Open Dynamics Engine を用いたスノーボードロボットシミュレータの開発

Exploration vs. exploitation is a well known issue in Evolutionary Algorithms. Accordingly, an unbalanced search can lead to premature convergence. GASOM, a novel Genetic Algorithm, addresses this problem by intelligent exploration techniques. The approach uses Self-Organizing Maps to mine data from the evolution process. The information obtained is successfully utilized to enhance the search strategy and confront genetic drift. This way, local optima are avoided and exploratory power is maintained. The evaluation of GASOM on well known problems shows that it effectively prevents premature convergence and seeks the global optimum. Particularly in deceptive and missleading functions it showed outstanding performance. Additionally, representing the search history by the Self-Organizing Map provides a visually pleasing insight into the state and course of evolution.

Intelligent Exploration for Genetic Algorithms: Using Self-Organizing Maps in Evolutionary Computation

This paper presents a new fuzzy bee colony optimization method to find the optimal distribution of the membership functions in the design of fuzzy controllers for complex nonlinear plants. We used interval type-2 and type-1 fuzzy logic systems in dynamically adapting the alpha and beta parameter values of the bee colony optimization algorithm (BCO). Simulation results with a type-1 fuzzy logic controller for benchmark control plants are presented. The advantage of using interval type-2 fuzzy logic systems for dynamic adjustment of parameters in BCO applied in fuzzy controller design is verified with two benchmark problems. We considered different levels and types of noise in the simulations to analyze the approach of interval type-2 fuzzy logic systems to find the best values of alpha and beta for BCO when applied in the design of fuzzy controllers.

A new fuzzy bee colony optimization with dynamic adaptation of parameters using interval type-2 fuzzy logic for tuning fuzzy controllers

In recent years, voltage limit violation and power system load-generation imbalance, i.e., line loading limit violation have been responsible for several incidents of major network collapses leading to partial or even complete blackouts. Alleviation of line overloads is the suitable corrective action in this regard. The control action strategies to limit the line loading to the security limits are generation rescheduling/load shedding. In this paper, an approach based on radial basis function neural network (RBFN) is presented for corrective action planning to alleviate line overloading in an efficient manner. Effectiveness of the proposed method is demonstrated for overloading alleviation under different loading/contingency conditions in 6-bus system and 24-bus RTS system.

Corrective action planning using RBF neural network

In this paper, we propose a method called Convolutional Neural Network-Markov Random Field (CNN-MRF) to estimate the crowd count in a still image. We first divide the dense crowd visible image into overlapping patches and then use a deep convolutional neural network to extract features from each patch image, followed by a fully connected neural network to regress the local patch crowd count. Since the local patches have overlapping portions, the crowd count of the adjacent patches has a high correlation. We use this correlation and the Markov random field to smooth the counting results of the local patches. Experiments show that our approach significantly outperforms the state-of-the-art methods on UCF and Shanghaitech crowd counting datasets.

Image Crowd Counting Using Convolutional Neural Network and Markov Random Field

A co-design architecture for the field programmable gate array (FPGA)-based classification system is proposed. In this architecture, a neural network (NN) is trained by the particle swarm optimization algorithm (PSO). The NN is implemented in hardware, and the PSO training is executed by a processor. The ARM processor and the NIOS II processor are used. This architecture maintains the operating speed of the NN. This approach also has the flexibility to change the parameters or even the PSO algorithm without affecting the FPGA part. Three publicly recognized databases (Iris, Balance-scale, Credit approval) are used to evaluate the system. Experimental results showed that the co-design was successfully implemented. In addition, compared to the NIOS II approach, the ARM approach had the advantages regarding the operating speed and the logic utilization.

Hardware/Software Co-design for a Neural Network Trained by Particle Swarm Optimization Algorithm

In this research, a fuzzy controller system is useful as a key technique to design a machine controller because a fuzzy system can be designed by fuzzy sets, which can be represented using linguistic descriptions of natural language. Those natural languages are able capture human knowledge in fuzzy rules. A fuzzy controller system consists of fuzzy rules and involved techniques. Recently, the research of the fuzzy control has controlled various machines. We verified that fuzzy control can be applicable to a line tracing car control, which is running at a high speed. In our research, designers took a huge amount of time to adjust and setup the fuzzy controller. In order to apply the Neural Network (NN) controller, we designed a simple network system, and setup a PSO_VC and parameters. The PSO_VC (a PSO with Velocity Control) is a speed control strategy for all moving particles on the PSO. In order to configure the NN, a very powerful and quick PSO_VC is used. The PSO_VC is able to adjust all weights and threshold levels. In this paper we present the performances of accurate control and compare results between fuzzy controller and the NN controller.

PSO training of the Neural Network application for a controller of the line tracing car

The adaptive neuro-fuzzy inference system (ANFIS) is employed in a vast range of applications because of its smoothness (by Fuzzy Control (FC)) and adaptability (by Neural Network (NN)). Although ANFIS is better in nonlinear optimization, two major loopholes need to be addressed thoroughly. They are the curse of dimensionality and computational complexity. To overcome these complications, a novel usage of the ANFIS model is proposed as Cascaded ANFIS in this paper. As the primary source of this algorithm, a general two-input one-output ANFIS algorithm is used. The novel algorithm has two main modules called pair selection and training model. Pair selection is responsible for selecting the best match for the inputs, while the training module generates the output. The cascaded behavior of the novel algorithm generates additional iterations to advance the best solution. Even though the number of parameters that need to be adjusted is increasing at each additional iteration, the complexity of the algorithm may stay stable. Two-hybrid state-of-the-art algorithms are used to compare the performance of the novel algorithm, namely, Particle Swarm Optimization-based ANFIS (ANFIS-PSO) and Genetic Algorithm-based ANFIS (ANFIS-GA). Furthermore, individual performance is presented for seven publicly recognized datasets. The results have demonstrated that, for some datasets, the root means square error (RMSE) can be minimum as 0.0001.

A Novel Optimization Algorithm: Cascaded Adaptive Neuro-Fuzzy Inference System

Hydropower stands as a crucial source of power in the current world, and there is a vast range of benefits of forecasting power generation for the future. This paper focuses on the significance of climate change on the future representation of the Samanalawewa Reservoir Hydropower Project using an architecture of the Cascaded ANFIS algorithm. Moreover, we assess the capacity of the novel Cascaded ANFIS algorithm for handling regression problems and compare the results with the state-of-art regression models. The inputs to this system were the rainfall data of selected weather stations inside the catchment. The future rainfalls were generated using Global Climate Models at RCP4.5 and RCP8.5 and corrected for their biases. The Cascaded ANFIS algorithm was selected to handle this regression problem by comparing the best algorithm among the state-of-the-art regression models, such as RNN, LSTM, and GRU. The Cascaded ANFIS could forecast the power generation with a minimum error of 1.01, whereas the second-best algorithm, GRU, scored a 6.5 error rate. The predictions were carried out for the near-future and mid-future and compared against the previous work. The results clearly show the algorithm can predict power generation’s variation with rainfall with a slight error rate. This research can be utilized in numerous areas for hydropower development.

/pdf/a-cascaded-adaptive-network-based-fuzzy-inference-system-for-2h3qz6ev.pdf

Yukinobu Hoshino

Papers

Hardware/Software Co-design for a Neural Network Trained by Particle Swarm Optimization Algorithm

PSO training of the Neural Network application for a controller of the line tracing car

A Novel Optimization Algorithm: Cascaded Adaptive Neuro-Fuzzy Inference System

A Cascaded Adaptive Network-Based Fuzzy Inference System for Hydropower Forecasting

Research of Foresight Knowledge by CMAC based Q-learning in Inhomogeneous Multi-Agent System