China
Africa
Explore chapters and articles related to this topic
AI Technology in Networks-on-Chip
Published in Om Prakash Jena, Sudhansu Shekhar Patra, Mrutyunjaya Panda, Zdzislaw Polkowski, S. Balamurugan, Industrial Transformation, 2022
Hu et al. (2020) use graph rewriting for adjustment of application-specific NoC where the Design Space Exploration (DSE) is as a Markov Decision Process (MDP). It uses Monte Carlo Tree Search (MCTS), an reinforcement learning technique as a search heuristic, and it shows better efficiency compared to Simulated Annealing (SA) and Genetic Annealing (GA). The authors proposed an RNN-based model for latency estimation. It is faster compared to cycle-accurate SystemC simulations and offers similar speedup as queuing theory (QT). Compared to SystemC simulation (SC), RNN provides a speedup of 5.85× to 7.38×, and QT provides a speedup of 6.78× to 8.21×.
Offshore responses using an efficient time simulation regression procedure
Published in J. Parunov, C. Guedes Soares, Trends in the Analysis and Design of Marine Structures, 2019
S.Z.A. Syed Ahmad, M.K. Abu Husain, N.I. Mohd Zaki, M.H. Mohd, G. Najafian
To conclude, Figure 4 shows large sampling variability, while Figure 5 displays a small sampling variability. It indicates the greater number of simulated records, the less will be the sampling variability error. That is the reason why MCTS method requires a huge number huge of simulated response records only to reduce its sampling variability in obtaining reliable results.
An improved Monte Carlo Tree Search approach to workflow scheduling
Published in Connection Science, 2022
Hok-Leung Kung, Shu-Jun Yang, Kuo-Chan Huang
Monte Carlo Tree Search (MCTS) is a heuristic searching algorithm consisting of four major steps. The entire search process of MCTS can be illustrated in Figure 8. When we apply MCST to workflow scheduling, each node in the MCTS tree represents a probable partial or complete schedule of a workflow to be executed on a set of processors. Nodes with complete schedules are called terminal nodes. The children of a node represent all possibilities of choosing an unscheduled ready task and allocating it to one of the processors. The root is a special node with an empty schedule where no tasks have been scheduled. MCTS starts with a single root node, and the tree gradually grows as illustrated in Figure 8 as the search process proceeds.
Extreme response prediction for fixed offshore structures by efficient time simulation regression procedures. Part 2: model validation
Published in Ships and Offshore Structures, 2023
S. Z. A. Syed Ahmad, M. K. Abu Husain, N. I. Mohd Zaki, N. A. Mukhlas, G. Najafian
Time domain analysis based on numerical schemes is commonly used to analyse the non-linear structural responses. The difficulty based on time-domain simulation may become computationally demanding because it considers all sort of nonlinearities in the calculation. Moreover, using the conventional MCTS method leads to increasing the simulation time and simulation number to achieve the accuracy level. Therefore, the novel design via the ETS-RegSE model is presented to assess offshore structures’ responses accurately and efficiently. The summary, which estimates the probability of the extreme offshore responses can be related to Figure 3.
An improved relief feature selection algorithm based on Monte-Carlo tree search
Published in Systems Science & Control Engineering, 2019
Jianyang Zheng, Hexing Zhu, Fangfang Chang, Yunlong Liu
Monte Carlo Tree Search (MCTS) is a method for finding the optimal decisions in a given domain by taking random samples in the decision space and building a search tree according to the results of samples. The basic process of MCTS is to construct the search tree iteratively until the pre-defined termination condition is reached (Liu et al., 2016). The node in the search tree represents the state of the domain, and the directed links from a node to its child nodes represent actions leading to subsequent states (Browne et al., 2012). Four phases exist in each search iteration (Hunt, Marin, & Stone, 1966; Chaslot, Bakkes, Szita, & Spronck, 2008).