ML Tools

Tuesday, October 22, 2019

Posted by Tushar Nitave, Dept. of Computer Science, Illinois Tech, Chicago

Use of machine learning based solutions to solve real-life problems is becoming popular not only among tech giants but also among startups. From ML-based company to software-native company everyone wants to leverage benefits of artificial intelligence. In order make this possible lots of ML tools are also developed. Here is the list of ML tools for specific tasks:

1. End-to-end platforms: DataRobot, Polyaxon, MLflow (open source), Peltarion, Petuum, Domino Data Lab, BigHead (open source), Google's AI Platform, Spell, MSFT Azure, AWS Sagemaker, Flyodhub.
2. Model deployment platforms: Kubeflow, Seldon, Algorithmia, Datmo, Atalaya.
3. Data science collaboration: Dataiku.
4. Notebooks: Deepnote, Neptune, Jupyter, Google's Colab.
5. Data pipeline building: Tecton, Pachyderm, Airflow, Google Dataflow, Amazon step functions.
6. Workflow Orchestration: Airflow (open source), Luigi.
7. Distributed programming frameworks (batch): Hadoop, Spark.
8. Distributed programming frameworks (real time): Spark Streaming, Flink, Kafka streams.
9. Version controlling for models and data: DVC, MissingLink, Verta.
10. Experiment tracking and analysis: Weights and Biases, Comet, MissingLink.
11. Data labelling: Scale, Mechanical Turk, Heartex, Deepen.
12. Data labelling: Scale, Mechanical Turk, Heartex, Deepen.
13. ML task specific end-to-end platforms: platform.ai (computer vision).
14. Trained model marketplaces: Runway (computer vision).
15. Distributed training acceleration: EngineML, Horovod (open source).
16. Model auditing and explainability: Untangle, Fiddler.

Source: Blog post: 3 business personas in ML, their challenges and opportunities.