You can deploy this model and start earning money today!

Higgs32/
tornet-ml-higgins

Name: tornet-ml-higgins
Availability: InStock

Views

No views yet

NOTE: ALL ten folds are in the ensemblev8+v9.zip, the five are in the zip for v8 file. A like would be appreciated!

Model Name	Threshold	Model parameters	AUC	AUCPR	BinaryAccuracy	TruePositives	FalsePositives	TrueNegatives	FalseNegatives	Precision	Recall	FalseAlarmRate	F1	ThreatScore
baseline	0.0101	4665409	0.8742	0.5349	0.9456	915	635	28841	1076	0.5903	0.4596	0.4097	0.5168	0.3484
WRN	0.4444	229759	0.882	0.5498	0.9456	973	693	28783	1018	0.5840	0.4887	0.416	0.5321	0.3625
gated	0.5353	514905	0.8862	0.5574	0.9436	976	761	28715	1015	0.5619	0.4902	0.4381	0.5236	0.3547
WRN+Gated	0.4898	744664	0.8928	0.5705	0.9474	975	640	28836	1016	0.6037	0.4897	0.3963	0.5408	0.3706
PReLU_textbook	0.4242	102524	0.8976	0.5853	0.9481	1032	673	28803	959	0.6053	0.5183	0.3947	0.5584	0.3874
old_ensemble	0.3899	512620	0.9047	0.5947	0.9479	1020	669	28807	971	0.6039	0.5123	0.3961	0.5543	0.3835
ensemble_V8TH	0.3999	1050714	0.911	0.6179	0.9497	1068	659	28817	923	0.6184	0.5364	0.3816	0.5745	0.403
ensemble_v9	0.46	1379898	0.9115	0.6143	0.9492	1038	644	28832	953	0.6171	0.5213	0.3829	0.5652	0.3939
ensemble_v8+v9	0.4353	2430612	0.9128	0.6195	0.9504	1050	621	28855	941	0.6284	0.5274	0.3716	0.5735	0.402

Name	Params	fold 1	fold 2	fold 3	fold 4	fold 5	Average
baseline	4.5M	0.5121	0.4405	0.4647	0.4544	0.4451	0.4634
WRN	230k	0.6184	0.5546	0.5986	0.6152	0.6072	0.5988
Gated	600k	0.6401	0.5516	0.5247	0.5878	0.5716	0.5752
PReLU	102k	0.6314	0.6210	0.6006	0.6266	0.6132	0.6186
ensemble_v8		0.6338	0.6359	0.6141	0.6342	0.6211	0.62782
fold 1 script cross val	~300k	0.6627	0.6312	0.6171	0.6488	0.6187	0.6357

Instructions on evaluation ensemble usage can be found at: https://github.com/Higgs32584/tornet-Higgins Instructions for downloading the data can be found at: https://github.com/mit-ll/tornet