ModelOp Center 2.4 - Actionable Monitoring
AI/ML models need to be monitored continuously to ensure that they are operating within the business and risk constraints set by model risk/compliance, data scientists, and the lines of business. Anthony Sepci, Partner - Risk Analytics, KPMG
CHICAGO (PRWEB) June 30, 2021
ModelOp, the pioneer of ModelOps software, today announced actionable monitoring that enables organizations to continuously optimize model performance and enforce risk and compliance controls--across all types of models, all model execution environments, and on-premises, cloud, and hybrid/multi-cloud.
Al/ML models are typically monitored for performance and quality issues like drift, availability, and interpretability – but that is not enough. ModelOp Center’s actionable monitoring also enforces risk controls and immediately initiates remediation actions. Remediation actions are orchestrated to ensure model issues are fully resolved with minimal interruption to the business and resolutions are fully documented for auditability.
A comprehensive set of out-of-the-box monitors for detecting and enforcing risk and compliance controls ensure models are constantly running in a compliant state, using model-specific thresholds.
“With the increasing complexity and variety of models that are used for critical business decisioning in regulated industries, enforcing risk, business, and compliance controls is essential,” says Dave Trier, Vice President of Product at ModelOp. “And automated remediation allows problems to be resolved quickly and with complete reproducibility and auditability, ensuring models are online, compliant and continuing to drive the revenue contribution for which they are intended.”
The new v2.4 release of ModelOp Center includes:
- Comprehensive set of out-of-the-box monitors for model performance, drift, ethical fairness, and risk monitors for measuring population and characteristic stability
- Ability to enable model monitoring across all model execution environments--open source, cloud, proprietary--to provide a single enterprise monitoring solution
- Automated monitoring execution with customizable orchestration of model-specific remediation actions
- Intelligent threshold monitoring that analyses a collection of thresholds instead of a single threshold for problem determination
- Traceability of any issues from time of detection to time of resolution
- Ability for 2nd line risk teams to set thresholds based on policies, approvals and conditions
“With AI/ML models, it is no longer sufficient to check in on models at annual or quarterly reviews. AI/ML models need to be monitored continuously to ensure that they are operating within the business and risk constraints set by model risk and compliance teams, data scientists, and the lines of business,” says Anthony Sepci, Partner-Risk Analytics, KPMG.
ModelOp Center automates the governance, monitoring and orchestration of AI and analytic models across platforms and teams, resulting in reliable, compliant and scalable AI initiatives. Models are monitored and governed throughout their life cycle to ensure complete auditability, visibility and reporting for any and all production models.
Additional resources:
- ModelOps Made Easy with ModelOp Center video
- Ethical Fairness with ModelOp Center video
- Model Monitoring: The Path to Reliable AI
- ModelOp Center v2.4 overview
About ModelOp:
ModelOp, the pioneer of ModelOps software, enables enterprises to address the critical governance and scale challenges necessary to fully unlock the transformational value of enterprise AI and Machine Learning investments. Core to any AI orchestration platform, G2000 companies use ModelOp Center to govern, monitor and orchestrate models across the enterprise and deliver reliable, compliant and scalable AI initiatives.