Who am I?
● Backend Engineer, Data Team at Compass
● Python
● Deep Learning
Slide 7
Slide 7 text
No content
Slide 8
Slide 8 text
Let’s dive in!
Slide 9
Slide 9 text
No content
Slide 10
Slide 10 text
Deep
Learning
Slide 11
Slide 11 text
No content
Slide 12
Slide 12 text
No content
Slide 13
Slide 13 text
No content
Slide 14
Slide 14 text
$ gem install rails
$ rails new blog
$ cd blog
$ bin/rails server
$ pip install Django
$ django-admin startproject
blog
$ cd blog
$ python manage.py runserver
Slide 15
Slide 15 text
No content
Slide 16
Slide 16 text
No content
Slide 17
Slide 17 text
Magic!
Slide 18
Slide 18 text
Deep
Learning
Slide 19
Slide 19 text
No content
Slide 20
Slide 20 text
Common roadblocks
What about the math?
Having to learn everything about everything before you even let yourself start
Thinking you can’t do this because you don’t have a PhD
Slide 21
Slide 21 text
No content
Slide 22
Slide 22 text
the fast.ai formula:
Slide 23
Slide 23 text
Math is important
Theory is important
But why not start with the knowledge and
skills that you have?
You know how to translate ideas into
working code.
Slide 29
Slide 29 text
No content
Slide 30
Slide 30 text
You know how to read the
documentation and source of libraries.
Slide 31
Slide 31 text
No content
Slide 32
Slide 32 text
You’ve probably spent hours of your life
debugging.
Slide 33
Slide 33 text
No content
Slide 34
Slide 34 text
Putting it together
Slide 35
Slide 35 text
No content
Slide 36
Slide 36 text
Kaggle Kernels
Slide 37
Slide 37 text
No content
Slide 38
Slide 38 text
fast.ai library
Slide 39
Slide 39 text
Communities
fast.ai forums
Pytorch forums
Kaggle discussion threads
AI Saturdays
Medium
Twitter
Slide 40
Slide 40 text
http://tools.google.com/seedbank/
Slide 41
Slide 41 text
What can I do?
Slide 42
Slide 42 text
Near term questions
What happens if I train for longer?
What happens if I change the learning rate?
Can I apply this to another dataset?
Can I use a different architecture?
What is training loss? What is validation loss?
Slide 43
Slide 43 text
Understand, implement, and explain
https://medium.com/@radekosmulski/do-smoother-ar
eas-of-the-error-surface-lead-to-better-generalization-
b5f93b9edf5b
http://teleported.in/posts/cyclic-learning-rate/