Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

13
Machine Learning Group University College Dublin 4.30 Machine Learning 4.30 Machine Learning Pádraig Cunningham
  • date post

    23-Jan-2016
  • Category

    Documents

  • view

    222
  • download

    0

Transcript of Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Page 1: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Machine Learning GroupUniversity College Dublin

4.30 Machine Learning4.30 Machine Learning

Pádraig Cunningham

Page 2: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

2

OutlineOutline

Week 1 Introduction & General Overview of Matrix Decomposition Nearest Neighbour Classifiers Tutorial

Week 2: Neural Networks Simple Perceptron, Backpropagation Other Architectures: Hopfield, Self-Organising Maps Tutorial

Week 3 Support Vector Machines Kernel Methods & Evaluation Tutorial

Week 4 Decision Trees Naïve Bayes Tutorial

Page 3: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

3

OutlineOutline

Week 5: Ensemble Techniques Bagging Boosting Tutorial

Week 6: Unsupervised Learning Hierarchical Clustering Other Clustering Algorithms: k-Means, Spectral Clustering Tutorial

Week 7: Dimension Reduction Principle Components Analysis, LSI, SVD Feature Selection Tutorial

Later 2 revision tutorials

Coursework3-4 pieces, 15 hours, Weka & Java

Page 4: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

4

Why Machine LearningWhy Machine Learning

Recent progress in algorithms and theory Loads of processing power Computational power is available Growing flood of

online data Amazon Google

Page 5: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

5

3 niches for ML3 niches for ML

Data mining: using historical data to improve decisions medical records medical knowledge

Software applications that cannot be programmed by hand. autonomous driving speech recognition i.e. weak theory domains.

Self customising programs Personalised Newspaper E-mail filtering

Page 6: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

6

Data-mining in medical recordsData-mining in medical recordsQuality Assurance in Maternity Care.http://svr-www.eng.cam.ac.uk/projects/qamc/qamc.html

Page 7: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

7

Rule LearningRule Learning

The QAMC system uses Decision /trees (I think!) It is also possible to extract rules from data:-  

If No previous normal delivery, andAbnormal 2nd Trimester Ultrasound, andMalpresentation at admission

Then Probability of Emergency C-Section is 0.6

 Over training dat 26/41 = 0.63Over test data: 12/20 = 0.6

<Rule taken from Machine Learning by Tom Mitchell>

Page 8: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

8

Spam FilteringSpam Filtering

For Machine Learning… Lots of training data High dimensionality data (lots of features) Email is a diverse concept

Porn, mortgage, religion, cheap drugs… Work, family, play…

Spam Filtering is a challenge because… Arms race: spammers vs filters False Positives are unacceptable

Spam is a changing concept

Page 9: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

9

ALVINALVIN

Problems too difficult to program by hand

Alvin drives at 70mph on motorways

Page 10: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

10

Autonomous VehiclesAutonomous Vehicles

DARPA Grand Challenge 2005 Winner: Stanley from Stanford

Various modules use ML

Page 11: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

11

SmartRadioSmartRadio

Internet-based music radio Personalised

Collaborative Recommendation Content-Based Recommendation

supported by knowledge discovery from log data supported by feature extraction from sound files

feature seleciton refinement

Page 12: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

12

Smart RadioSmart Radio

Smart Radio is a web based client-server music application which allows listeners build, manage and share music programmes

The project was set up to look at a possible model for:

The regulated distribution of music on the web

A personalised stream of music service

To provide an architecture and data to test our data mining and collaborative filtering algorithms

Page 13: Machine Learning Group University College Dublin 4.30 Machine Learning Pádraig Cunningham.

Intro to ML

13

ML DimensionsML Dimensions

Lazy v’s Eager k-NN v’s rule learning

Supervised v’s Unsupervised Symbolic v’s Sub-symbolic