Showing posts with label python. Show all posts
Showing posts with label python. Show all posts

Ottawa Python Authors Meetup: Artificial Intelligence with Python

>> Friday, July 29, 2016

Last night, I attended my first Ottawa Python Authors Meetup.  It was the first time that I had attended despite wanting to attend for a long time.  (Mr. Releng also works with Python and thus every time there's a meetup, we discuss who gets to go and who gets to stay home and take care of little Releng.  It depends on if the talk to more relevant to our work interests.)

The venue was across the street from Confederation Park aka land of Pokemon.


I really enjoyed it.  The people I chatted with were very friendly and welcoming. Of course, I ran into some people I used to work with, as is with any tech event in Ottawa it seems. Nice to catch up!

The venue had the Canada Council for the Arts as a tenant, thus the quintessentially Canadian art.


The speaker that night was Emily Daniels, developer from Halogen Software who spoke on Artificial Intelligence with Python. (Slides here, github repo here).  She mentioned that she writes Java during the day but works on fun projects in Python at night.  She started the talk by going through some examples of artificial intelligence on the web.  Perhaps the most interesting one I found was a recurrent neural network called Benjamin which generates movie script ideas and was trained on existing sci-fi movies and movie scripts.  Also, a short film called Sunspring was made of one of the generated scripts.  The dialogue is kind of stilted but it is interesting concept.

 After the examples, Emily then moved on to how it all works. 

Deep learning is a type of machine learning that drives meaning out of data using a hierarchy of multiple layers that mimics the neural networks of our brain.

She then spoke about a project she wrote to create generative poetry from a RNN (recurrent neural network).  It was based on a RNN tutorial that she heavily refactored to meet her needs.  She went through the code that she developed to generate artificial prose from the works of H.G. Wells and Jane Austen.  She talked about how she cleaned up the text to remove EOL delimiters, page breaks, chapters numbers and so on. And then it took a week to train it with the data.

She then talked about another example which used data from Jack Kerouac and Virginia Woolf novels, which she posts some of the results to twitter.


She also created a twitter account which posts generated text from her RNN that consumes the content of Walt Whitman and Emily Dickinson. (I should mention at this point that she chose these authors for her projects because copyrights have expired on these works and they are available on the Gutenberg project)

After the talk, she field a number of audience questions which were really insightful. There were discussions on the inherent bias in the data because it was written by humans that are sexist and racist.  She mentioned that she doesn't post the results of the model automatically to twitter because some of them are really inappropriate since these novels since they learned from text that humans wrote who are inherently biased.

One thing I found really interesting is that Emily mentioned that she felt a need to ensure that the algorithms and data continue to exist, and that they were faithfully backed up.  I began to think about all the Amazon instances that Mozilla releng had automatically killed that day as our capacity had peaked and declined.  And of the great joy I feel ripping out code when we deprecate a platform.  I personally feel no emotional attachment to bring down machines or deleting used code.
 
Perhaps the sense of a need for a caretaker for these recurrent neural networks and the data they create is related to the fact that the algorithms that output text that is a simulacrum for the work of an author that we enjoy reading.  And perhaps that is why we maybe we aren't as attached to a ephemeral pool of build machines as we are are to our phones.  Because the phone provides a sense human of connection to the larger world when we may be sitting alone.

Thank you Emily for the very interesting talk, to the Ottawa Python Authors Group for organizing the meetup, and Shopify for sponsoring the venue.  Looking forward to the next one!

Further reading

Read more...

Resources for learning Python

>> Monday, June 17, 2013

When I started working at Mozilla, I didn't know Python very well.  All of the code that drives our continuous integration is written in it so I've had a spent a lot of time becoming more proficient.  I've taken a number of free self-study classes/tutorials that others might find useful getting up to speed with Python.

Google Python class  - Two day class that describes how to manipulate strings, dictionaries, lists. File I/O, regular expressions and sorting algorithms.  Class lectures that were recorded at a Google training event are available on youtube, and the exercises are pretty straightforward with a solution key to check your work.  All the exercises are done using the Python installation on your local machine.

Code academy has a Python class.  I did the exercises for the topics that weren't covered in the Google class, including list comprehensions, lambda expressions and OO in Python.  The exercises are completed and validated on the website.

The best course I've taken so far has been the nine week Interactive Programming in Python class offered by Coursera. Coursera is a company that offers university courses for free.  You receive a grade and certification of completion,  but not an actual credit at the institution. The Python course I took is in conjunction with Rice University.  For this course, you have to register for the course when it's run and complete the work each week.   There are an hour or so of video lectures, a quiz or two and an assignment to complete for around 8-10 hours of work each week.  After you submit your assignment, you're asked to review the work of five other students.  This is how to scale marking assignments for the thousands of students enrolled in the class :-).  The code for your assignments is written in the browser,  on the CodeSkulptor website.

Syllabus for coursera course
Each week the assignment is to implement a different game which was a lot of fun.   We also learned how to implement the calculations for velocity, reflection and collisions to make the objects in our games appear as a realistic game.   The final project is to implement a version of the Asteroids game complete with missiles and exploding rocks.  I've also never done much with a the UI aspects of programming in the past, so it was interesting to learn how to incorporate these elements.
Screenshot of final asteroids game
This iteration of the course is almost over and the next one starts in October.  Essentially the course provides lots of time to practice writing code, which is the only path to getting better.  I found the course to be really well organized, and the video lectures were succinct in getting to the point of the lecture.  No profs going off on tangents about random subjects :-)

Mr. Releng also took this course at the same time I did as he is also using Python at work.  I said to him one day that I certainly wouldn't wanted to have attended all my university courses online as I always liked to ask questions in class.  With the pre-recorded lectures, the only way to ask questions is on the associated forums.  He agreed, but said while this was probably not the future of university, it was certainly great for continuing education.

Other Python resources I looked at  include http://www.learnpython.org/.  The pyvideo website includes talks from many of the PyCon and related conferences here which are interesting to watch.

What resources did you use to learn Python?

Read more...

  © Blogger template Simple n' Sweet by Ourblogtemplates.com 2009

Back to TOP