On Tue, May 31, 2016 at 7:51 AM, Christian Weilbach < whitesp...@polyc0l0r.net> wrote:
> > Almost all of the development in deep learning is done in Python, so > having to reproduce this work on a different runtime (and language) > seems non-Clojure-like for me (compared to being hosted on the JVM and > leveraging this ecosystem). deeplearning4j is already attempting it > from the Java side with the argument of distributed scaling, but > again, there is a ton of work done for the Python toolkits including > massive scaling (e.g. with tensorflow) and most research is there, so > my question would be what the actual goals for a Clojure stack would be? Good question. Basic answer is that we are in the early stages of discovering the right abstractions for these systems; Clojure provides a strong platform for exploring the basic issues. The biggies for me are: 1. Model composition. Most of these systems have some flavor of compilation, often with a heavy dose of mutability in the construction of the graph. I'd argue its easier and simpler to explore the design space starting from immutable data representations. Given the speed of evolution in model architectures, this seems pretty important. 2. Introspection & interactivity. With Clojurescript (and even just Clojure) we have excellent options for creating interactive UIs on top of the models. Yes, theres IPython, but thats not a really equivalent to the power provided by Clojurescipt. Now, I'm all in favor of not reinventing the wheel, thats why I'm wondering what the best available foundation would be. JyNI is quite interesting and I hope it takes off. But just having a wrapper on a high-level Python lib probably isn't worth the hassle, unless you get some fundamentally new leverage. Poking around I actually discovered something closer to what I want: https://github.com/dmlc/mxnet Its pretty much the only DL platform intended to be consumed as a library, from any language. And there are already sensible JVM bindings to it. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en --- You received this message because you are subscribed to the Google Groups "Clojure" group. To unsubscribe from this group and stop receiving emails from it, send an email to clojure+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
