📂Machine Learning

Using Machine Learning Datasets in Julia

Description

The MLDatasets.jl^{1 2} package allows for the use of the following datasets. Datasets with links have their usage explained in their respective documents.

Vision

• CIFAR10
• CIFAR100
• EMNIST
• FashionMNIST
• MNIST
• Omniglot
• SVHN2
• convert2image

Mesh

• FAUST

Miscellaneous

• BostonHousing
• Iris
• Mutagenesis
• Titanic

Text

• PTBLM
• SMSSpamCollection
• UD_English

Graphs

• CiteSeer
• Cora
• Graph
• HeteroGraph
• KarateClub
• MovieLens
• OGBDataset
• OrganicMaterialsDB
• PolBlogs
• PubMed
• Reddit
• TUDataset

For one-hot encoding this data or training methods, refer to the following.

• How to one-hot encode data
• How to implement MLP and train the MNIST dataset

Example

CIFAR10

julia> Train_X2, Train_Y2 = CIFAR10.traindata()

julia> size(Train_X2)
(32, 32, 3, 50000)

julia>typeof(Train_X2)
Base.ReinterpretArray{N0f8, 4, UInt8, Array{UInt8, 4}, false}

julia> size(Train_Y2)
(50000,)

julia>typeof(Train_Y2)
Vector{Int64} (alias for Array{Int64, 1})

julia> for i in 1:7
        save("CIFAR10_$i.png", colorview(RGB, CIFAR10.convert2image(CIFAR10.traintensor(i))))
        end

Picking the first 7 pictures would look like this.

Environment

• OS: Windows11
• Version: Julia v1.8.2, MLDatasets v0.7.6