Machine Learning Meetup Notes: 2008-12-17

From Noisebridge
(Difference between revisions)
Jump to: navigation, search
(New page: These are the mostly-raw notes from the machine learning group meeting on 2008-12-17. People attending: Bill, Jean, Jeremy, Josh, Matt, Mike, Praveen (and ruben made a cameo) Here's what...)
 
(fixing links -- to give external links a name, use a space (not a pipe))
 
Line 7: Line 7:
 
'''matt'''
 
'''matt'''
 
* interested in swarms and modeling multiple agents as applied to robots
 
* interested in swarms and modeling multiple agents as applied to robots
* [http://www.amazon.com/Intelligence-Jeff-Hawkins/dp/0805074562|Jeff Hawkins, On Intelligence (at Amazon)]
+
* [http://www.amazon.com/Intelligence-Jeff-Hawkins/dp/0805074562 Jeff Hawkins, On Intelligence (at Amazon)]
* [http://www.onintelligence.org/||On Intelligence, the website]
+
* [http://www.onintelligence.org/ On Intelligence, the website]
  
 
'''bill'''  
 
'''bill'''  
Line 20: Line 20:
 
* fun problem is document clustering, semantic relatedness in documents
 
* fun problem is document clustering, semantic relatedness in documents
 
* interested in unsupervised and semisupervised learning
 
* interested in unsupervised and semisupervised learning
* [http://glaros.dtc.umn.edu/gkhome/views/cluto/|Cluto, a set of clustering tools with great visualization]
+
* [http://glaros.dtc.umn.edu/gkhome/views/cluto/ Cluto, a set of clustering tools with great visualization]
 
* boosting was mentioned
 
* boosting was mentioned
  
Line 28: Line 28:
 
* interested in creating a taxonomy/survey of machine learning techniques
 
* interested in creating a taxonomy/survey of machine learning techniques
 
* genetic neural networks google talk
 
* genetic neural networks google talk
* [http://www.cs.waikato.ac.nz/ml/weka/|Weka homepage], a machine learning toolkit
+
* [http://www.cs.waikato.ac.nz/ml/weka/ Weka homepage], a machine learning toolkit
* [http://www.youtube.com/watch?v=_m97_kL4ox0|Polyworld: Using Evolution to Design Artificial Intelligence], a google techtalk about evolving neural network topologies via genetic algorithms
+
* [http://www.youtube.com/watch?v=_m97_kL4ox0 Polyworld: Using Evolution to Design Artificial Intelligence], a google techtalk about evolving neural network topologies via genetic algorithms
* [http://www.youtube.com/watch?v=AyzOUbkUf3M|The Next Generation of Neural Networks], a google techtalk on Restricted Boltzmann Machines, used for the NIST digit recognition task.
+
* [http://www.youtube.com/watch?v=AyzOUbkUf3M The Next Generation of Neural Networks], a google techtalk on Restricted Boltzmann Machines, used for the NIST digit recognition task.
* [http://kittenwars.com|An example of human-driven binary comparison of attributes]
+
* [http://kittenwars.com An example of human-driven binary comparison of attributes]
  
 
'''jean'''  
 
'''jean'''  
Line 37: Line 37:
 
* interested in exploring/surveying different techniques (taxonomy)
 
* interested in exploring/surveying different techniques (taxonomy)
 
* in particular with applications in eeg/fmri us
 
* in particular with applications in eeg/fmri us
* [http://www.pymvpa.org/|Python MultiVariate Pattern Analysis], a python packages that implements a lot of Multivariate pattern analysis algorithms
+
* [http://www.pymvpa.org/ Python MultiVariate Pattern Analysis], a python packages that implements a lot of Multivariate pattern analysis algorithms
* [http://www.gnu.org/software/octave/|GNU Octave], the GNU implementation of matlab  (good matlab replacement for PCA)
+
* [http://www.gnu.org/software/octave/ GNU Octave], the GNU implementation of matlab  (good matlab replacement for PCA)
  
 
'''praveen'''
 
'''praveen'''
Line 45: Line 45:
 
* looking to collaborate on anything, wanting to get more hands on PCA experience
 
* looking to collaborate on anything, wanting to get more hands on PCA experience
 
* perl script that will make million$
 
* perl script that will make million$
* [http://www.swarm.org/|Swarm Development Group Wiki],
+
* [http://www.swarm.org/ Swarm Development Group Wiki],
* [http://en.wikipedia.org/wiki/Self-organizing_map|Self-organizing maps], a neural network technique he's worked with in the past
+
* [http://en.wikipedia.org/wiki/Self-organizing_map Self-organizing maps], a neural network technique he's worked with in the past
* [http://www.pbs.org/wgbh/nova/apegenius/program.html|Ape Genius (NOVA)], source of some anecdotes we shared (chapter 6)
+
* [http://www.pbs.org/wgbh/nova/apegenius/program.html Ape Genius (NOVA)], source of some anecdotes we shared (chapter 6)
  
 
'''ruben110'''
 
'''ruben110'''

Latest revision as of 15:33, 26 December 2008

These are the mostly-raw notes from the machine learning group meeting on 2008-12-17.

People attending: Bill, Jean, Jeremy, Josh, Matt, Mike, Praveen (and ruben made a cameo)

Here's what the people who attended were interested in:

matt

bill

  • in the past work on optimization of functions (optimization theory), interested in just learning anything

mike

  • interested in learning and absorbing new info

jeremy

josh

jean

  • working at ucsf doing cognitive research
  • interested in exploring/surveying different techniques (taxonomy)
  • in particular with applications in eeg/fmri us
  • Python MultiVariate Pattern Analysis, a python packages that implements a lot of Multivariate pattern analysis algorithms
  • GNU Octave, the GNU implementation of matlab (good matlab replacement for PCA)

praveen

  • currently @ linden lab
  • worked with gene expresssion data mining, music preference and basket analysis,
  • looking to collaborate on anything, wanting to get more hands on PCA experience
  • perl script that will make million$
  • Swarm Development Group Wiki,
  • Self-organizing maps, a neural network technique he's worked with in the past
  • Ape Genius (NOVA), source of some anecdotes we shared (chapter 6)

ruben110

  • facial recognition anecdotes

[edit] Collaborative project kick off for year end 2008

  • Create a survey/taxonomy list of ML techniques -- when setting out to do a particular machine learning task, what is the shortlist of techniques/tools/insights?
Personal tools