Скачать или смотреть AI Abstraction and Reasoning | More Human-like AI | Kaggle Competition Part 1: Overview

"Can a computer learn complex, abstract tasks from just a few examples?" Current Machine Learning models are data-hungry, which is different from how human intelligence behaves. In this competition, we are going to create an algorithm that uses a very small training set and is able to reason about what the test output should be.

This is part 1 of a new series on the Kaggle competition Abstraction and Reasoning. In this video, I'll go over the competition objectives, what data is provided, and how we are evaluated.

Resources as mentioned in the video:
Join competition and acquire data here: https://www.kaggle.com/c/abstraction-...
Starter Notebook: https://www.kaggle.com/inversion/abst...
My version of it (modified plot task function to plot every pair in the task): https://www.kaggle.com/qcw171717/abst...
GitHub repo: https://github.com/fchollet/ARC