Learn why Two Sigma standardized on our optimization solution to scale their research efforts. Read More
Optimize Machine Learning Models for Production
Putting models into production is one of the greatest challenges for data science teams, and hyperparameter tuning is one of the main barriers to that process. Our optimization solution accelerates the model development process for machine learning practitioners and enables experiment reproducibility during that process. Even after being put in production, models can experience drift as data changes, and SigOpt allows data science teams to build a robust process around continuously optimizing models.
Seamless Integration into Your Model
Practitioners integrate SigOpt into their model either via one of our supported client libraries or directly through the REST API. SigOpt integrates seamlessly with any of the popular machine learning and includes a dedicated integration for scikit-learn.
Some Supported Languages
Some Supported Frameworks
Advanced Features for Machine Learning
The SigOpt optimization engine includes numerous advanced features specifically designed to supercharge the optimization of machine learning models.
- Conditional Parameters enables simultaneous model selection and hyperparameter tuning.
- Multimetric Optimization facilitates the exploration of two distinct metrics simultaneously.
- High Parallelism allows customers to run parallel jobs on their compute cluster to obtain results faster.
Track and Analyze Experiments
Experiment reproducibility is a big problem for data scientists and machine learning practitioners. As modelers explore new datasets, perform feature engineering, and test different techniques, it is important to track the exact hyperparameters (and other metadata) that led to certain results. Metadata reporting and the experiment insights dashboard allow users to keep track of their experiments and understand insights from their results.