Amazon SageMaker for Data Scientists
Amazon SageMaker for Data Scientists
Tens of thousands of data scientists use Amazon SageMaker, because SageMaker makes it easy to solve business problems using machine learning (ML). SageMaker Studio provides a fully integrated development environment (IDE) for ML so you can prepare data, and build, train, and deploy models with a single, visual experience. Overall, data science teams can be up to 10x more productive using SageMaker.
Transparency
Biases are imbalances in the accuracy of predictions across different groups, such as age or income bracket. Biases can result from the data or algorithm used to train your model. The field of machine learning provides an opportunity to address biases by detecting them in your data and model.
Detect bias and understand predictions
Amazon SageMaker Clarify provides data to improve model quality through bias detection during data preparation and after training. SageMaker Clarify also provides model explainability reports so stakeholders can see how and why models make predictions.
Collect and prepare training data
Amazon SageMaker offers all the tools you need to create high quality training data. You can easily access data from AWS and third-party data sources, label your data, automatically cleanse and transform data, and visualize data in order to engineer model features.
Prepare data for ML in minutes
With SageMaker Data Wrangler’s data selection tool, you can quickly select data from multiple data sources, such as Amazon Athena, Amazon Redshift, AWS Lake Formation, Amazon S3, and the Amazon SageMaker Feature Store. You can write queries for data sources and import data directly into SageMaker from various file formats, and use SageMaker Data Wrangler’s visualization templates and built-in data transforms to ensure data prepared will result in accurate ML models.
Data labeling
Amazon SageMaker Ground Truth makes it helps you build highly accurate training datasets for machine learning. Get started with labeling your data in minutes through the SageMaker Ground Truth console using custom or built-in data labeling workflows including 3D point clouds, video, images, and text.
Low latency feature store
Amazon SageMaker Feature Store is a fully managed repository to store, update, retrieve, and share machine learning features. SageMaker Feature Store serves the exact same features in batch for training and in real-time for inference so you don’t need to write code to keep features consistent. You can easily add new features, update existing ones, retrieve features in batches for training, and get the same features with single-digit millisecond latency for real-time inference.
Build models
After data is prepared, Amazon SageMaker provides all the tools you need to iteratively try different modeling techniques in order to evaluate their performance. You can pick different algorithms, including over 15 that are built in and optimized for SageMaker, and over 150 pre-built models from popular model zoos available with just a few clicks. Inside SageMaker Studio, you can run the models on a small scale to see results and view reports on their performance so you can come up with high-quality working prototypes.
One-click Jupyter Notebooks
Amazon SageMaker Studio Notebooks are one-click Jupyter notebooks that can be spun up quickly. The underlying compute resources are fully elastic, so you can easily dial up or down the available resources and the changes take place automatically in the background without interrupting your work. Notebooks can be shared with a single click, your colleagues get the exact same notebook, saved in the same place.
Built-in algorithms
Amazon SageMaker also offers over 15 built-in algorithms available in pre-built container images that can be used to quickly train and run inference.
Local mode
Amazon SageMaker makes it possible to test and prototype locally. The Apache MXNet and TensorFlow Docker containers used in SageMaker are available on GitHub. You can download these containers to your local environment and use the SageMaker Python SDK to test your scripts before deploying to SageMaker training or hosting environments.
Reinforcement learning
Amazon SageMaker supports reinforcement learning in addition to traditional supervised and unsupervised learning. SageMaker has built-in, fully-managed reinforcement learning algorithms, including some of the newest and best performing in the academic literature.
Train and tune models
Amazon SageMaker provides everything you need to train and tune models. You can easily manage different training runs to isolate and measure the impact of changing datasets, algorithm versions, and model parameters or take advantage of automatic model tuning.
Organize, track, and evaluate training runs
Amazon SageMaker Experiments automatically captures training input parameters, configurations, and results, and stores them as ‘experiments’. You can browse active experiments, search for previous experiments by their characteristics, review previous experiments with their results, and compare experiment results visually.
Detect and debug problems
Amazon SageMaker Debugger captures metrics in real-time so you can correct performance problems quickly before the model is deployed to production.
Managed spot training
Amazon SageMaker provides Managed Spot Training to help you to reduce training costs by up to 90%. This capability uses Amazon EC2 Spot instances, which is spare AWS compute capacity. Training jobs are automatically run when compute capacity becomes available and are made resilient to interruptions caused by changes in capacity, allowing you to save cost when you have flexibility with when to run training jobs.
Automatic model tuning
Amazon SageMaker can automatically tune your model by adjusting thousands of different combinations of algorithm parameters to arrive at the most accurate predictions the model is capable of producing saving weeks of effort. Automatic model tuning uses machine learning to quickly tune your model to be as accurate as possible.
Deploy models to production
Amazon SageMaker makes it easy to generate predictions by providing everything you need to deploy machine learning models in production and monitor model quality over time.
Automatic workflows
Amazon SageMaker Pipelines help you create, automate, and manage end-to-end ML workflows at scale using CI/CD practices. Once the workflows are created, they can be visualized and managed in SageMaker Studio. SageMaker Pipelines takes care of all the heavy lifting involved with managing dependencies between each step of the ML workflow. You can re-run complete workflows at any time with updated data to keep your models accurate, and share workflows with other teams to collaborate on projects.
Continuously monitor models
Amazon SageMaker Model Monitor automatically detects model and concept drifts and provides detailed alerts that help identify the source of the problem so you can improve model quality over time. All models trained in SageMaker automatically emit key metrics that can be collected and viewed in SageMaker Studio.
Human review
Many machine learning applications require humans to review low confidence predictions to ensure the results are correct. Amazon Augmented AI provides built-in human review workflows for common machine learning use cases.
Batch transform
Amazon SageMaker Batch Transform eliminates the need to resize large datasets for batch processing jobs. Batch Transform allows you to run predictions on large or small batch datasets using a simple API.
Multi-model endpoints
Amazon SageMaker provides a scalable and cost-effective way to deploy large numbers of custom machine learning models. SageMaker Multi-Model endpoints enable you to deploy multiple models with a single click on a single endpoint and serve them using a single serving container.