Get the latest tech news

Lessons learned from manually classifying CIFAR-10 (2011)


Musings of a Computer Scientist.

Sometimes you can clearly see the animal or object and classify it based very highly-informative distinct parts (for example, you find ears of a cat). My impression from this exercise is that it will be hard to go above 80%, but I suspect improvements might be possible up to range of about 85-90%, depending on how wrong I am about the lack of training data. I have trouble exactly articulating what I learned, but overall I feel like I gained more intuition for image classification tasks and more appreciation for the difficulty of the problem at hand.

Get the Android app

Or read this on Hacker News

Read more on:

Photo of lessons

lessons

Photo of CIFAR-10

CIFAR-10

Related news:

News photo

Why Tech Companies Are Not Your Friends: Lessons From Roku

News photo

Let’s not make the same mistakes with AI that we made with social media | Social media’s unregulated evolution over the past decade holds a lot of lessons that apply directly to AI companies

News photo

A Year Later, Lessons From Silicon Valley Bank’s Epic Collapse