Unstable results

[alishan2040]

Hello @raghavchalapathy ,

Thanks for sharing the implementation of OC-NN (a cool idea indeed). I tried the implementation of OC-NN from this file < https://github.com/raghavchalapathy/oc-nn/blob/master/src/models/OC_NN.py>. However, the results are quite unstable between multiple executions of the same code (keeping the same hyper-parameters and same random_seed=42). For example, execution 1: f1 0.901 execution 2: f1 0.16 execution 3: f1 0.91 execution 4: f1 0.24 and so on.

I want to confirm if this is the correct implementation or should I look into this file: https://github.com/raghavchalapathy/oc-nn/blob/master/src/models/OneClass_SVDD.py as you have mentioned this file in a separate issue

Note: I trained OC-NN on my custom data.

Your response is appreciated. Thanks,