I was able to squeeze in the beta AWS ML exam the week before Christmas. Given that it was several weeks ago, some of the other resources on Medium may be more informative, but I’ll throw my two cents out here for anyone who may be interested.
Generally speaking, know about different types of machine learning models (particularly those supported by SageMaker) and in what sorts of situations they’re applicable. These include:
- binary classification
- multiclass classification
- logistic regression
Other things you should understand:
- k-folds cross validation
- labels
- features
- splitting data up for training and evaluation
- the two options provided by Amazon ML for pre-arranged splitting
- sequential split
- random split
- the two options provided by Amazon ML for pre-arranged splitting
- all of SageMaker’s inbuilt algorithms – there were quite a few questions covering XGBoost and RCF
- deep learning concepts
- RNNs
- CNNs
- LSTMs
- regularatization
- networking concepts as they relate to ML services (mostly SageMaker endpoints)
Those are the high points. Needless to say, SageMaker is the most exclusively-covered service, but you should be familiar with services like Rekognition, Transcribe, Translate, Comprehend, and Lex as well.
Good luck!