Thinking Out Loud for a presentation...

Only two days left to put in a proposal for, I'll just brainstorm some topics here:

  • Neural Networks
    • Crash Course -- introduction to Neural Networks for traditional programmers (no "big" math), what is back prop, what is an auto-encoder (what is drop-out), what is LSTM(RNN); each in code, with preferably *just* numpy or similar libraries (i.e. no Theano)
    • Practical Application -- build a trivial tool for processing something with e.g. Theano (or more likely a higher-level library), I *want* to do an LSTM doing basic voice recognition, but maybe something more along the face-recognition side would be more reasonable
    • What do We Need to Build Skynet? -- where are we currently and what are the missing pieces to build something that will take over and kill us all?
  • Introductory
    • Socket Programming
    • Twisted
    • Profiling (I'm a bit tired of this one)
    • Cython
    • Ctypes
    • Graphics/GUI/Thick Clients
      • Pyside/PyQt (likely not, doesn't seem like most people are going to be interested)
      • wxPython (thinking no, again, not a big interest area)
      • PyOpenGL (again, likely not, not a hot topic)
      • Pygame
    • Fabric (no, not likely all that interesting a topic)
    • Handling XML (with lxml)
    • Gstreamer (video processing)
  • Teaching
    • Python and Pygame for grades 4-8 under the TDSB
  • Language Details
    • Metaclasses
    • Descriptors
    • AST
  • Practicality
    • Performance
    • Fussy Firmware Packager
    • Inotify (implementation in ctypes)
    • Multicast Meetups -- make your machines connect automatically to local resources
    • Decoupling Processes -- creating more robust architectures by isolating functional units (microservices)
    • Tracking Time and Invoicing -- gluing together custom invoicing solutions from plain-text files
  • Having Fun
    • Dumb Ideas by the Dozen -- exploring problem spaces without worrying about failure
    • Denormalization in Code -- don't repeat yourself, Ancient Greek vs. Roman ideals of Beauty
    • Dodging Silver Bullets -- how to recognize impossible promises before you start the project
    • Saying No -- how saying yes leads you down the dark and dangerous paths; how dark and dangerous paths let you grow as a human being; how growth is sometimes overrated
    • Building Hydras -- connecting Python to random things as a strategy for world domination. Being the "second best language" for everything


Comments are closed.


Pingbacks are closed.