):  
Anya Reitz ◽  
Richard Krahenbuhl ◽  
Yaoguo Li

There is presently an increased need to monitor production efficiency as heavy oil reservoirs become more economically viable. We present a feasibility study of monitoring steam-assisted gravity drainage (SAGD) reservoirs using time-lapse gravimetry and gravity gradiometry. Even though time-lapse seismic has historically shown great success for SAGD monitoring, the gravimetry and gravity gradiometry methods offer a low-cost interseismic alternative that can complement the seismic method, increase the survey frequency, and decrease the cost of monitoring. In addition, both gravity-based methods are directly sensitive to the density changes that occur as a result of the replacement of heavy oil by steam. Advances in technologies have made both methods viable candidates for consideration in time-lapse reservoir monitoring, and we have numerically evaluated their potential application in monitoring SAGD production. The results indicate that SAGD production should produce a strong anomaly for both methods at typical SAGD reservoir depths. However, the level of detail for steam-chamber geometries and separations that can be recovered from the gravimetry and gravity gradiometry data is site dependent. Gravity gradiometry shows improved monitoring ability, such as better recovery of nonuniform steam movement due to reservoir heterogeneity, at shallower production reservoirs. Gravimetry has the ability to detect SAGD steam-chamber growth to greater depths than does gravity gradiometry, although with decreasing resolution of the expanding steam chambers.


2014 ◽  
Vol 575 ◽  
pp. 910-916
Author(s):  
Samir Mekid

This paper reports on the impact of Value Analysis Value Engineering (VAVE) applied to an industrial product where manufacturability and choice of manufacturing methods can be of great benefit in reducing cost and sometimes manufacturing time in a mass production line. The purpose of this requested task is to reduce the cost of a specific product using VAVE analysis through better value estimation related to suitable manufacturing process while maintaining similar or better technical performance. A couple of technical design solutions are proposed and discussed with study cases. The work is progressing towards the final stage.


2008 ◽  
Vol 12 (1) ◽  
pp. 25-32
Author(s):  
Jose Villacis Gonzalez

The Rubiks cube is a special game and a very particular puzzle. The 3-dimensional cube is made up of six faces, or boundary sections, of the same size. Each face, or section, consists of several two dimensional square parts, or cubelets. Every cubelet has the same surface area, and each of the six faces has the same number of cubelets. Therefore, the cubes surface is entirely covered with isocubelets. The cubelets are painted in six different colours, and it is possible to create a design where each face shows only one colour. Such is the object of the game: to turn the cubelets and sections of the cube so that only one (different) colour shows on each one of the six faces. If one manages to master the puzzle, the cube will show six faces of the same size, each coloured differently. The cubelets and sections of the cube can be turned both horizontally and vertically in order to change colours while trying to determine the appropriate combination to complete the puzzle. This approach is linked to a particular function in microeconomics that deals with the relationship between two magnitudes: on the one hand, the moves needed to achieve the desired final design; and on the other hand, the cost linked to the required production processes. This analytical model must use combinatorial mathematics equipment because, after all, the key factor in solving the Rubiks cube is the way in which the cubelets and sections are arranged.


2021 ◽  
pp. 257-264
Author(s):  
Huale Li ◽  
Rui Cao ◽  
Xiaohan Hou ◽  
Xuan Wang ◽  
Linlin Tang ◽  
...  

Tumor Biology ◽  
2019 ◽  
Vol 41 (9) ◽  
pp. 101042831987374
Author(s):  
Stewart Sell ◽  
Zoran Ilic

Differentiation therapy is directed to the self-renewing cancer stem cells, as well as their progeny transit amplifying cells, to force them to mature to terminal differentiation. Differentiation therapy is effective in treatment of neuroblastomas and myeloid leukemias. Checkpoint inhibition therapy removes blocks to cancer reactive T-killer cells and allows them to react to malignant cells and limit the growth of cancer. The percentage of patients with a given cancer that responds to either therapy is less than hoped for, and the duration of response is variable. Multiplying the response rate (percentage of patients responding to therapy) by the duration of response may be used to derive a survival score for patients treated with differentiation therapy or checkpoint inhibition. By this criterion, differentiation therapy gives better survival scores than checkpoint inhibition. Yet, checkpoint inhibition is considered a great success, mostly because it may be applied to many different types of cancer, and differentiation therapy is considered relatively ineffective because it is limited to a few specific cancers. On the other hand, the cost of checkpoint inhibition treatment is 10–20 times more per patient than that of differentiation therapy. Hopefully, future combined treatments and advances in both approaches will increase the effectiveness of these cancer treatments.


2021 ◽  
Author(s):  
Stephanie Eugenie Brigitte McArdle ◽  
Kinana Habra ◽  
Joshua R D Pearson

Monolayer cell culture, while useful for basic in vitro studies, are not physiologically relevant. Spheroids, on the other hand provide a more complex 3-dimensional (3D) structure which more resemble the in vivo tumour growth thereby allowing results obtained with those on proliferation, cell death, differentiation, metabolism, and various anti-tumour therapies to be more predictive of in vivo outcomes. However, the cost associated with their generation often involve expensive, plate, media, and growth supplements, which have limited their use for high throughput experiments. The protocol herein presents a novel and rapid generation for single spheroids of various cancer cell lines, U87 MG; SEBTA-027; SF188, brain cancer cells, DU-145, TRAMP-C1, prostate cancer cells, in 96-round bottom well plates. Cells are washed with anti-adherent solution, and the homogeneous compact spheroid morphology was evidenced as early as 24 hours after 10 minutes centrifugation for the seeded cells. By using confocal microscopy, the proliferating cells were traced in the rim and the dead cells were found inside the core region of the spheroid. The H&E stain of spheroid slices and the western blotting were utilised to investigate the tightness of the cell packaging by adhesion proteins. Carnosine was used as an example of treatment for U87 single spheroids. The protocol allows the rapid generation of spheroids, which will help towards reducing the number of tests performed on animals.


Author(s):  
O. N. Kalinina

The article shows results of researching the projects that exist in St. Petersburg. Constructive and cost parameters of power stations, which pass and distribute electric power were found and possibility to estimate the degree of parametric similarity of the project being devised with the analogues project by using the method of cluster analysis by homogeneity principle. It was found out that the procedure of clustering gives an opportunity to get homogeneous in quality samples of needed projects, which in its turn provides a chance to choose the nearest analogue and on its foundation to forecast the cost of the project. On the basis of the method of expert estimations and respective coefficients methodology of modeling the project cost was designed, whose goal is to evaluate the precise cost of the project, which differs from the analogue by parameters. This methodology can allow us with more trustworthiness to identify the future cot of the project and to give possibility to change the cost. The approach put forward by the authors is built on the basis of mathematic model data that speed up the process of selecting analogues and modeling cost of the project.


Author(s):  
Sheng-Jun Huang ◽  
Nengneng Gao ◽  
Songcan Chen

Multi-instance multi-label learning(MIML) has been successfully applied into many real-world applications. Along with the enhancing of the expressive power, the cost of labelling a MIML example increases significantly. And thus it becomes an important task to train an effective MIML model with as few labelled examples as possible. Active learning, which actively selects the most valuable data to query their labels, is a main approach to reducing labeling cost. Existing active methods achieved great success in traditional learning tasks, but cannot be directly applied to MIML problems. In this paper, we propose a MIML active learning algorithm, which exploits diversity and uncertainty in both the input and output space to query the most valuable information. This algorithm designs a novel query strategy for MIML objects specifically and acquires more precise information from the oracle without addition cost. Based on the queried information, the MIML model is then effectively trained by simultaneously optimizing the relative rank among instances and labels.


Author(s):  
Adam Bodayuk

The subject of the study is the value of geological objects investigated for the needs of defining a number of concepts related to economic and fiscal geology. The purpose of this article is to substantiate the nature of resource-resource fiscalization of subsoil use and methods of estimated calculation of its indicators. Methodology of work – the method of monographic analysis (in determining the cost-resource approach); method of abstract-logical analysis (in defining the system of concepts of economic and fiscal geology); method of generalization (when forming conclusions and suggestions). Methodology of work – method of abstract-logical analysis (in defining the system of concepts of fiscal geology, in particular, fiscal subsoil-resource product); monographic analysis (when using the resource-cost approach), volumetric- geochemical method (to quantify the estimated and prospective resources by the results of geochemical studies); comparison and generalization (when forming conclusions and proposals). Results of work. According to the results of the conducted researches of the essence, content, processes of subsoil use and methods of determining its cost indicators, according to literature sources and official documents, certain fiscal concepts of fiscal geology are substantiated. The essence of resource-based fiscalisation in subsoil use is defined as the conduct of certain studies of subsoil objects, the allocation of subsoil products, the determination of their resource indicators, and the establishment using official and non-official methods of value of these objects.The value-for-money approach to the evaluation of the subsoil resource product (required) is substantiated. This approach first distinguishes the explorable geological object, further distinguishes it from subsoil resources, then provides these products with specific methods of value estimation. These estimates determine the expected sale price of the property, the cost of renting the property, permanent ownership, permanent use, etc., as well as payments to the state treasury. Conclusions and suggestions. Cost and resource fiscalisation is interpreted as the determination of the fiscal value of subsoil resources products, on the basis of which the fiscal payments are calculated. The methodologies for calculating the cost of mining are necessary with certain additions and practical calculations combined into a recommendation document, discuss it, check it in practice, make certain adjustments and officially approve it for use in determining the fiscal value. Keywords: subsoil, minerals, resources, costs, expenses, extraction, payment.


Sign in / Sign up

Export Citation Format

Share Document