MySQL HeatWave ML fully automates the machine learning lifecycle and stores all trained models within the MySQL database, eliminating the need to move the data or model to an ML tool or service.
On top of this, Oracle has announced that Oracle MySQL HeatWave now supports ML within the database, as a complement to previously available transaction processing and analytics, being the only MySQL cloud service to do so.
“MySQL HeatWave is one of the fastest growing cloud services at Oracle. An increasing number of customers have migrated from Amazon and other cloud database services to MySQL HeatWave, and have seen significant performance gains and lower costs,” said Edward Screven, chief corporate architect, Oracle.
Machine Learning in MySQL applications
Until now, adding machine learning capabilities to MySQL applications has been prohibitively difficult and time-consuming for developers.
First, there is the process of extracting data out of the database and into another system to build and deploy ML models. This approach creates multiple silos for applying machine learning to application data and introduces latency as the data moves. It also leads to the proliferation of data outside of the database, making it more vulnerable to security threats and adding complexity for developers to program across multiple environments.
Second, existing services expect developers to be experts in guiding the ML model training process; otherwise, the model is not optimal, which degrades the accuracy of the predictions. Lastly, most existing ML solutions do not include the functionality to provide explanations as to why the models that developers build offer specific predictions.
MySQL HeatWave ML solves these problems by natively integrating machine learning capabilities within the MySQL database, eliminating the need to extract, transform, and load (ETL) the data to another service.
HeatWave ML fully automates the training process and builds a model with the best algorithm, optimal features, and optimal hyperparameters for a given data set and specified task. All models generated by HeatWave ML can provide explanations about the model and prediction.
MySQL HeatWave ML, the easiest, fastest and most affordable way for developers to add powerful machine learning capabilities to their MySQL applications
HeatWave ML Capabilities
HeatWave ML offers the following capabilities compared to other cloud database services:
- Fully Automated Model Training: All stages of creating a model with HeatWave ML are fully automated and do not require any developer intervention. This results in a fitted model that is more accurate, requires no manual work, and the training process is always complete.
- Model and Inference Explanations: Model explainability helps developers understand the behavior of a machine learning model. For example, if a bank denies a customer a loan, the bank needs to be able to determine which parameters of the model have been taken into account, or if the model contains any bias.
- Hyper Parameter Tuning: HeatWave ML implements a new gradient-search-based reduction algorithm for hyperparameter tuning. This allows the hyperparameter search to run in parallel without compromising the accuracy of the model. Hyperparameter tuning is the most time-consuming stage in ML model training, and this unique capability gives HeatWave ML a significant performance advantage over other cloud services for building machine learning models.
- Algorithm Selection: HeatWave ML uses the notion of proxy models – which are simple models that display the properties of a complete complex model – to determine the best ML algorithm for training. Using a simple proxy model, the selection of the algorithm is done very efficiently, without loss of precision.
- Intelligent Data Sampling: During model training, HeatWave ML samples a small percentage of the data to improve performance. This sampling is done in such a way that all representative data points are captured in the sample data set.
- Feature Selection: feature selection helps determine the attributes of the training data that influence the behavior of the machine learning model to make predictions.
George is Digismak’s reported cum editor with 13 years of experience in Journalism