search

ECP-CANDLE / Benchmarks

ECP-CANDLE Benchmarks
MIT License
57 stars 83 forks source link

Data preprocessing: normalization #36

Open Deathn0t opened 5 years ago

Deathn0t commented 5 years ago

Hello,

I was looking at this baseline: https://github.com/ECP-CANDLE/Benchmarks/blob/release_01/Pilot1/Combo/combo_baseline_keras2.py

I was asking myself if the data are already preprocessed/normalized because I don't see any preprocessing function.

hyoo commented 5 years ago

Check ComboDataLoader and generator classes.

Sent from my iPhone

On Nov 15, 2018, at 3:27 PM, Romain Egele notifications@github.com<mailto:notifications@github.com> wrote:

Hello,

I was looking at this baseline: https://github.com/ECP-CANDLE/Benchmarks/blob/release_01/Pilot1/Combo/combo_baseline_keras2.py

I was asking myself if the data are already preprocessed/normalized because I don't see any preprocessing function.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/ECP-CANDLE/Benchmarks/issues/36, or mute the threadhttps://github.com/notifications/unsubscribe-auth/ABLd8xMHNx81XMSpf3mY_x0ASZJjNUEyks5uvdwbgaJpZM4Yi9f-.

Deathn0t commented 5 years ago

Oh yes I see, you are using the scaling argument.

so the mse compute on the validation set is using preprocessed data (evaluate_score), isn't it better to reverse the preprocessing before computing the mse between y_orig, y_pred ?