This post consist of the notes that are based on the series of AWS SageMaker videos provided by Amazon Web Services Amazon SageMaker Technical Deep Dive Series.

Fully-Managed Notebook Instances with Amazon SageMaker

Create your Notebook Instance:

  1. Pick the right family:
    • t - tiny < m, c - computed optimized, p - GPU
  2. Pick the right size:
    • From medium to very very large.
  3. Pick the right version:
    • ml.t3.medium: 3 means the latest version in the T family

There are ~200 example notebooks in SageMaker Notebook, definitely try them out.

Built-in Machine Learning Algorithms with Amazon SageMaker

** = Distributed Training*

<> = incremental training

Type of Problem Algorithms
Classification Linear Learner *XGBoostKNNFatorization Machines
Computer Vision Image Classification <>Object DetectionSemantic Segmentation
Topic Modeling LDANTM
Working with Text Blazing Text- Supervised- Unsupervised *
Recommendation Fatorization Machines * (+ KNN)
Forecasting DeepAR *
Anomaly Detection Random Cut Forests *IP Insights *
Clustering KMeans *KNN
Sequence Translation Seq2Seq *
Regression Linear LearnerXGBoostKNN
Feature Reduction PCA Object2Vec