Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya...
-
Upload
conrad-stanley -
Category
Documents
-
view
225 -
download
0
description
Transcript of Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya...
![Page 1: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/1.jpg)
Web-Mining Agents
Prof. Dr. Ralf MöllerUniversität zu Lübeck
Institut für Informationssysteme
Tanya Braun (Übungen)
![Page 2: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/2.jpg)
ClassificationArtificial Neural Networks
SVMs
R. MoellerInstitute of Information Systems
University of Luebeck
![Page 3: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/3.jpg)
Agenda• Neural Networks• Single-layer networks (Perceptrons)
– Perceptron learning rule– Easy to train
• Fast convergence, few data required– Cannot learn „complex“ functions
• Support Vector Machines• Multi-Layer networks
– Backpropagation learning– Hard to train
• Slow convergence, many data required• Deep Learning
![Page 4: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/4.jpg)
![Page 5: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/5.jpg)
![Page 6: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/6.jpg)
XOR problem
![Page 7: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/7.jpg)
XOR problem
![Page 8: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/8.jpg)
![Page 9: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/9.jpg)
(learning rate)
Proof omitted since neural networks are not in the focus of this lecture
![Page 10: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/10.jpg)
Support Vector Machine Classifier
• Basic idea– Mapping the instances from the two
classes into a space where they become linearly separable. The mapping is achieved using a kernel function that operates on the instances near to the margin of separation.
• Parameter: kernel type
![Page 11: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/11.jpg)
y = +1
y = -1
Nonlinear Separation
![Page 12: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/12.jpg)
![Page 13: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/13.jpg)
margin separator
support vectors
Support Vectors
![Page 14: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/14.jpg)
Literature
Mitchell (1989). Machine Learning. http://www.cs.cmu.edu/~tom/mlbook.html
Duda, Hart, & Stork (2000). Pattern Classification. http://rii.ricoh.com/~stork/DHS.html
Hastie, Tibshirani, & Friedman (2001). The Elements of Statistical Learning. http://www-stat.stanford.edu/~tibs/ElemStatLearn/
![Page 15: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/15.jpg)
Literature (cont.)
Russell & Norvig (2004). Artificial Intelligence. http://aima.cs.berkeley.edu/
Shawe-Taylor & Cristianini. Kernel Methods for Pattern Analysis. http://www.kernel-methods.net/
![Page 16: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/16.jpg)
![Page 17: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/17.jpg)
![Page 18: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/18.jpg)
Z = y1 AND NOT y2 = (x1 OR x2) AND NOT(x1 AND x2)
![Page 19: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/19.jpg)
![Page 20: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/20.jpg)
![Page 21: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/21.jpg)
W1
W2
W3
f(x)
1.4
-2.5
-0.06
David Corne: Open Courseware
![Page 22: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/22.jpg)
2.7
-8.6
0.002
f(x)
1.4
-2.5
-0.06
x = -0.06×2.7 + 2.5×8.6 + 1.4×0.002 = 21.34
David Corne: Open Courseware
![Page 23: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/23.jpg)
A datasetFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
David Corne: Open Courseware
![Page 24: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/24.jpg)
Training the neural network Fields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
David Corne: Open Courseware
![Page 25: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/25.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Initialise with random weights
David Corne: Open Courseware
![Page 26: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/26.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Present a training pattern
1.4
2.7
1.9
David Corne: Open Courseware
![Page 27: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/27.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Feed it through to get output
1.4
2.7 0.8
1.9
David Corne: Open Courseware
![Page 28: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/28.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Compare with target output
1.4
2.7 0.8
01.9 error 0.8
David Corne: Open Courseware
![Page 29: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/29.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Adjust weights based on error
1.4
2.7 0.8
0 1.9 error 0.8
David Corne: Open Courseware
![Page 30: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/30.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Present a training pattern
6.4
2.8
1.7
David Corne: Open Courseware
![Page 31: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/31.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Feed it through to get output
6.4
2.8 0.9
1.7
David Corne: Open Courseware
![Page 32: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/32.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Compare with target output
6.4
2.8 0.9
1 1.7 error -0.1
David Corne: Open Courseware
![Page 33: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/33.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
Adjust weights based on error
6.4
2.8 0.9
1 1.7 error -0.1
David Corne: Open Courseware
![Page 34: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/34.jpg)
Training dataFields class1.4 2.7 1.9 03.8 3.4 3.2 06.4 2.8 1.7 14.1 0.1 0.2 0etc …
And so on ….
6.4
2.8 0.9
1 1.7 error -0.1
Repeat this thousands, maybe millions of times – each timetaking a random training instance, and making slight weight adjustments Algorithms for weight adjustment are designed to makechanges that will reduce the error
David Corne: Open Courseware
![Page 35: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/35.jpg)
The decision boundary perspective…Initial random weights
David Corne: Open Courseware
![Page 36: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/36.jpg)
The decision boundary perspective…Present a training instance / adjust the weights
David Corne: Open Courseware
![Page 37: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/37.jpg)
The decision boundary perspective…Present a training instance / adjust the weights
David Corne: Open Courseware
![Page 38: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/38.jpg)
The decision boundary perspective…Present a training instance / adjust the weights
David Corne: Open Courseware
![Page 39: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/39.jpg)
The decision boundary perspective…Present a training instance / adjust the weights
David Corne: Open Courseware
![Page 40: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/40.jpg)
The decision boundary perspective…Eventually ….
David Corne: Open Courseware
![Page 41: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/41.jpg)
The point I am trying to make• Weight-learning algorithms for NNs are dumb
• They work by making thousands and thousands of tiny adjustments, each making the network do better at the most recent pattern, but perhaps a little worse on many others
• But, by dumb luck, eventually this tends to be good enough to learn effective classifiers for many real applications
David Corne: Open Courseware
![Page 42: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/42.jpg)
Some other points
If f(x) is non-linear, a network with 1 hidden layer can, in theory, learn perfectly any classification problem. A set of weights exists that can produce the targets from the inputs. The problem is finding them.
David Corne: Open Courseware
![Page 43: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/43.jpg)
Some other ‘by the way’ pointsIf f(x) is linear, the NN can only draw straight decision boundaries (even if there are many layers of units)
David Corne: Open Courseware
![Page 44: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/44.jpg)
Some other ‘by the way’ pointsNNs use nonlinear f(x) so theycan draw complex boundaries,but keep the data unchanged
David Corne: Open Courseware
![Page 45: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/45.jpg)
Some other ‘by the way’ pointsNNs use nonlinear f(x) so they SVMs only draw straight lines, can draw complex boundaries, but they transform the data firstbut keep the data unchanged in a way that makes that OK
David Corne: Open Courseware
![Page 46: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/46.jpg)
Deep Learning
aka or related toDeep Neural Networks
Deep Structural LearningDeep Belief Networks
etc,
![Page 47: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/47.jpg)
The new way to train multi-layer NNs…
David Corne: Open Courseware
![Page 48: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/48.jpg)
The new way to train multi-layer NNs…
Train this layer first
David Corne: Open Courseware
![Page 49: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/49.jpg)
The new way to train multi-layer NNs…
Train this layer first
then this layer
David Corne: Open Courseware
![Page 50: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/50.jpg)
The new way to train multi-layer NNs…
Train this layer first
then this layer
then this layer
David Corne: Open Courseware
![Page 51: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/51.jpg)
The new way to train multi-layer NNs…
Train this layer first
then this layer
then this layerthen this layer
David Corne: Open Courseware
![Page 52: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/52.jpg)
The new way to train multi-layer NNs…
Train this layer first
then this layer
then this layerthen this layer
finally this layerDavid Corne: Open Courseware
![Page 53: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/53.jpg)
The new way to train multi-layer NNs…
EACH of the (non-output) layers is trained
to be an auto-encoderBasically, it is forced to learn good features that describe what comes from the previous layer
David Corne: Open Courseware
![Page 54: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/54.jpg)
an auto-encoder is trained, with an absolutely standard weight-adjustment algorithm to reproduce the input
David Corne: Open Courseware
![Page 55: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/55.jpg)
an auto-encoder is trained, with an absolutely standard weight-adjustment algorithm to reproduce the input
By making this happen with (many) fewer units than the inputs, this forces the ‘hidden layer’ units to become good
feature detectors
David Corne: Open Courseware
![Page 56: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/56.jpg)
intermediate layers are each trained to be auto encoders (or similar)
David Corne: Open Courseware
![Page 57: Web-Mining Agents Prof. Dr. Ralf Mller Universitt zu Lbeck Institut fr Informationssysteme Tanya Braun (bungen)](https://reader035.fdocuments.net/reader035/viewer/2022062317/5a4d1b8f7f8b9ab0599c0996/html5/thumbnails/57.jpg)
Final layer trained to predict class based on outputs from previous layers
David Corne: Open Courseware