Many problems modeled by Markov decision processes (MDPs) have very large state and/or action spaces, leading to the well-known curse of dimensionality that makes solution of the resulting models intractable. In other...
详细信息
Many problems modeled by Markov decision processes (MDPs) have very large state and/or action spaces, leading to the well-known curse of dimensionality that makes solution of the resulting models intractable. In other cases, the system of interest is complex enough that it is not feasible to explicitly specify some of the MDP model parameters, but simulated sample paths can be readily generated (e.g., for random state transitions and rewards), albeit at a non-trivial computational cost. For these settings, we have developed various sampling and population-based numerical algorithms to overcome the computational difficulties of computing an optimal solution in terms of a policy and/or value function. Specific approaches presented in this survey include multi-stage adaptive sampling, evolutionary policy iteration and evolutionary random policy search.
The massive nature of modern university programming courses increases the burden on academic workers. The Digital Teaching Assistant (DTA) system addresses this issue by automating unique programming exercise generati...
详细信息
The massive nature of modern university programming courses increases the burden on academic workers. The Digital Teaching Assistant (DTA) system addresses this issue by automating unique programming exercise generation and checking, and provides means for analyzing programs received from students by the end of semester. In this paper, we propose a machine learning-based approach to the classification of student programs represented as Markov chains. The proposed approach enables real-time student submissions analysis in the DTA system. We compare the performance of different multi-class classification algorithms, such as support vector machine (SVM), the k nearest neighbors (KNN) algorithm, random forest (RF), and extreme learning machine (ELM). ELM is a single-hidden layer feedforward network (SLFN) learning scheme that drastically speeds up the SLFN training process. This is achieved by randomly initializing weights of connections among input and hidden neurons, and explicitly computing weights of connections among hidden and output neurons. The experimental results show that ELM is the most computationally efficient algorithm among the considered ones. In addition, we apply biology-inspired algorithms to ELM input weights fine-tuning in order to further improve the generalization capabilities of this algorithm. The obtained results show that ELMs fine-tuned with biology-inspired algorithms achieve the best accuracy on test data in most of the considered problems.
This paper discusses dual rotor axial flux machines with surface mounted and spoke type ferrite permanent magnets (PMs) with concentrated windings;they are introduced as alternatives to a generator with surface mounte...
详细信息
This paper discusses dual rotor axial flux machines with surface mounted and spoke type ferrite permanent magnets (PMs) with concentrated windings;they are introduced as alternatives to a generator with surface mounted Nd-Fe-B magnets which is analyzed in [10]. The output power, voltage, speed and air gap clearance for all the generators are identical. The machine designs are optimized for minimum mass using a population-based algorithm, assuming the same efficiency as the Nd-Fe-B machine. A finite element analysis (FEA) is applied to predict the performance, electromotive force, developed torque, cogging torque, no load losses, leakage flux and efficiency of both ferrite generators and that of the Nd-Fe-B generator. To minimize cogging torque, different rotor pole topologies and different pole arc to pole pitch ratios are investigated by means of 3D FEA. It was found that the surface mounted Ferrite generator topology is unable to develop the nominal electromagnetic torque, and has higher torque ripple and is heavier than the spoke type machine. Furthermore, it was shown that the spoke type Ferrite PM generator has favorable performance and could be an alternative to rare-earth PM generators, particularly in wind energy applications. Finally, the analytical and numerical results are verified using experimental results.
Differential Evolution (DE) is a population-based algorithm which has been successfully used to solve optimization problems. DE algorithm begins with an initial population with some randomly generated candidate soluti...
详细信息
Differential Evolution (DE) is a population-based algorithm which has been successfully used to solve optimization problems. DE algorithm begins with an initial population with some randomly generated candidate solutions. During evolutionary process, the population of candidate solutions is evolved toward the promising region by using the specific operations. The population in the DE algorithm can resemble an especial perspective of a small society which has individuals to seek a common goal. In a society, the election system is commonly used as an effective approach;which is employed to determine one or several representatives who are responsible to make major decisions. Some machine learning algorithms are inspired from the society election system to develop an enhanced algorithm from a pool of potential algorithms with the complementary performances. This study is motivated from the election systems of societies which can be applied on population-based algorithms, here DE algorithm as a case study. We propose an election-based discrete DE algorithm which uses the information of all candidate solutions to create a new trial solution as a president candidate solution. During optimization phases, after applying the evolutionary operators, all candidate solutions vote to select the values of president's variables. In the proposed method, a majority voting method is applied to choose a value for each variable of the president candidate solution. We employ the discrete DE (DDE) algorithm as the parent algorithm to develop election-based discrete DE (EDDE) which is evaluated on the fifteen discrete benchmark functions. Simulation results confirm that EDDE obtains a promising performance on the majority of these functions.
暂无评论