Francesco Pisu
Francesco Pisu
My understanding is that explainer.shap_values(X) is the old way of using the SHAP APIs whereas now we should use explainer(X). You'll find many inconsistencies with plot APIs too. E.g.: shap.dependence_plot...
Any update on this ?
Hello Menelau, thank you very much for your time and valuable inputs. My dataset suffers from mild imbalance, but this time is the negative class (let's say absence of condition)...
-- EDIT: In my previous answer I was reporting wrong results due to some implementation errors on my part. I removed that part as I am now getting more realistic...
Is there any update on this regard ?
@gregpost do you think the [ASOCA dataset](https://asoca.grand-challenge.org/) could be integrated with CAT08 ? I didn't manage to download CAT08 but I have ASOCA. Whoever has any idea about how the...
@journey1234-liu No updates: CAT08 can't be shared anymore; ASOCA apparently had the same fate, unfortunately
@gregpost BASED. Thank you. Did you have to balance positive/negative ostia and seed patches and offset/no offset centerline patches when generating the train and val datasets?
Are you using TotalSegmentator for 3d segmentation ? Pytorch's MPS backend doesn't support 3d ops yet, hence even if TotalSegmentator is doing everything right your computation would still be slow...
@wxc-2020 How many workers for preprocessing ? nr_thr_resamp and nr_thr_saving My run crashed very recently because I spawned too many workers (31 to be precise) and I drained all 64...