scholarly journals Robust design using multiobjective optimisation and artificial neural networks with application to a heat pump radial compressor

2022 ◽  
Vol 8 ◽  
Author(s):  
Soheyl Massoudi ◽  
Cyril Picard ◽  
Jürg Schiffmann
Keyword(s):  
Neural Networks ◽  
Training Data ◽  
Small Scale ◽  
High Fidelity ◽  
A Posteriori ◽  
Multiobjective Optimisation ◽  
Radial Compressor ◽  
Speed Up ◽  
Objective Robustness

Abstract Although robustness is an important consideration to guarantee the performance of designs under deviation, systems are often engineered by evaluating their performance exclusively at nominal conditions. Robustness is sometimes evaluated a posteriori through a sensitivity analysis, which does not guarantee optimality in terms of robustness. This article introduces an automated design framework based on multiobjective optimisation to evaluate robustness as an additional competing objective. Robustness is computed as a sampled hypervolume of imposed geometrical and operational deviations from the nominal point. In order to address the high number of additional evaluations needed to compute robustness, artificial neutral networks are used to generate fast and accurate surrogates of high-fidelity models. The identification of their hyperparameters is formulated as an optimisation problem. In the frame of a case study, the developed methodology was applied to the design of a small-scale turbocompressor. Robustness was included as an objective to be maximised alongside nominal efficiency and mass-flow range between surge and choke. An experimentally validated 1D radial turbocompressor meanline model was used to generate the training data. The optimisation results suggest a clear competition between efficiency, range and robustness, while the use of neural networks led to a speed-up by four orders of magnitude compared to the 1D code.

scholarly journals Opportunities and Constraints in Applying Artificial Neural Networks (ANNs) in Food Authentication. Honey—A Case Study

2021 ◽  
Vol 11 (15) ◽  
pp. 6723
Author(s):  
Ariana Raluca Hategan ◽  
Romulus Puscas ◽  
Gabriela Cristea ◽  
Adriana Dehelean ◽  
Francois Guyon ◽  
...  
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Training Sample ◽  
Training Data ◽  
Elemental Content ◽  
Neural Structure ◽  
Food Authentication ◽  
Data Set ◽  
Artificial Neural

The present work aims to test the potential of the application of Artificial Neural Networks (ANNs) for food authentication. For this purpose, honey was chosen as the working matrix. The samples were originated from two countries: Romania (50) and France (53), having as floral origins: acacia, linden, honeydew, colza, galium verum, coriander, sunflower, thyme, raspberry, lavender and chestnut. The ANNs were built on the isotope and elemental content of the investigated honey samples. This approach conducted to the development of a prediction model for geographical recognition with an accuracy of 96%. Alongside this work, distinct models were developed and tested, with the aim of identifying the most suitable configurations for this application. In this regard, improvements have been continuously performed; the most important of them consisted in overcoming the unwanted phenomenon of over-fitting, observed for the training data set. This was achieved by identifying appropriate values for the number of iterations over the training data and for the size and number of the hidden layers and by introducing of a dropout layer in the configuration of the neural structure. As a conclusion, ANNs can be successfully applied in food authenticity control, but with a degree of caution with respect to the “over optimization” of the correct classification percentage for the training sample set, which can lead to an over-fitted model.

Understanding Time-Series Networks: A Case Study in Rule Extraction

1997 ◽  
Vol 08 (04) ◽  
pp. 373-384 ◽  
Author(s):  
Mark W. Craven ◽  
Jude W. Shavlik
Keyword(s):  
Neural Networks ◽  
Time Series ◽  
Decision Tree ◽  
Predictive Accuracy ◽  
Training Data ◽  
Concept Representation ◽  
Symbolic Representations ◽  
Noisy Time Series ◽  
Trained Network

A significant limitation of neural networks is that the representation they learn are usually incomprehensible to humans. We have developed an algorithm, called TREPAN, for extracting comprehensible, symbolic representations from trained neural networks. Given a trained network, TREPAN produces a decision tree that approximates the concept represented by the network. In this article, we discuss the application of TREPAN to a neural network trained on a noisy time series task: predicting the Dollar–Mark exchange rate. We present experiments that show that TREPAN is able to extract a decision tree from this network that equals the network in terms of predictive accuracy, yet provides a comprehensible concept representation. Moreover, our experiments indicate that decision trees induced directly from the training data using conventional algorithms do not match the accuracy nor the comprehensibility of the tree extracted by TREPAN.

scholarly journals Biological sequence modeling with convolutional kernel networks

2019 ◽  
Vol 35 (18) ◽  
pp. 3294-3302 ◽  
Author(s):  
Dexiong Chen ◽  
Laurent Jacob ◽  
Julien Mairal
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Hybrid Approach ◽  
Training Data ◽  
Supplementary Information ◽  
Small Scale ◽  
Biological Sequences ◽  
Homology Detection ◽  
Binding Prediction ◽  
Sequence Modeling

Abstract Motivation The growing number of annotated biological sequences available makes it possible to learn genotype-phenotype relationships from data with increasingly high accuracy. When large quantities of labeled samples are available for training a model, convolutional neural networks can be used to predict the phenotype of unannotated sequences with good accuracy. Unfortunately, their performance with medium- or small-scale datasets is mitigated, which requires inventing new data-efficient approaches. Results We introduce a hybrid approach between convolutional neural networks and kernel methods to model biological sequences. Our method enjoys the ability of convolutional neural networks to learn data representations that are adapted to a specific task, while the kernel point of view yields algorithms that perform significantly better when the amount of training data is small. We illustrate these advantages for transcription factor binding prediction and protein homology detection, and we demonstrate that our model is also simple to interpret, which is crucial for discovering predictive motifs in sequences. Availability and implementation Source code is freely available at https://gitlab.inria.fr/dchen/CKN-seq. Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals Upscaling models of solute transport in porous media through genetic programming

2007 ◽  
Vol 9 (4) ◽  
pp. 251-266 ◽  
Author(s):  
David J. Hill ◽  
Barbara S. Minsker ◽  
Albert J. Valocchi ◽  
Vladan Babovic ◽  
Maarten Keijzer
Keyword(s):  
Porous Media ◽  
Genetic Programming ◽  
Solute Transport ◽  
Heterogeneous Porous Media ◽  
Training Data ◽  
Solute Flux ◽  
Small Scale ◽  
Transport Models ◽  
Averaged Equations

Due to the considerable computational demands of modeling solute transport in heterogeneous porous media, there is a need for upscaled models that do not require explicit resolution of the small-scale heterogeneity. This study investigates the development of upscaled solute transport models using genetic programming (GP), a domain-independent modeling tool that searches the space of mathematical equations for one or more equations that describe a set of training data. An upscaling methodology is developed that facilitates both the GP search and the implementation of the resulting models. A case study is performed that demonstrates this methodology by developing vertically averaged equations of solute transport in perfectly stratified aquifers. The solute flux models developed for the case study were analyzed for parsimony and physical meaning, resulting in an upscaled model of the enhanced spreading of the solute plume, due to aquifer heterogeneity, as a process that changes from predominantly advective to Fickian. This case study not only demonstrates the use and efficacy of GP as a tool for developing upscaled solute transport models, but it also provides insight into how to approach more realistic multi-dimensional problems with this methodology.

scholarly journals Biological Sequence Modeling with Convolutional Kernel Networks

2017 ◽  
Author(s):  
Dexiong Chen ◽  
Laurent Jacob ◽  
Julien Mairal
Keyword(s):  
Neural Networks ◽  
Convolutional Neural Networks ◽  
Hybrid Approach ◽  
Training Data ◽  
Small Scale ◽  
Biological Sequences ◽  
Biological Sequence ◽  
Homology Detection ◽  
Binding Prediction ◽  
Sequence Modeling

AbstractThe growing number of annotated biological sequences available makes it possible to learn genotype-phenotype relationships from data with increasingly high accuracy. When large quantities of labeled samples are available for training a model, convolutional neural networks can be used to predict the phenotype of unannotated sequences with good accuracy. Unfortunately, their performance with medium- or small-scale datasets is mitigated, which requires inventing new data-efficient approaches. In this paper, we introduce a hybrid approach between convolutional neural networks and kernel methods to model biological sequences. Our method enjoys the ability of convolutional neural networks to learn data representations that are adapted to a specific task, while the kernel point of view yields algorithms that perform significantly better when the amount of training data is small. We illustrate these advantages for transcription factor binding prediction and protein homology detection, and we demonstrate that our model is also simple to interpret, which is crucial for discovering predictive motifs in sequences. The source code is freely available at https://gitlab.inria.fr/dchen/CKN-seq.

scholarly journals Deep Forest: Towards An Alternative to Deep Neural Networks

2017 ◽  
Author(s):  
Zhi-Hua Zhou ◽  
Ji Feng
Keyword(s):  
Neural Networks ◽  
Large Scale ◽  
Deep Neural Networks ◽  
Parallel Implementation ◽  
Parameter Tuning ◽  
Training Data ◽  
Small Scale ◽  
Great Effort ◽  
Training Cost ◽  
Deep Forest

In this paper, we propose gcForest, a decision tree ensemble approach with performance highly competitive to deep neural networks in a broad range of tasks. In contrast to deep neural networks which require great effort in hyper-parameter tuning, gcForest is much easier to train; even when it is applied to different data across different domains in our experiments, excellent performance can be achieved by almost same settings of hyper-parameters. The training process of gcForest is efficient, and users can control training cost according to computational resource available. The efficiency may be further enhanced because gcForest is naturally apt to parallel implementation. Furthermore, in contrast to deep neural networks which require large-scale training data, gcForest can work well even when there are only small-scale training data.

Probabilistic Vehicle Identification Techniques for Semiautomated Transportation Security

2005 ◽  
Vol 1917 (1) ◽  
pp. 190-198
Author(s):  
Narasimha Rao Medeme ◽  
Carlos C. Sun
Keyword(s):  
Neural Networks ◽  
Promising Result ◽  
Search Space ◽  
Transportation Systems ◽  
Training Data ◽  
Posteriori Probability ◽  
Data Sets ◽  
A Posteriori ◽  
Fusion Model ◽  
Transportation Security

Intelligent transportation systems can play a significant role in transportation security in addition to their traditional roles in transportation operations and management. A multidetector semiautomated vehicle surveillance framework is presented. The objective of the framework is to assist in the search for a vehicle of interest involved with security threats such as terrorism, abduction, or crime. When a vehicle of interest is wanted, this framework can be applied to reduce surveillance data sets and thus reduce time and labor. This system estimates the a posteriori probabilities that indicate the closeness of the match between a vehicle of interest and any vehicle in the search space. This paper explores the use of multidetector fusion of video and inductive loop data by means of a linear fusion model. This system classifies vehicle pairs into possible correct match or incorrect match classes and transforms the problem into the probabilistic domain by using Bayesian neural networks and probabilistic neural networks (PNNs). The use of Bayesian and PNN classifiers assumes equal losses. With Bayesian estimation and generalized regression neural networks, the a posteriori probability is used as a threshold representing unequal losses. A comparison between the traditional Bayesian approaches and the equivalent neural network methods is presented. The use of different feature combinations, methods to balance training data sets, forward sequential search, and combined and uncombined feature approaches is also investigated. Field arterial data from southern California show that, by retaining only 29% of the search space, the framework produces 92% accuracy, which is a promising result.

scholarly journals PREDIKSI PERILAKU POLA JUMLAH MAHASISWA MENGGUNAKAN JARINGAN SYARAF TIRUAN DENGAN METODE BACKPROPAGATION

2018 ◽  
Vol 1 (2) ◽  
pp. 145
Author(s):  
Yustria Handika Siregar
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Training Data ◽  
School Year ◽  
Optimal Output ◽  
Architectural Patterns ◽  
Backpropagation Method ◽  
Artificial Neural ◽  
Academic Year

Abstrack - This study aims to predict the behavior of student patterns so that they can predict based on the number of students. To achieve optimal output, this study uses Artificial Neural Networks with the Backpropagation method. Case study conducted at the Asahan University Faculty of Engineering. The data used are data on the number of students in the academic year 2011 to 2013 as training data and 2014 school year data until 2016 as testing data. Furthermore, the data is analyzed with several network architectural patterns and the best patterns will be selected to be implemented into the Matlab R2010 program. The system results show a correlation between the number of students that occurred.   Keywords - Prediction, Artificial Neural Networks, Backpropagation Method, Number of Students

scholarly journals Predicting subsurface soil layering and landslide risk with Artificial Neural Networks: a case study from Iran

2011 ◽  
Vol 62 (5) ◽  
pp. 477-485 ◽  
Author(s):  
Farzad Farrokhzad ◽  
Amin Barari ◽  
Lars Ibsen ◽  
Asskar Choobbasti
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Training Data ◽  
Landslide Risk ◽  
Northern Iran ◽  
Soil Layers ◽  
Subsurface Soil ◽  
Soil Layering ◽  
Artificial Neural

Predicting subsurface soil layering and landslide risk with Artificial Neural Networks: a case study from Iran This paper is concerned principally with the application of Artificial Neural Networks (ANN) in geotechnical engineering. In particular the application of ANN is discussed in more detail for subsurface soil layering and landslide analysis. Two ANN models are trained to predict subsurface soil layering and landslide risk using data collected from a study area in northern Iran. Given the three-dimensional coordinates of soil layers present in thirty boreholes as training data, our first ANN successfully predicted the depth and type of subsurface soil layers at new locations in the region. The agreement between the ANN outputs and actual data is over 90 % for all test cases. The second ANN was designed to recognize the probability of landslide occurrence at 200 sampling points which were not used in training. The neural network outputs are very close (over 92 %) to risk values calculated by the finite element method or by Bishop's method.

Sign in / Sign up

Export Citation Format

Share Document