ezraerb / neuralnetsensorclassifier Goto Github PK
View Code? Open in Web Editor NEWLicense: GNU General Public License v3.0
License: GNU General Public License v3.0
These programs classify data from a gas sensor array. The array was used to collect time series of sensor data under various conditions. The goal is to classify each time series to the correct set of conditions. A classic gradient descent neural network was used to do the classification. The network was explictly implemented instead of using a package to learn the algorithm. The program input is two files, one of sensor data tine series and one of the category each belongs in. The data came from a public data set available here: https://archive.ics.uci.edu/ml/datasets/Gas+sensors+for+home+activity+monitoring The time series are first preprocessed by removing background data and then normalizing to the range 0..1 based on the lowest and higest values seen over all samples. The samples are then split by category and then further split into three groups: training, validation, and test. The split by category ensure that no group has a bias toward a particular category, which can affect the accuracy of the network. The network is trained time series by time series, using the classic early stopping algorithm. The algorighm has a patience period of 3 epochs, meaning it stops after the epocs of increasing validation error. The performance of the final network is tested on the remaining samples to produce a confusion matrix. The precision, recall, and balance F statistic are then calculated for each category, and overall, based on the matrix. Initial results showed that certain sensors produced very noisy data that was confusing the network. Two data plot utilities, SamplePlotter and CrossSampleSensorPlotter were written to visualize the output of various sensors. With these results, sensors were exluded from the netword to potentially improve classification results. Results show excluding noisy sensors did improve classification, but only to a point. Excluding more than one-third of sensors led to worse classification and significantly increased training time. Sample results: All sensors: predicted actual banana background wine banana 5.0 2.0 2.0 background 2.0 5.0 0.0 wine 3.0 1.0 6.0 Category: banana Precision: 0.5 Recall: 0.555555555556 Balanced F Statistic: 0.5 26315789474 Category: background Precision: 0.625 Recall: 0.714285714286 Balanced F Statisti c: 0.666666666667 Category: wine Precision: 0.75 Recall: 0.6 Balanced F Statistic: 0.666666666667 Overall: Precision: 0.625 Recall: 0.62328042328 Balanced F Statistic: 0.61988304 0936 Ignore last two: predicted actual banana background wine banana 4.0 4.0 1.0 background 1.0 5.0 1.0 wine 1.0 1.0 8.0 Category: banana Precision: 0.666666666667 Recall: 0.444444444444 Balanced F Sta tistic: 0.533333333333 Category: background Precision: 0.5 Recall: 0.714285714286 Balanced F Statistic: 0.588235294118 Category: wine Precision: 0.8 Recall: 0.8 Balanced F Statistic: 0.8 Overall: Precision: 0.655555555556 Recall: 0.65291005291 Balanced F Statistic: 0 .640522875817 Ignore last three: predicted actual banana background wine banana 5.0 3.0 1.0 background 0.0 7.0 0.0 wine 2.0 1.0 7.0 Category: banana Precision: 0.714285714286 Recall: 0.555555555556 Balanced F Sta tistic: 0.625 Category: background Precision: 0.636363636364 Recall: 1.0 Balanced F Statistic: 0.777777777778 Category: wine Precision: 0.875 Recall: 0.7 Balanced F Statistic: 0.777777777778 Overall: Precision: 0.741883116883 Recall: 0.751851851852 Balanced F Statistic: 0.726851851852 The code was written on Python 2.7. It requires NumPy and matplotlib. Copyright (C) 2016 Ezra Erb This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License version 3 as published by the Free Software Foundation. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. I'd appreciate a note if you find this program useful or make updates. Please contact me through LinkedIn or github (my profile also has a link to the code depository)
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.