CMAC Models Learn to Play
نویسنده
چکیده
Traditional reinforcement learning methods require a function approx-imator (FA) for learning value functions in large or continuous state spaces. We describe a novel combination of CMAC-based FAs and adap-tive world models (WMs) estimating transition probabilities and rewards. Simple variants are tested in multiagent soccer environments where they outperform the evolutionary method PIPE which performed best in previous comparisons.
منابع مشابه
Adaptive Noise Cancellation Using Deep Cerebellar Model Articulation Controller
This paper proposes a deep cerebellar model articulation controller (DCMAC) for adaptive noise cancellation (ANC). We expand upon the conventional CMAC by stacking single-layer CMAC models into multiple layers to form a DCMAC model and derive a modified backpropagation training algorithm to learn the DCMAC parameters. Compared with conventional CMAC, the DCMAC can characterize nonlinear transfo...
متن کاملTwo Novel Learning Algorithms for CMAC Neural Network Based on Changeable Learning Rate
Cerebellar Model Articulation Controller Neural Network is a computational model of cerebellum which acts as a lookup table. The advantages of CMAC are fast learning convergence, and capability of mapping nonlinear functions due to its local generalization of weight updating, single structure and easy processing. In the training phase, the disadvantage of some CMAC models is unstable phenomenon...
متن کاملThe One-time Learning Hierarchical Cmac and the Memory Limited Ca-cmac for Image Data Compression
Two methods to compress transmitted image data are proposed in this paper. The first method is the one-time learning hierarchical CMAC method and the second is the memory limited CA-CMAC method for image data compression and reconstruction. The one-time learning hierarchical CMAC method is used when a coarse image needs to be sent to the receiver initially and then the image quality is graduall...
متن کاملKinematics Control of Redundant Manipulators Using CMAC Neural Network
The inverse kinematics problems of redundant manipulators have been investigated for many years. The conventional method of solving this problem analytically is by applying the Jacobian Pseudoinverse Algorithm. It is effective and able to resolve the redundancy for additional constraints. However, its demand for computational effort makes it not suitable for real-time control. Recently, neural ...
متن کاملClosed-loop method to improve image PSNR in pyramidal CMAC networks
A closed-loop method to improve image the peak signal to noise ratio (PSNR) in pyramidal cerebellar model arithmetic computer (CMAC) networks is proposed in this paper. We propose a novel coding procedure, which can make the CMAC network learn the feature of the transmitted image with only one-shot training, so some sampled data of the original image can quickly be sent to reconstruct a coarse ...
متن کامل