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/
rob1 has joined #mlpack
robb_ has joined #mlpack
< robb_>
Hey guys, so the FFN template requires that the InitializationRuleType has a default constructor, but it doesn't
< robb_>
in the case of ConstInitialization
< robb_>
So you can't use ConstInitialization as a template argument
< robb_>
Shouldn't the default constructor of ConstInitialization initialize the weights to 0?
rob1 has quit [Ping timeout: 256 seconds]
< robb_>
Since ZeroInitialization got removed
Mina has quit [Quit: Page closed]
robb_ has quit [Quit: Page closed]
bishwa has joined #mlpack
bishwa has quit [Quit: Page closed]
Cross has joined #mlpack
Cross has quit [Client Quit]
aman_p has joined #mlpack
Shashwat has joined #mlpack
Shashwat has quit [Client Quit]
< Soonmok>
Hi! I'm trying to run mlpack with gpu using nvblas. I installed cuda tool-kit which include nvblas, but when I run mlpack application, it doesn't seem that it use gpu (I checked it using command Nvidia-smi). what should I check to use nvidia gpu?
< ShikharJ>
Soonmok: I think Ryan would be the person to help you regarding that. Let's wait for his reply.
< ShikharJ>
zoq: rcurtin: I think the wiki page for testing guidelines is ready now.
aman_p has quit [Ping timeout: 246 seconds]
rajiv_ has joined #mlpack
< rajiv_>
Hi, I don't understand how the Backward() function in the ann/layers works. I started with the simplest linear_impl.hpp, I don't understand the line: g = weight.t() * gy; It would be great if someone could help me out
rajiv_ has quit [Client Quit]
< ayesdie>
ShikharJ: Hi ShikharJ, my name is Ayush Chamoli, and I'm a student in India and also an aspirant for GSoC 2019 and I would like to apply for the `Essential Deep Learning Modules` which looks interesting to me. I've been handling some issues and got accustomed to the mlpack's codebase. I followed the work you did on `GAN` and `RBM` last year and found that you had some PR that were still open. If you don't mind, can I work on
< ayesdie>
those completing those?
< ayesdie>
on completing those*
< ShikharJ>
ayesdie: Yeah sure, if you see any incomplete open PR that isn't being worked upon, feel free to fix it up.
< ShikharJ>
rajiv_: I'm assuming you know how backpropagation works? Between that and the documentation for Backward() function, I think you should be fine? Can you be specific regarding your query?
< ayesdie>
I was asking for the #1477 CIFAR-10 test that was still open
tymefighter has joined #mlpack
< ShikharJ>
ayesdie: Yeah, any includes that one too :)
kstv_13 has joined #mlpack
< ayesdie>
Alright, and sorry for misunderstanding.
awad has joined #mlpack
< awad>
i wanna ask about graph data meaning in string processing utilities and what are other kind of data we need to support and at this step i understand we need only to convert data using techniques like one hot enconding or dictionary and how to recover it and test this algorithms is there any thing else after that i will give this algorithm to whom and thanks in advance
< awad>
i mean graph data is it edge and nodes like facebook or other meaning
tymefighter has quit [Ping timeout: 256 seconds]
awad has quit [Quit: Page closed]
kstv_13 has quit [Ping timeout: 245 seconds]
aman_p has joined #mlpack
kstv_13 has joined #mlpack
kstv_13 has quit [Client Quit]
awad has joined #mlpack
awad has quit [Client Quit]
rajiv_ has joined #mlpack
< rajiv_>
ShikharJ: I was a bit confused with the notation but now I understood :)
< rajiv_>
ShikharJ: I wanted to know about the Backward() as I'm working on dense blocks now. Dense block is a combination of bn-relu-conv layers and that unit can be called a conv_block. So, a dense block contains many conv_blocks. How should I perform Backward() in this case?
< zoq>
mulx10: I don't think LoadCSV can load text.
Prinshu_Kumar has joined #mlpack
st_sg has joined #mlpack
st_sg has quit [Client Quit]
mulx10 has joined #mlpack
< mulx10>
zoq: data::Load with DataSetInfo can
< mulx10>
which in turn calls LoadCSV, but it loads a part of it. I actually tweaked its expression to load the text as well.
< mulx10>
currently I 'm testing it.
aman_p has quit [Ping timeout: 250 seconds]
mulx10 has quit [Quit: Page closed]
st_sg has joined #mlpack
st_sg has quit [Client Quit]
rob has joined #mlpack
rob is now known as Guest1399
aman_p has joined #mlpack
Prinshu_Kumar has quit [Quit: Page closed]
awad_ has joined #mlpack
< zoq>
mulx10: Ah, nice!
Guest1399 has quit [Ping timeout: 256 seconds]
Sergobot has joined #mlpack
< Sergobot>
Hi everyone! My nane is Sergey and I'm studying applied maths and CS in Saint-Petersburg. I want to contribute to mlpack as part of GSoC 2019. Can I talk to someone about it? :)
< zoq>
Sergobot: Hello and welcome, happy to help.
< Sergobot>
Awesome
< Sergobot>
So, I went through the list of ideas and liked NEAT the most
st_sg has joined #mlpack
< zoq>
Sergobot: Glad you liek the idea :)
< Sergobot>
It feels really interesting but a somewhat courageous to me, as I may lack neccessary experiences
st_sg has quit [Client Quit]
< Sergobot>
It says only basic C++ and understanding of how NNs work are required but I wonder if that's it
< Sergobot>
Like, the idea itself is far from trivial and it's surely same for the implementation
awad_ has quit [Ping timeout: 256 seconds]
< zoq>
I might be biased but the main idea behind NEAT is somewhat straightforward; My recommendation is to get familiar with the optimizer framework (ensmallen), the paper might be interesting as well (http://www.ensmallen.org/files/ensmallen_2018.pdf). Take a closer look at some existing methods, that might give you an impression of how the project could be structured.
< awad_>
i wanna ask about graph data meaning in string processing utilities and what are other kind of data we need to support and at this step i understand we need only to convert data using techniques like one hot enconding or dictionary and how to recover it and test this algorithms is there any thing else after that i will give this algorithm to whom and i mean graph data is it edge and nodes like facebook or other meaning and if it like fa
tavishjain has joined #mlpack
st_sg has joined #mlpack
awad_ has quit [Quit: Page closed]
st_sg has quit [Client Quit]
tavishjain has quit [Quit: Page closed]
awad has joined #mlpack
st_sg has joined #mlpack
st_sg has quit [Client Quit]
< awad>
any mentors here
st_sg has joined #mlpack
awad has quit [Quit: Page closed]
st_sg has quit [Client Quit]
aman_p has quit [Ping timeout: 255 seconds]
awad has joined #mlpack
st_sg has joined #mlpack
st_sg has quit [Client Quit]
< awad>
how can i contact with Ryan Curtin
< Sergobot>
Please be patient, mentors can't be avaliable 24/7. They have their own lives and stuff.
Guest65 has joined #mlpack
Guest65 has quit [Remote host closed the connection]
< awad>
it's my dream to be part of open source community and participate in GSOC sorry
zenzu has joined #mlpack
st_sg has joined #mlpack
st_sg has quit [Client Quit]
< Sergobot>
awad: No problem! That was only a friendly remark :)
< zenzu>
Hello everyone, my name is Tom and I would like to work on reinforcement learning project. I am beginner in AI but I have experience in C++, ROS and robotics. I would like to combine mentioned project with robotics if it would be possible. Looking forward to work together :)
zenzu has quit [Quit: Page closed]
st_sg has joined #mlpack
st_sg has quit [Client Quit]
awad has quit [Ping timeout: 256 seconds]
awad has joined #mlpack
zenzu has joined #mlpack
zenzu has quit [Client Quit]
st_sg has joined #mlpack
st_sg has quit [Client Quit]
aman_p has joined #mlpack
aman_p has quit [Remote host closed the connection]
aman_p has joined #mlpack
aman_p has quit [Client Quit]
aman_p has joined #mlpack
st_sg has joined #mlpack
aman_p has quit [Ping timeout: 245 seconds]
st_sg has quit [Client Quit]
st_sg has joined #mlpack
st_sg has quit [Client Quit]
rob_b has joined #mlpack
< rob_b>
what algorithms would you guys recommend for time series regression?