2021-03-31 MAR
Journal Club 9:30-10:30
Sparse deep learning - the neurobiology clue, the programable gated array experiment
Numenta whitepaper
Deep learning networks today have accomplished a great deal but are hitting bottlenecks as they scale to more complex tasks and bigger models. Researchers attempt to break through the bottleneck by adding more compute power and training data. These enormous models consume vast amounts of power, limiting scalability and creating environmental damage. We need a new algorithmic approach to achieve breakthroughs in performance and scalability.
Additional ref - A Thousand Brains: A New Theory of Intelligence (2021) Jeff Hawkins (Author), Richard Dawkins - foreword.
Hackathon - 10:30-11:30
Quest
One doesn't solve a problem by making it harder
a solution is something that makes the problem simpler
Example - let's remove modularity from Quest
Example, removing modularity from Quest
--- file ---
URL1 vars1
URL2 vars2
URL3 vars3
--- end ---
did it work?
Stata
STATA demonstration (looking over Mustapha's shoulder)
CloudRun
Daniel's soccer, UI;
Praful's tile server
Traversing Box Files
...
Firebase
https://console.firebase.google.com, update on an old (and future?) favorite ...