scholarly journals Electronic model of a dubois fuzzy integration neuron

2009 ◽  
Vol 7 (01) ◽  
Author(s):  
J.L. Pérez S. ◽  
A. Garcés M ◽  
F. Cabiedes C. ◽  
A. Miranda V.
Keyword(s):  
Membership Function ◽  
Integration Method ◽  
Threshold Value ◽  
Activation Function ◽  
Fuzzy Integral ◽  
Electronic Circuits ◽  
Electronic Model ◽  
Input Signals ◽  
Leaky Integrator ◽  
Electronic Neuron

In this work, we present a fuzzy electronic neuron that has a Dubois fuzzy integration method, an activation function with a fuzzy threshold, and a fuzzy response. We generated a fuzzy sum of the input signals and a shooting threshold value defined by means of a triangular or sinusoidal membership function. We present the electronic circuits, the oscilograms of the neuron responses, the value of the fuzzy integral, and we compare their behavior with those of a conventional leaky integrator neuron.

Multifunctional logic gates based on silicon hybrid plasmonic waveguides

2018 ◽  
Vol 32 (02) ◽  
pp. 1850008 ◽  
Author(s):  
Luna Cui ◽  
Li Yu
Keyword(s):  
Integrated Circuits ◽  
Logic Gates ◽  
Threshold Value ◽  
Multimode Interference ◽  
Plasmonic Waveguides ◽  
Optical Logic ◽  
Optical Logic Gates ◽  
Input Signals ◽  
All Optical ◽  
Hybrid Plasmonic Waveguides

Nano-scale Multifunctional Logic Gates based on Si hybrid plasmonic waveguides (HPWGs) are designed by utilizing the multimode interference (MMI) effect. The proposed device is composed of three input waveguides, three output waveguides and an MMI waveguide. The functional size of the device is only 1000 nm × 3200 nm, which is much smaller than traditional Si-based all-optical logic gates. By setting different input signals and selecting suitable threshold value, OR, AND, XOR and NOT gates are achieved simultaneously or individually in a single device. This may provide a way for ultrahigh speed signal processing and future nanophotonic integrated circuits.

scholarly journals Dynamic fuzzy logic functor

2008 ◽  
Vol 6 (02) ◽  
Author(s):  
J. L. Pérez-Silva ◽  
A. Garcés-Madrigal ◽  
A. Miranda-Vitela ◽  
F. Lara-Rosano.
Keyword(s):  
Fuzzy Logic ◽  
Functional Model ◽  
Threshold Value ◽  
Time Dependent ◽  
Threshold Function ◽  
Logical Connective ◽  
Input Signals ◽  
Operation Procedure ◽  
Frequency Threshold ◽  
Fuzzy Connectives

This is a fuzzy logical element from which we can generate all the fuzzy connectives as particular cases of it. This functor was a result of the frequency threshold dependent fuzzy logic. We will show that fuzzy logic connectives are generated by a defined group of inputs from this fuzzy functor as dependence of frequency threshold. If we apply time varying signals to the functor, we will see that the responses of the functor depend on the concordances of the input signals and the frequency threshold value. The value of the logical functor threshold can be a time dependent function in such a way that, when varying the threshold, the form that the functor takes as logical connective changes for each time dependent threshold function value in time. In this work we will define the logical functor, we will show its logical operation procedure and we will show the electronic circuit in which a functional model was implanted.

scholarly journals A Codesigned Integrated Photonic Electronic Neuron

2021 ◽  
Author(s):  
Lorenzo De Marinis ◽  
Alessandro Catania ◽  
Piero Castoldi ◽  
Giampiero Contestabile ◽  
Paolo Bruschi ◽  
...  
Keyword(s):  
Power Consumption ◽  
High Speed ◽  
Activation Function ◽  
Memory Storage ◽  
Cmos Process ◽  
Analog To Digital ◽  
Analog Electronics ◽  
Front End ◽  
Execution Speed ◽  
Electronic Neuron

In the modern era of artificial intelligence, increasingly sophisticated artificial neural networks (ANNs) are implemented, which pose challenges in terms of execution speed and power consumption. To tackle this problem, recent research on reduced-precision ANNs opened the possibility to exploit analog hardware for neuromorphic acceleration. In this scenario, photonic-electronic engines are emerging as a short-medium term solution to exploit the high speed and inherent parallelism of optics for linear computations needed in ANN, while resorting to electronic circuitry for signal conditioning and memory storage. In this paper we introduce a precision-scalable integrated photonic-electronic multiply-accumulate neuron, namely PEMAN. The proposed device relies on (i) an analog photonic engine to perform reduced-precision multiplications at high speed and low power, and (ii) an electronic front-end for accumulation and application of the nonlinear activation function by means of a nonlinear encoding in the analog-to-digital converter (ADC). The device, based on the iSiPP50G SOI process for the photonic engine and a commercial 28 nm CMOS process for the electronic front end, has been numerically validated through cosimulations to perform multiply-accumulate operations (MAC). PEMAN exhibits a multiplication accuracy of 6.1 ENOB up to 10 GMAC/s, while it can perform computations up to 56 GMAC/s with a reduced accuracy down to 2.1 ENOB. The device can trade off speed with resolution and power consumption, it outperforms its analog electronics counterparts both in terms of speed and power consumption, and brings substantial improvements also compared to a leading GPU.

Direct integration method based on dual neural networks to solve the structural reliability of fuzzy failure criteria

2019 ◽  
Vol 233 (19-20) ◽  
pp. 7183-7196 ◽  
Author(s):  
Juan Du ◽  
Haibin Li
Keyword(s):  
Neural Networks ◽  
Mathematical Model ◽  
Membership Function ◽  
Structural Reliability ◽  
Integration Method ◽  
Failure Criteria ◽  
Direct Integration ◽  
Fuzzy Reliability ◽  
Research Attention ◽  
Direct Integration Method

In practical engineering, fuzzy failure criteria can reflect the actual conditions of the normal use and durability of structures. Therefore, this topic has garnered considerable research attention. First, a fuzzy set and a membership function were proposed in this study. A fuzzy reliability mathematical model of structures was obtained by means of the fuzzy random event probability. Second, the distribution forms of common membership functions were introduced, and the optimal membership function was selected based on the Akaike information criterion. Third, considering the difficulty of calculating multiple integrals in the fuzzy reliability mathematical model, a direct integration method based on dual neural networks was introduced. This method provides a new approach for calculating structural reliability with the fuzzy failure criteria. Finally, the proposed method was verified by numerical examples. The results showed that this method could solve structural fuzzy reliability problems with multidimensional random variables with high computational efficiency and accuracy.

scholarly journals Applications of Autoencoders along with Deep Learning Techniques to generate valid molecules

2021 ◽  
Vol 2070 (1) ◽  
pp. 012125
Author(s):  
T Sesha Sai Aparna ◽  
T Anuradha
Keyword(s):  
Deep Learning ◽  
Drug Design ◽  
De Novo ◽  
Chemical Space ◽  
Solution Space ◽  
Threshold Value ◽  
Activation Function ◽  
Generative Models ◽  
Pharmaceutical Chemical ◽  
De Novo Drug Design

Abstract From the moment of identifying the fundamental cause of an illness to its availability in the marketplace, it takes an average of 10 years and almost $2.6 billion dollars to develop a medication. We’re actually hunting for a needle in a haystack, which takes a lot of time, effort, and money. In a solution space of between 1030 and 10100 synthetically viable compounds, we’re seeking for the one molecule that can turn off a disease at the molecular level. The chemical solution space is just too large to adequately screen for the desired molecule. Only a small percentage of the synthetically viable compounds for wet lab research are stored in pharmaceutical chemical repositories. Computational de novo drug design can be used to explore this vast chemical space and develop previously undesigned compounds. Computational drug design can cut the amount of time spent in the discovery phase in half, resulting in a shorter time to market and lower drug prices. Deep learning and artificial intelligence (AI) have opened up new perspectives in cheminformatics, especially in molecules generative models. Recurrent neural networks (RNNs) trained with molecules in the SMILES text format, in particular, are very good at exploring the chemical space. Two baseline models were created for generating molecules, one of the model includes an encoder that takes SMILES as input and then develops a deep generative LSTM model which acts as a hidden layer and the output from layers acts as an input to the decoder. The other baseline model acts the same as the above-mentioned model but it includes latent space, it is simply a representation of compressed data that bring related data points closer together physically. To learn data properties and find simpler data representations for analysis, and weights which are obtained from the previous model to generate more efficient molecules. Then created a custom function to play with the temperature of the softmax activation function which creates a threshold value for the valid molecules to generate. This model enables us to produce new molecules through successful exploration.

scholarly journals Flatten-T Swish: a thresholded ReLU-Swish-like activation function for deep learning

2018 ◽  
Vol 4 (2) ◽  
pp. 76 ◽  
Author(s):  
Hock Hung Chieng ◽  
Noorhaniza Wahid ◽  
Ong Pauline ◽  
Sai Raj Kishore Perla
Keyword(s):  
Deep Learning ◽  
Learning Community ◽  
Classification Accuracy ◽  
Deep Neural Network ◽  
Threshold Value ◽  
Activation Function ◽  
Activation Functions ◽  
Rectified Linear Unit ◽  
Overall Performance ◽  
Fully Connected

Activation functions are essential for deep learning methods to learn and perform complex tasks such as image classification. Rectified Linear Unit (ReLU) has been widely used and become the default activation function across the deep learning community since 2012. Although ReLU has been popular, however, the hard zero property of the ReLU has heavily hindering the negative values from propagating through the network. Consequently, the deep neural network has not been benefited from the negative representations. In this work, an activation function called Flatten-T Swish (FTS) that leverage the benefit of the negative values is proposed. To verify its performance, this study evaluates FTS with ReLU and several recent activation functions. Each activation function is trained using MNIST dataset on five different deep fully connected neural networks (DFNNs) with depth vary from five to eight layers. For a fair evaluation, all DFNNs are using the same configuration settings. Based on the experimental results, FTS with a threshold value, T=-0.20 has the best overall performance. As compared with ReLU, FTS (T=-0.20) improves MNIST classification accuracy by 0.13%, 0.70%, 0.67%, 1.07% and 1.15% on wider 5 layers, slimmer 5 layers, 6 layers, 7 layers and 8 layers DFNNs respectively. Apart from this, the study also noticed that FTS converges twice as fast as ReLU. Although there are other existing activation functions are also evaluated, this study elects ReLU as the baseline activation function.

scholarly journals Boolean Models Guide Intentionally Continuous Information and Computation Inside the Brain

2019 ◽  
Vol 12 (Issue 3) ◽  
pp. 90-98
Author(s):  
Germano Resconi
Keyword(s):  
Boolean Function ◽  
Back Propagation ◽  
Activation Function ◽  
Electrical Signal ◽  
Boolean Models ◽  
Input Signals ◽  
Long Time ◽  
Computer Logic ◽  
Logic Functions ◽  
Formal Neuron

In 1943 Machculloch and Pitts create the formal neuron where many input signals are linearly composed with different weights on the neuron soma. When the soma electrical signal goes over a specific threshold an output is produced. The main topic in this model is that the response is the same response as in a Boolean function used a lot for the digital computer. Logic functions can be simplified with the formal neuron. But there is the big problem for which not all logic functions, as XOR , cannot be designed in the formal neuron. After a long time the back propagation and many other neural models overcame the big problem in some cases but not in all cases creating a lot of uncertainty. The model proposed does not consider the formal neuron but the natural network controlled by a set of differential equations for neural channels that model the current and voltage on the neuron surface. The steady state of the probabilities is the activation state continuous function whose maximum and minimum are the values of the Boolean function associated with the activation time of spikes of the neuron. With this method the activation function can be designed when the Boolean functions are known. Moreover the neuron differential equation can be designed in order to realize the wanted Boolean function in the neuron itself. The activation function theory permits to compute the neural parameters in agreement with the intention.

scholarly journals Design, at transistor level, of a neuron with axonic delay

2009 ◽  
Vol 7 (01) ◽  
Author(s):  
E. Mateos Santillán ◽  
J. L. Pérez Silva
Keyword(s):  
Mathematical Model ◽  
Analog Computer ◽  
Simulation System ◽  
Fundamental Part ◽  
Ramp Function ◽  
Leaky Integrator ◽  
Electronic Neuron ◽  
The Mathematical Model ◽  
Type Response

An electronic neuron designed with only transistors, with the idea of being able to develop to future a VLSI integrated microcircuit is presented. The neuron is of leaky integrator type, with a ramp function with saturation type response and axonic delay. In this work we will present the mathematical model of our neuron, and its electronics main characteristics, as fundamental part of our simulation system, the neural analog computer.

Sign in / Sign up

Export Citation Format

Share Document