September 2022 Summaries

whylogs v1.1 is an open-source data logging API that simplifies monitoring of data and ML models. It allows users to create statistical profiles of datasets, monitor for data quality, data drift, model drift, and more in Python or Java environments. The latest version includes new features such as segments, image data logging, log rotation, conditional count metrics, string tracking, and model performance monitoring. These enhancements enable users to gain visibility within subgroups of data, monitor computer vision models, handle continuous data streams, detect specific values in datasets, track string data for NLP, and track model performance. whylogs v1.1 also supports integration with the WhyLabs Observatory for a customizable monitoring experience.

Model monitoring is crucial for financial fraud classification models as it ensures their effectiveness after deployment. Financial services firms can use the WhyLabs Observatory to monitor their machine learning models, ensuring data security and privacy while scaling with the volume of processed data. The platform monitors data across various stages, detects anomalies, and sends notifications to help save time from debugging. By monitoring data and models, financial institutions can minimize losses from fraud by identifying data quality issues and changes in a data's distribution.

The Robust & Responsible AI Newsletter - Issue #3 provides an overview of the latest developments in MLOps and Data-Centric AI. It highlights WhyLabs' new monitoring system, BentoML v1.0.0 release, and the upcoming Data-Centric AI Summit. Additionally, it features open source projects such as Union.ai's UnionML, Craiyon's text-to-image AI model, and Intel's transfer of CVAT development under OpenCV umbrella. The newsletter also includes articles on navigating the AI/Infrastructure tool maze, evolving MLOps trends, and Stitchfix's platform architecture design considerations. Furthermore, it discusses WhyLabs' recent releases in AI observability, including a case study with a Fortune 500 fintech company, monitoring image data, and the whylogs Apache Airflow provider. Finally, the newsletter lists upcoming events such as Intro to Observability Workshop, ODSC West, AWS re:Invent, and NeurIPS, along with an invitation to join the Robust & Responsible AI Community on Slack.

Apache Airflow is a powerful tool for creating, scheduling, and monitoring data pipelines. However, ensuring the quality of data processed by these pipelines requires additional tools like whylogs. The integration of whylogs with Apache Airflow allows users to monitor data and machine learning processes more effectively. By using whylogs operators in conjunction with Airflow's Directed Acyclic Graph (DAG), users can create constraints validators, generate drift reports, and profile their data for enhanced reliability and efficiency. This integration helps detect potential issues early on, ensuring the accuracy of results from data pipelines.