ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
firou1925 has joined #mlpack
firou1925_ has joined #mlpack
firou1925_ has quit [Quit: Page closed]
vivekp has joined #mlpack
pd09041999 has joined #mlpack
rf_sust2018 has joined #mlpack
chandramouli_r has joined #mlpack
Gus_ has joined #mlpack
< Gus_>
Hi. Can anybody give an example of a Convolutional Neural Network is implemented with mlpack? Specially the convolutions operators using different kernels
chandramouli_r has quit [Ping timeout: 256 seconds]
Mulx10 has quit [Ping timeout: 256 seconds]
< Gus_>
Great!
< Gus_>
This one is a very good and clear example for CNNs
< Gus_>
You guys are the bests. Really.
Gus_ has quit [Quit: Page closed]
chandramouli_r has joined #mlpack
dhawgupta has joined #mlpack
dhawgupta has quit [Quit: dhawgupta]
rf_sust2018 has quit [Ping timeout: 240 seconds]
firou1925 has quit [Ping timeout: 250 seconds]
afnan has joined #mlpack
afnan has quit [Client Quit]
dhawgupta has joined #mlpack
< chandramouli_r>
Can someone review my draft proposal and give your feedback ? Thanks in advance !
< dhawgupta>
Anybody who has been able to compile mlpack on Clion, I have done so on the normal command line process, and rebulding it everytime is a real pain, was looking to make use of CLion but the build process simply fails
< dhawgupta>
with mostly these logs ```-- Could NOT find OpenMP_C (missing: OpenMP_C_FLAGS OpenMP_C_LIB_NAMES)
< dhawgupta>
-- Could NOT find OpenMP_CXX (missing: OpenMP_CXX_FLAGS OpenMP_CXX_LIB_NAMES)
< dhawgupta>
-- Could NOT find OpenMP (missing: OpenMP_C_FOUND OpenMP_CXX_FOUND)
< dhawgupta>
-- Updated gitversion.hpp.
< dhawgupta>
CMake Warning at CMakeLists.txt:514 (message):
< dhawgupta>
txt2man not found; man pages will not be generated.
< dhawgupta>
-- Not building Markdown bindings.
< dhawgupta>
-- Found Python: /Users/dhawgupta/anaconda3/bin/python
< dhawgupta>
CMake Error at src/mlpack/tests/CMakeLists.txt:195 (file):
< dhawgupta>
file STRINGS file
< dhawgupta>
"/Users/dhawgupta/mlpack/src/mlpack/tests/mytest/dhawal1.h" cannot be read.
< dhawgupta>
CMake Error at src/mlpack/tests/CMakeLists.txt:208 (file):
< dhawgupta>
file STRINGS file
< dhawgupta>
"/Users/dhawgupta/mlpack/src/mlpack/tests/mytest/dhawal1.h" cannot be read.
< dhawgupta>
See also "/Users/dhawgupta/mlpack/cmake-build-debug/CMakeFiles/CMakeOutput.log".
< dhawgupta>
Problems were encountered while collecting compiler information:
< dhawgupta>
<built-in>:1:10: fatal error: '/Users/dhawgupta/mlpack/cmake-build-debug/src/mlpack/cotire/mlpack_CXX_prefix.hxx' file not found
< dhawgupta>
<built-in>:1:10: fatal error: '/Users/dhawgupta/mlpack/cmake-build-debug/src/mlpack/tests/cotire/mlpack_test_CXX_prefix.hxx' file not found```
< chandramouli_r>
does it compile with -fopenmp
< chandramouli_r>
?
< chandramouli_r>
maybe there
< chandramouli_r>
maybe there's a problem with the version of cmake you use.
< chandramouli_r>
Try adding -fopenmp to the CMakeLists.txt that might work!
pd09041999 has quit [Ping timeout: 268 seconds]
chandramouli_r has quit [Ping timeout: 256 seconds]
heisenbug_ has joined #mlpack
pd09041999 has joined #mlpack
< heisenbug_>
hey, I haven's received any comments on my proposal, can I know the status? I know you guy are very busy with all your GSoC and other commitments. Thank you.
dhawgupta has quit [Quit: dhawgupta]
dhawgupta has joined #mlpack
dhawgupta has quit [Quit: dhawgupta]
yashMustak has joined #mlpack
< yashMustak>
Hey, I wanted to ask whether implementation of A.I. algorithms in my GSoC proposal will be worth and appealing to the goals of mlpack?
jeffin has quit [Remote host closed the connection]
jeffin has joined #mlpack
lozhnikov has quit [Ping timeout: 268 seconds]
dhawgupta_ has joined #mlpack
dhawgupta has quit [Ping timeout: 250 seconds]
dhawgupta_ is now known as dhawgupta
pd09041999 has quit [Ping timeout: 246 seconds]
< sreenik>
rcurtin zoq I have just uploaded my proposal on mlpack-Tensorflow Translator. I know I am late, please have a look if possible :)
< sreenik>
Also, I'll update the Abstract before the final submission
heisenbug_ has joined #mlpack
pd09041999 has joined #mlpack
dhawgupta has quit [Quit: dhawgupta]
< jeffin>
Sreenik :i also went through the code and couldn't find it, may be the code in mlpack/models for digit recognizer, we could try calling model.model() , if it throws error that means it's not there*
< jeffin>
I was work on it to print the summay of layers*
< jeffin>
Something like model.summary() function in keras*
< jeffin>
Working*
dhawgupta has joined #mlpack
Soonmok has joined #mlpack
jeffin has quit [Read error: Connection reset by peer]
heisenbug_ has quit [Ping timeout: 256 seconds]
jeffin has joined #mlpack
pd09041999 has quit [Ping timeout: 250 seconds]
pd09041999 has joined #mlpack
lozhnikov has joined #mlpack
dhawgupta has quit [Quit: dhawgupta]
dhawgupta has joined #mlpack
< sreenik>
jeffin: You are right. There must be some other way to get the layers. Maybe the source code of ffn_impl.hpp can prove helpful, will need digging the code up a bit
< jeffin>
Sreenik : i did go through it closely , couldn't find a catch , let's see if you could find some treasure*
< sreenik>
Haha, ahoy!
dhawgupta has quit [Quit: dhawgupta]
chandramouli_r has joined #mlpack
dhawgupta has joined #mlpack
dhawgupta has quit [Client Quit]
chandramouli_r has quit [Ping timeout: 256 seconds]
pd09041999 has quit [Ping timeout: 268 seconds]
chandramouli_r has joined #mlpack
Hsankesara has joined #mlpack
pd09041999 has joined #mlpack
pd09041999 has quit [Excess Flood]
pd09041999 has joined #mlpack
pd09041999 has quit [Excess Flood]
pd09041999 has joined #mlpack
pd09041999 has quit [Remote host closed the connection]
Hsankesara has quit [Quit: Page closed]
chandramouli_r has quit [Ping timeout: 256 seconds]
< mulx10>
I have been trying hard to dig up the resources. Still, I have not been able to progress considerably.
< mulx10>
Namrata is helping. Despite that I need help.
< mulx10>
Thank You
mulx10 has quit [Client Quit]
< jeffin>
Mux10 : i haven't worked with the Network so unsure , but i will read and let you know :-)
Yashwants19 has joined #mlpack
< Yashwants19>
Hi zoq, Shikhar J : I have posted my proposal on GSoC dashboard. rcurtin had reviewed my proposal.If you will also review my proposal and leave some comments.It would be great for me.
< Yashwants19>
Thank you.
Yashwants19 has quit [Quit: Page closed]
heisenbug_ has joined #mlpack
Soonmok has quit [Quit: Connection closed for inactivity]
rf_sust2018 has joined #mlpack
rf_sust20181 has joined #mlpack
rf_sust2018 has quit [Ping timeout: 250 seconds]
< heisenbug_>
Is anyone up for a discussion on the project String Processing Utilities?
< ShikharJ>
Toshal: I can't seem to open your draft in Google docs, as it is in pdf format. So I can't comment on it at places.
< heisenbug_>
Hey, even I submitted the draft but didn't got any comments...
rf_sust20181 has quit [Ping timeout: 246 seconds]
< ShikharJ>
heisenbug_: I will only be reviewing the proposals which adhere to the areas I'm interested/ I'm comfortable in mentoring. I think I have mentioned general advice in a previous comment.
< jeffin>
Shikharj : under ann/activation function / softsign is the fornula for softsign derivative correct..??
< heisenbug_>
You were listed as Potential Mentor for that project, so I thought you will be reviewing it...
< ShikharJ>
Please be patient regarding the process.
< heisenbug_>
I followed your general advice while drafting the proposal.
< jeffin>
I learned it as 1/(1+ abs(x))^2
< heisenbug_>
Yea ShikharJ
< ShikharJ>
heisenbug_: I thought you mentioned String Processing. Sorry for the confusion :)
< heisenbug_>
Yea, actually I was thinking of another application. That would be my second proposal. I submitted a draft for Application of ANN Algorithms Implemented in mlpack
< heisenbug_>
Although I was not sure if I can submit two proposals to same organisation.
< ShikharJ>
heisenbug_: I think that you can.
< heisenbug_>
Yea, I went through the FAQ and didnt find anything that say I cannot, so I hope I can.
< zoq>
heisenbug_: Please keep in mind, we do this in our free time; at this point (one day before the dealine), we can't provide feedback for every appplication, we do our best to provide as much feedback as we can.
< zoq>
heisenbug_: You can submit up to 3 applications.
< ShikharJ>
jeffin: I'll get back to you regarding that.
< heisenbug_>
Yea, I am sorry.
< zoq>
heisenbug_: No worries.
< heisenbug_>
So, can we discuss a bit about the string utilities, if you can...
i8hantanu has quit [Quit: Connection closed for inactivity]
< zoq>
heisenbug_: I'm about join a meeting, there are some additional information in the mailinglist archive. Take a look at the posts from last year as well.
< heisenbug_>
Ohk, thank you for your time.
< ShikharJ>
jeffin: I think that we're doing std::pow(1 - std::abs(y), 2). Where y = x / (1 + std::abs(x)); So essentially, we reach back to the same expression that you mentioned, however, here the advantage is that we don't need to keep a track of whether x was positive or negative. It is a clever way of computing the derivative, thanks to zoq.
< jeffin>
Shikhaj : oh yes , i didn't see that. Thanks :)
< jeffin>
May be next time i should work out the formula on paper, before asking :)
heisenbug_ has quit [Ping timeout: 256 seconds]
saksham189 has joined #mlpack
pd09041999 has joined #mlpack
saksham189 has quit [Ping timeout: 256 seconds]
atulim has joined #mlpack
< atulim>
@rcurtin @zoq I have submitted my Gsoc Proposal on Gsoc portal. Kindly take a look at it and suggest changes. Sorry for the delay.
< favre49>
zoq: I submitted the final proposal, thanks for all your help :)
< favre49>
rcurtin: I was wondering, what is the convention behind the naming of the ensmallen versions?
dhawgupta has quit [Quit: dhawgupta]
dhawgupta has joined #mlpack
< rcurtin>
favre49: whatever's on my mind that day? no formal convention :)
shrbo has joined #mlpack
shrbo has quit [Ping timeout: 256 seconds]
nvnmo has quit [Ping timeout: 256 seconds]
shrbo has joined #mlpack
shrbo has quit [Client Quit]
dhawgupta has quit [Quit: dhawgupta]
dhawgupta has joined #mlpack
dhawgupta has quit [Client Quit]
dhawgupta has joined #mlpack
favre49 has quit [Quit: Page closed]
lozhnikov has quit [Ping timeout: 246 seconds]
Marelis has joined #mlpack
lozhnikov has joined #mlpack
vivekp has quit [Ping timeout: 245 seconds]
vivekp has joined #mlpack
abernauer has joined #mlpack
< abernauer>
Shared my draft proposal if it could get comments would appreciate it.
abernauer has quit []
dhawgupta has quit [Quit: dhawgupta]
dhawgupta has joined #mlpack
kanishq24 has quit [Read error: Connection reset by peer]
kanishq24 has joined #mlpack
dhawgupta has quit [Quit: dhawgupta]
Suryo has joined #mlpack
< Suryo>
rcurtin, zoq: I'd like to request the same... Some last moment comments on my proposal. I don't know what went wrong the other day when I uploaded the link to the updated draft...
Suryo has quit [Client Quit]
dhawgupta has joined #mlpack
< rcurtin>
Suryo: definitely no time from my end, sorry about that
< rcurtin>
remember that if we need clarification on anything after the application is submitted we can just get in touch and ask :)
dhawgupta has quit [Client Quit]
mrohit[m] has left #mlpack []
dhawgupta has joined #mlpack
Toshal has joined #mlpack
dhawgupta has left #mlpack []
akfluffy has joined #mlpack
< akfluffy>
what are the activation function layers for neural networks that range from -1 to 1?
< Toshal>
ShikharJ: Actually you can comment on the proposal in pdf itself. Select the text on which you want to comment. A message symbol with a plus on it will appear on right edge of the text/page with that you can add the comment.
< Toshal>
Please let me know if this is not clear.
< Toshal>
akfluffy: TanHLayer is one of the activation layer in mlpack.
< akfluffy>
Thanks, I tried that but I forgot "Layer", I just wrote TanH
pd09041999 has quit [Ping timeout: 250 seconds]
< akfluffy>
also, is it possible to just apply an activation function to a single neuron in a layer
abhinavsagar has quit [Quit: Connection closed for inactivity]
Suryo has joined #mlpack
< Suryo>
rcurtin: that's reassuring! Thanks!
Suryo has quit [Client Quit]
jeffin has quit [Ping timeout: 245 seconds]
sreenik has quit [Quit: Page closed]
< akfluffy>
are there any plans to implement the a3c algorithm for reinforcement learning
< zoq>
akfluffy: Not really, I think there are methods that provide better results, are easier to implement and faster as well.
< akfluffy>
like regular q_learning?
< zoq>
or PPO
< akfluffy>
I thought a3c was able to provide better results than DQN
< akfluffy>
hmm
< zoq>
right, wasn't really thinking about DQN
Marelis has quit [Ping timeout: 256 seconds]
< akfluffy>
what are the other policy based methods we have implemented besides DQN?
< akfluffy>
oh, aren't the async methods similar to a3c?
< zoq>
yeah, but I would rather implement something like PPO instead of A3C, at least at this point.
< Toshal>
zoq: I saw your comment on my timeline. Thanks for the feed back. Can I suggest two timelines? One which will be more flexible than the current one and current one itself.
< zoq>
Toshal: What about you mention some "ideas" to strip down the timeline, if necessary? That way you can keep the original timeline and there is no need to create another one.
< akfluffy>
zoq: are there any plans to implement PPO?
pistacja has joined #mlpack
pistacja has quit [Client Quit]
saksham189 has quit [Quit: Page closed]
< akfluffy>
also, is it possible to add a Batch Normalization layer not at the input? What if I wanted to normalize data further in
< zoq>
akfluffy: It's not on my current list.
< zoq>
akfluffy: you could use e.g. batch normalization at any level
heisenbug_ has joined #mlpack
< akfluffy>
thanks
< heisenbug_>
Hey, thanks for comments on my proposal. I went through the suggestions and replied accordingly. I just want to thank all the mentors and people who are helping us through the process. Thank you for your constant guidance.
< heisenbug_>
regarding string utilities why cant we use unicode?
< heisenbug_>
We can use something like hash table or any other more efficient maping algorithm and apply that.
< heisenbug_>
Also in that project do we have to implement more than one algorithm?
< heisenbug_>
Coz I think the main trade off would be how are we mapping these encodings coz after processing we again have to convert it back to test and process of conversion may occur many times depending on the algo/application. So using some better algo for mapping would be my first step. Any suggestoins?
< heisenbug_>
*suggestions?
pistacja has joined #mlpack
pistacja has quit [Ping timeout: 256 seconds]
akfluffy has quit [Remote host closed the connection]
GZennaro has joined #mlpack
< GZennaro>
Hello, I'd like to have some suggestions about the proposal for GSoC 2019
< GZennaro>
I'm a computer science student in NY and I'd love to contribute to your project. My primary coding language is C++
< heisenbug_>
You can submit the draft they will review it and will comment on it.
< heisenbug_>
Although I dont know they will now, coz its late now as less than 24hrs are remaining so...
< GZennaro>
what are you looking for in a candidate?
< heisenbug_>
I am myself a candidate.
< heisenbug_>
You can check the logs, they had some general suggestions.