/ Machine Learning

Lagos AI hack - the first few hours

Before speaking about the hack, it is important to touch base on AI Saturdays(Lagos) - One of the communities devoted to making rigorous AI education accessible and free, using materials used at top universities. AI Saturdays is run in 50+ cities across the world to date. During the program, we learnt topics like Linear Algebra, Statistics, Deep Learning and more. The Hackathon was an opportunity for us to use our knowledge to tackle a near real world problem: Classifying Food

The First Few Hours.

I came 1 hour late for the hackathon.
We were disoriented ensuring everyone was registered.
We kicked off one hour later after everyone was sorted out into teams.
Internet was a challenge the first few mins, but that got better over time. The sweets did a good job at boosting morale.
The biggest 1st hour challenge was deciding which environment to use for the challenge.
The popular choices were

  1. Google compute engine
  2. Local machine
  3. Colaboratory: Google's free python research environment.

The Next few hours

Most people had chosen to go with colab. The challenge here was getting the data into colab. A few suggestions came out

  1. Upload the 1.4Gb file to Google Drive - that was proving dificult (Slow internet)
  2. Use the Kaggle CLI - also tricky setting up Kaggle cli to work in collaboratory
  3. Using Good ol python file download.
    A team came up with a Kernel on Kaggle for pulling the data, most people ended up using good ol' python for this task