Download presentation
Presentation is loading. Please wait.
Published byDerick Morgan Modified over 9 years ago
1
Jonathan Reagan Umass Dartmouth CSUMS Summer 11 August 3 rd 2011
2
What is a Neural Network? How does it work? Why do we care? Results Issues encountered Future work
3
Input Layers Hidden Layers Output Layers
4
Not realistic to study every possible case Smaller sample can be used to model the entire case Assume connections hold
5
(input)=[age, income, credit score, etc] (output)=[dependability] We want weights of α’s X* α (Hidden)=Y
6
Use the learning method to find α I Y-Xα I=0
7
PerceptronLeast Square NAccuracys Failed TrialsNAccuracys Failed Trials MinAVGMaxN/Total MinAVGMaxN/Total 20.40480.52960.60810/1002.4081.5310.63060/100 30.39680.52470.61610/1003.4000.5248.61770/100 40.39680.52920.63060/1004.3984.5195.61940/100 50.40810.53120.62740/1005.4048.5154.63060/100 60.39840.54460.61612/1006.4081.5248.62260/100 70.41450.53120.61949/1007.3645.5308.63060/100 80.41940.5440.617720/1008.3790.5374.63060/100 90.39520.54440.62134/1009.3742.5410.63230/100 100.40480.54650.62949/10010.3726.5412.63710/100
10
2 million Convergence Failed Trials 4 million convergence Failed Trials NT(Time)SecondsN/TotalNT(Time)SecondsN/Total 20.030/5020.030/50 30.040/5030.050/50 40.230/5040.240/50 50.930/5050.880/50 610.490/50610.40/50 750.24.2/50779.8.2/50 8143.07.8/508260.2.7/50 9259.815/509484.714/50 10363.622/5010727.522/50 11483.928/5011928.228/50 12560.134/50121096.133/50 13661.740/50131287.238/50 14732.743/50141416.242/50 15750.444/50151463.744/50 16778.146/50161508.946/50 17819.348/50171607.748/50 1883250/50181665.450/50
11
Random Data can’t be learned Deterministic Data can be learned Adding Random variance decreases Accuracy More values of N the Better But more values of N take Longer
12
Increase the speed of the Neural Network Find more applicable data for testing of the Neural Network Try multiple layer Neural Networks and Compare
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.