A novel one-layer recurrent neural network for the l1-regularized least square problem

Mohammadi, Majid; Tan, Yao‐Hua; Hofman, Wout; Mousavi, S. Hamid

doi:10.1016/j.neucom.2018.07.007

“…The decrease in the mean square error value also proves the working of feed-back gradient descent optimization mechanism of algorithm, through which the parameters are made to fit into the model correctly by improving the weight of the input data 31 . It also proves the convergence of the data reliable to the robust model for the given dataset for the constant parameters 32 .…”

Section: Mean Square Errorsupporting

confidence: 59%

Construction of Feed Forward MultiLayer Perceptron Model For Genetic Dataset in Leishmaniasis Using Cognitive Computing

Mahalingam

¹

,

Kabra

²

,

Singh

³

2018

Preprint

0

View full text Add to dashboard Cite

Leishmaniasis is an endemic parasitic disease, predominantly found in the poor locality of Africa, Asia and Latin America. It is associated with malnutrition, weak immune system of people and their housing locality. At present, it is diagnosed by microscopic identification, molecular and biochemical characterisation or serum analysis for parasitic compounds. In this study, we present a new approach for diagnosing Leishmaniasis using cognitive computing. The Genetic datasets of leishmaniasis are collected from Gene Expression Omnibus database and it's then processed. The algorithm for training and developing a model, based on the data is prepared and coded using python. The algorithm and their corresponding datasets are integrated using TensorFlow dataframe. A feed forward Artificial Neural Network trained model with multi-layer perceptron is developed as a diagnosing model for Leishmaniasis, using genetic dataset. It is developed using recurrent neural network. The cognitive model of the trained network is interpreted using the maps and mathematical formula of the influencing parameters. The credit of the system is measured using the accuracy, loss and error of the system. This integrated system of the leishmaniasis genetic dataset and neural network proved to be the good choice for diagnosis with higher accuracy and lower error. Through this approach, all records of the data are effectively incorporated into the system. The experimental results of feed forward multilayer perceptron model after normalization; mean square error (219.84), loss function (1.94) and accuracy (85.71%) of the model, shows good fit of model with the process and it could possibly serve as a better solution for diagnosing Leishmaniasis in future, using genetic datasets.The code is available in Github repository: https://github.com/shailzasingh/Machine-Learning-code-for-analyzing-genetic-dataset-in-Leishmaniasis

show abstract

“…The latter observation describes a form of stability, which coincides with the conclusion presented in Theorem 2. In addition, it is quite expected in compressive sampling that augmenting the number of measurements m generally improves the recovering quality (progressing from ill‐posed and under‐determined systems to well‐posed ones) 20,21,35–36 . This fact is also noticed in the different plots of Figures 2 and 3.…”

Section: Simulation and Discussionmentioning

confidence: 64%

“…In the last decades, numerous optimization issues that arise in various engineering applications require real‐time processing 8,20,21 . Classical digital techniques, as gradient projection‐based algorithms or subgradient strategies and so forth, are mostly not convenient from the standpoint of computational time or resource allocation in computer networks 16,17,22–23 .…”

Section: Introductionmentioning

confidence: 99%

LPNN‐based approach for LASSO problem via a sequence of regularized minimizations

ZeglaouiAnis

¹

,

HoumiaAnouar

²

,

Mejai

³

et al. 2021

Adaptive Control & Signal

0

View full text Add to dashboard Cite

Summary In compressive sampling theory, the least absolute shrinkage and selection operator (LASSO) is a representative problem. Nevertheless, the non‐differentiable constraint impedes the use of Lagrange programming neural networks (LPNNs). We present in this article the 𝒫k‐LPNN model, a novel algorithm that tackles the LASSO minimization together with the underlying theory support. First, we design a sequence of smooth constrained optimization problems, by introducing a convenient differentiable approximation to the non‐differentiable l1‐norm constraint. Next, we prove that the optimal solutions of the regularized intermediate problems converge to the optimal sparse signal for the LASSO. Then, for every regularized problem from the sequence, the 𝒫k‐LPNN dynamic model is derived, and the asymptotic stability of its equilibrium state is established as well. Finally, numerical simulations are carried out to compare the performance of the proposed 𝒫k‐LPNN algorithm with both the LASSO‐LPNN model and a standard digital method.

show abstract

“…is a (3p + 3)-dimensional column vector, w s ≥ 0,(s � 1, 2, 3), and 3 s�1 w s � 1. To solve problem ( 20)-( 23) we first use the following proposition [72]. Let u, v ∈ R n be auxiliary variables such that…”

Section: An Optimization Modelmentioning

confidence: 99%

A Generalized Bridge Regression in Fuzzy Environment and Its Numerical Solution by a Capable Recurrent Neural Network

Karbasi

¹

,

Rabiei

²

,

Nazemi

³

2020

Journal of Mathematics

View full text Add to dashboard Cite

Bridge regression is a special family of penalized regressions using a penalty function ∑ A j γ with γ ≥ 1 that for γ = 1 and γ = 2 , it concludes lasso and ridge regression, respectively. In case where the output variable in the regression model was imprecise, we developed a bridge regression model in a fuzzy environment. We also exhibited penalized fuzzy estimates for this model when the input variables were crisp. So, we perform the presented optimization problem for the model that leads to a multiobjective program. Also, we try to determine the shrinkage parameter and the tuning parameter from the same optimization problem. In order to estimate fuzzy coefficients of the proposed model, we introduce a hybrid scheme based on recurrent neural networks. The suggested neural network model is constructed based on some concepts of convex optimization and stability theory which guarantees to find the approximate parameters of the proposed model. We use a simulation study to depict the performance of the proposed bridge technique in the presence of multicollinear data. Furthermore, real data analysis is used to show the performance of the proposed method. A comparison between the fuzzy bridge regression model and several other shrinkage models is made with three different well-known fuzzy criteria. In this study, we visualize the performance of the model by Taylor’s diagram and Bubble plot. Also, we examine the predictive ability of the model, thus, obtained by cross validation. The numerical results clearly showed higher accuracy of the proposed fuzzy bridge method compared to the other existing fuzzy regression models: fuzzy bridge regression model, multiobjective optimization, recurrent neural network, stability convergence, and goodness-of-fit measure.

show abstract

A novel one-layer recurrent neural network for the l1-regularized least square problem

Cited by 13 publications

References 35 publications

Construction of Feed Forward MultiLayer Perceptron Model For Genetic Dataset in Leishmaniasis Using Cognitive Computing

Construction of Feed Forward MultiLayer Perceptron Model For Genetic Dataset in Leishmaniasis Using Cognitive Computing

LPNN‐based approach for LASSO problem via a sequence of regularized minimizations

A Generalized Bridge Regression in Fuzzy Environment and Its Numerical Solution by a Capable Recurrent Neural Network

Contact Info

Product

Resources

About