Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[WIP][HEP3] Implement convolution for DNA sequence #16

Closed
wants to merge 95 commits into from
Closed

[WIP][HEP3] Implement convolution for DNA sequence #16

wants to merge 95 commits into from

Conversation

hannes-brt
Copy link
Owner

I am merging my code for training conv-nets from DNA sequence into Hebel. This should be done by the end of January 2015. Please follow this issue if you are interested in using Hebel for learning from DNA sequence or would like to test it.

Hannes Bretschneider added 30 commits June 5, 2013 11:57
basic convolution routines for sequences
a2fe24f
use dynamic memory in convolve_sequence_gradient
a8a9d01
convolution gradient now working for all filter sizes that are multip…
50b96ed 
…les of 4
reorder output of max pooling layer
4438f6e
implemented SequenceConvolutionNet
43c8321
factor choice of activation function into method subclassed from Hidd…
d3e0cfb 
…enLayer
implemented bias for filters, changed max pooling to use argmax
ec6666f
remove compile-time constants from kernels, make more compatible with…
c5c247a 
… HiddenLayer
adapted new interface for updating parameters
e9ac90e
fixed idiotic parameter updating scheme from last night
5b5ac34
compatibility for double and float, gradient check
757899d
fixed convolution gradients (debug code remains)
4d1aac7
removed debug code
278bf28
fixed tests, removed loop over filters
1fd1879
removed unnecessary conditional
42f6732
fix gradient check
b147bd0
fixed memory layout in max-pooling functions
4e6ab06
all tests pass!!!
da1e184
L1 and L2 weight decay
5915b6b
add dropout
737dd63
Switched input encoding to using just a single byte per nucleotide
27c667e
Gradients work on new bit-encoding; all tests pass
12596f1
convert sequences to upper case before encoding
94a583e
modified all kernel functions to work on a subarray
452ebbd
changed data layout
b4c6308
removed unused variables
1641a9f
Implemented convolutions on separate subregions
8c904da
got MultiSequenceConvolutionLayer working
1943af0
fixed some bugs in MultiSequenceConvolutionLayer
01fb042
enable multiple fully connected layers in MultiSequenceConvolutionLayer
134d24a
@hannes-brt hannes-brt self-assigned this Jan 3, 2015
@hannes-brt hannes-brt changed the title Implement convolution for DNA sequence [HEP3] Implement convolution for DNA sequence Jan 3, 2015
@hannes-brt hannes-brt changed the title [HEP3] Implement convolution for DNA sequence [WIP][HEP3] Implement convolution for DNA sequence Jan 22, 2015
@ddofer
Copy link

ddofer commented Oct 26, 2015

Any plans to add this?

@hannes-brt
Copy link
Owner Author

I'm sorry, but Hebel is no longer under development. It now makes more sense to add this functionality to a deep learning framework of your choice, e.g. Theano, Chainer, or Torch.

@hannes-brt hannes-brt closed this Oct 26, 2015
@ddofer
Copy link

ddofer commented Oct 26, 2015

Ok, thanks.
(Shame,I wanted to try this quickly with biological sequences, without
trying to "translate" the various methods of importing data).
On Oct 26, 2015 3:22 PM, "Hannes Bretschneider" notifications@github.com
wrote:

I'm sorry, but Hebel is no longer under development. It now makes more
sense to add this functionality to a deep learning framework of your
choice, e.g. Theano, Chainer, or Torch.

—
Reply to this email directly or view it on GitHub
#16 (comment).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees

@hannes-brt hannes-brt

Labels
Hebel Enhancement Proposal
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

None yet
2 participants
@hannes-brt @ddofer