Welcome to the Kaggle subgroup of TensorFlow NW!
Purpose
This subgroup is for members of the TensorFlow NW meetup interested in applying/testing/honing TensorFlow tools on Kaggle competitions.
About TensorFlow NW
TensorFlow NW is a Portland, OR, area Meetup. The group is for anyone who's interested in learning about or advancing their knowledge of TensorFlow's use cases and applications.
About Kaggle
Kaggle is a place for data scientists and machine learning engineers of all levels to hone their craft though solving real-world problems on hundreds of datasets. The platform also hosts competitions where data scientists can work in teams to apply state-of-the-art technologies on current business problems.
And the best part is, Kaggle supports running Python notebooks and R markdown files in browser so you don’t have to wrestle with libraries, dependencies, or finicky source code. Check out this intro video to learn more.
Getting started
This subgroup is designed for those who are ready to jump right in and compete. If you are new to Kaggle, Python, or TensorFlow, we would still love for you to participate and join (but you'll have to work on your own to catch up).
Here’s what we suggest:
- Request to join the TensorFlow NW slack workspace by contacting Bryan Holman at bryanpholman [at] gmail [dot] com. Once you are given access, introduce yourself to the other members of the group!
- Register on kaggle.com and check out the site!
- Get introduced to Kaggle Kernels!
- Enter your first Kaggle competition, recognizing hand-written numbers with the classic MNIST dataset. We have prepared this guide for entering the competition, building your first kernel, and submitting predictions.
If any of these steps is beyond your capabilities, please see the resources section below. Also feel free to reach out on the slack workspace, and come to the subgroup meetings! We meet the 1st and 3rd Saturdays of every month at The Tech Academy Portland.
Resources
While TensorFlow was originally developed for Python, APIs/interfaces exist for R, Java, and more! But as a group, we typically work in Python.
If you are new to python, TensorFlow, machine learning or deep learning, please do not worry! Everyone was new to these technologies at one point. Fortunately there are hundreds of hours of amazing content online, taught by field experts, that can be accessed for very little cost (or even free!).
If you are new to Python, we recommend the following:
- A (very) brief, but free crash course at learnpython.org.
- MIT's Introduction to Computer Science and Programming Using Python.
- Jose Portilla’s Python for Data Science and Machine Learning Bootcamp on Udemy (typically less than $20).
- Mode Analytics free Python tutorial.
- Codecademy's introduction to Python.
If you are new to TensorFlow:
- Google’s Machine Learning Crash Course with TensorFlow (free).
- Google's TensorFlow tutorials (free).
- Stanford's cs20: TensorFlow for Deep Learning Research provides course slides, lecture notes, and code-alongs for free.
New to machine learning? Check out:
- Google’s Machine Learning Crash Course with TensorFlow (free).
- Andrew Ng’s machine learning course (free).
New to deep learning?
- Andrew Ng’s deep learning specialization ($50 per month).
- Fast.ai, which leverages the PyTorch framework instead of TensorFlow but is free!
- Stanford’s cs231n YouTube lectures give an in-depth look at convolutional neural networks and computer vision.
- Stanford’s cs224n Youtube lectures provide an extensive overview of recurrent neural networks and natural language processing.