Orange Forum • View topic - GSoC- Introduction and Some Questions (Neural Networks)

GSoC- Introduction and Some Questions (Neural Networks)

General discussions about Orange and with Orange connected things (data mining, machine learning, bioinformatics...).

GSoC- Introduction and Some Questions (Neural Networks)

Postby mudit3774 » Sun Apr 01, 2012 19:26

Hello,

I am Mudit Raj Gupta, fourth year student of M.Sc. (Hons.) Chemistry and B.E. (Hons.) Electronics and Instrumentation at BITS-Pilani (http://www.bits-pilani.ac.in/). I am interested in Machine Learning, specifically various neural network algorithms and Bio-inspired evolutionary algorithms. I have an interest in implementing various existing algorithms and developing modified/new algorithms related to the same.

I successfully completed my Google Summer of Code - 2011 for the Center for the study of Complex systems - University of Michigan. I implemented various bio-inspired evolutionary algorithms (ant colony, bees etc.) in Repast S. My contributions to Repast S was a part of the latest release of the software. The detailed documentation and code can be found here: http://code.google.com/p/cscs-repast-demos/wiki/Mudit I have done a formal course on Neural networks and it's applications and implemented various nets in my assignments like: Hopfield Network, Perceptron Network, Maxnet, Spatio-Temporal Memories, Bidirectional Associative Memories and Back Propagation. I also completed a project on "Use of Evolutionary Algorithms to train the weights and thresholds of a back-propagating neural networks". The details can be found here: http://code.google.com/p/aco-bp/

I have also worked on various projects related to implementation of Bio-Inspired Evolutionary Algorithms in C/C++. You can check the code and some documentation on the same on all my projects on code.google.com my user profile is : http://code.google.com/u/110675325175605367090/

I am interested in applying for the project - "Neural Networks". I am willing to write and submit a proposal as soon as possible. I would like to know your view on about how many implementations you expect during the summer and from the tentative list, what all algorithms are of special interest to the community.

I propose to implement some from : Single Layer perceptron classifiers (Discrete, Continuous and multi category), Multi-layer Feed forward (Linearly non-separable pattern classification, Back propagation), Single Layer Feedback Network (Hopfield), Associative Memories (Bi directional associative memory, patio-Temporal associations etc.), Matching and Self Organization (MAXNET, Hamming Net, Self Organization Maps, Cluster Discovery) and some recent topics like Listnet etc.


I have surveyed some of the libraries (already mentioned in the discussion) and although they have pretty good collection of models, still we can plan on integrating some and add a few new ones which are of specific interest to the community and optimize.

I hope to be hearing from you soon. I am sorry for a delayed contact. I would also like to request you to please let me know about any specific detail related to the project that is required in the proposal (apart from the ones mentioned on the page)

Thank you for your time. Hope to hear from you soon.

Best Regards,

Mudit Raj Gupta

Re: GSoC- Introduction and Some Questions (Neural Networks)

Postby jurezb » Tue Apr 03, 2012 15:14

I propose to implement some from : Single Layer perceptron classifiers (Discrete, Continuous and multi category), Multi-layer Feed forward (Linearly non-separable pattern classification, Back propagation), Single Layer Feedback Network (Hopfield), Associative Memories (Bi directional associative memory, patio-Temporal associations etc.), Matching and Self Organization (MAXNET, Hamming Net, Self Organization Maps, Cluster Discovery) and some recent topics like Listnet etc.


The only requirement is a solid implementations of the multi-layer feedforward newtork. Instead of implementing 10 different architectures I would love to see a single well-tested, well-documented, lightning fast implementation of the MLP with elegant source code.


Return to General Discussions