ML demo code

Specific Domains Machine Learning
1

I am new to Machine Learning in Julia and am going through the tutorial here. The code for the tutorial is here.

While trying to run the following line I am getting an error:

Metalhead.imagenet_classes[rand(1:1000,1,1)]

Error:

UndefVarError: imagenet_classes not defined

Any suggestions please?

Also are there any other good tutorials for machine learning and deep learning in Julia?

Thanks!!!

2

You need to add the Metalhead.jl package, and then add the line using Metalhead

3

I have already done that. I am still getting the same error!
Btw I am on Julia 0.6. Is that causing the problem?

4

It looks like the tutorial is broken or is working inside JuliaBox only.

I get the same error with Julia v1.0.1. However, I think that we have to use Metalhead.ImageNet.imagenet_labels instead of Metalhead.imagenet_classes.

Also are there any other good tutorials for machine learning and deep learning in Julia?

Unfortunately, most Julia tutorials still suffer from changes that are caused by Julia v1.0+.

5

Yes, perhaps we should have a Doctober or Upgradadoc to root out all those vestigial relics of v.0.6-isms.

6

Updating documentation is one thing. First, packages have to work with Julia v1.0+. There are many packages that look entirely deprecated, even inside Julia organizations. I started to list data science, machine learning and AI packages that still work and are not completely deprecated:
https://www.simonwenkel.com/2018/10/05/Julia-for-datascience-machine-learning-and-artificial-intelligence.html

1 Like
7

I don’t have permission to access that page.

8

Let’s not “root out” v0.6 docs/packages/etc. Some of us are stuck with v0.6 for a while.

1 Like
9

No, i think we should remove all obsolete and non-functional example code from 1.0 tutorials and documentation. You should be OK with earlier revisions should you choose to remain on an earlier system.

10

I think versioned documentation is appropriate. We already have it. Just the v1.0+ docs need to be updated.

We agree.