Tune two deep learning models and compare performance against other traditional ML algorithms -- 3
Budget: $30 – $250 CAD
We have two deep learnings models that have already been created in PyTorch:
1. One multitasking feed forward model
2. One feed forward model with one task head
I need you to:
1. Tune the hyperparameters of these models with a Bayesian approach
2. Demonstrate the performance of this model without PCA (currently PCA is used) through confusion matrix, training times, training validation loss and also AUC ROC for the classifiers
3. Use a feature reduction model that preserves feature importance unlike PCA and assess its performance
4. Compare its performance against classical ML approaches including: extra tree, KNN, random forest, SGD classifier and SVM, generating the same output metrics of performance
1. One multitasking feed forward model
2. One feed forward model with one task head
I need you to:
1. Tune the hyperparameters of these models with a Bayesian approach
2. Demonstrate the performance of this model without PCA (currently PCA is used) through confusion matrix, training times, training validation loss and also AUC ROC for the classifiers
3. Use a feature reduction model that preserves feature importance unlike PCA and assess its performance
4. Compare its performance against classical ML approaches including: extra tree, KNN, random forest, SGD classifier and SVM, generating the same output metrics of performance