Sigmoid function

About: Sigmoid function is a research topic. Over the lifetime, 2228 publications have been published within this topic receiving 59557 citations. The topic is also known as: S curve.

Research on Two Kinds of Adaptive Synchronization Method for Uncertain Fourth Order Chaotic Systems based on Sigmoid Function and Soft Function Method

Abstract: 1 College of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China 2 College of Computer and Communication Engineering, Zhengzhou University of Light Industry, Zhengzhou 450002, China 3 Department of Control Engineering, Naval Aeronautical and Astronautical University, Yantai Shandong 264001, China 4 Department of Armament Engineering, Naval Aeronautical and Astronautical University, Yantai Shandong 264001, China a shangzl@zzuli.edu.cn, b cjcheng@zzuli.edu.cn, c leijunwei@126.com, d richkey1980@gmail.com,

Random number generation system and program

Abstract: PROBLEM TO BE SOLVED: To provide a random number generation system, which arithmetic load is reduced, which can generate random numbers widely and successively for changes of an external input value.SOLUTION: An output function of a neuron, which configures a chaos neural network, for an internal state of the neuron is assumed as an asymmetric segment linear function in place of a sigmoid function. And by assuming output of the asymmetric segment linear function as a double variable, lower 3 bits of the mantissa part are truncated and numbers upper than the lower 3 bits are taken out.

Fine tuning with sigmoid functions in robust fixed point transformation

Abstract: In this paper a novel implementation of the adaptive controllers designed by the use of Robust Fixed Point Transformation is studied. Instead guaranteeing global stability the so designed controllers work smoothly in a bounded region of operation. Both the limits of this region as well as the performance of the controller depends on the basic component of the RFPT-based design, i.e. on the properties of a sigmoid function that can be defined in various manners. In this case a special, easily parameterizable sigmoid was chosen that is widely used in the daily engineering practice, a truncated linear function. This function has a single parameter, its slope, that in the same time determines the width of the window within which the adaptive nature of the controller is guaranteed. It was found that by varying this parameter either the precision of the controller or the frequency of the necessary adaptive tuning can be improved. This statement is substantiated by simulations.

Investigation of Node Pruning Criteria for Neural Networks Model Compression with Non-Linear Function and Non-Uniform Network Topology

Abstract: This paper investigates node-pruning-based compression for non-uniform deep learning models such as acoustic models in automatic speech recognition (ASR). Node pruning for small footprint ASR has been well studied, but most studies assumed a sigmoid as an activation function and uniform or simple fully-connected neural networks without bypass connections. We propose a node pruning method that can be applied to non-sigmoid functions such as ReLU and that can deal with network topology related issues such as bypass connections. To deal with non-sigmoid functions, we extend a node entropy technique to estimate node activities. To cope with non-uniform network topology, we propose three criteria; inter-layer pairing, no bypass connection pruning, and layer-based pruning rate configuration. The proposed method as a combination of these four techniques and criteria was applied to compress a Kaldi's acoustic model with ReLU as a non-linear function, time delay neural networks (TDNN) and bypass connections inspired by residual networks. Experimental results showed that the proposed method achieved a 31% speed increase while maintaining the ASR accuracy to be comparable by taking network topology into consideration.

An Agent-Based Framework for Policy Simulation: Modeling Heterogeneous Behaviors With Modified Sigmoid Function and Evolutionary Training

Abstract: This article proposes an agent-based policy simulation framework that can be applied to the cases satisfying: 1) the agents try to maximize some intertemporal preference and 2) the impacts of different factors on agents’ behavioral tendency are monotonic. By combining the simulation and optimization methods, this framework balances the flexibility and validity of agent-based models (ABMs): the sigmoid function is modified and used to model agents’ decision-making rules, and the evolutionary training method is used to calibrate agents’ behavioral parameters. Based on an example for the emission trading scheme, the application of the framework is presented and evaluated in detail.