Oracle Corp. today is unleashing the benefits of its own internal artificial development experience to customers with a collection of pre-defined services that can be incorporated into user-developed applications without requiring data science experience.

Oracle Cloud Infrastructure AI services have been pre-trained on business-oriented data and can be supplemented with custom training by an organization. The six discrete modules cover two domains, according to Greg Pavlik, chief technology officer of Oracle Cloud. One is perceptual services like speech recognition and computer vision. The other is decisioning enabled by observations and data.

The functions will be accessed through application program interface calls to make integration with existing or newly developed applications easier, Pavlik said. “You don’t have to understand the science,” he said. “While these are useful services for data scientists, the key here is to bring the power of adaptive machine learning to people outside of the data science community.”

Oracle is a separate cloud platform called the OCI Data Science service that is optimized for Python development and model training by data science professionals. The OCI AI Services were built on top of that infrastructure and incorporates the same technology used in the vendor’s commercial applications such as speech recognition.

“Language recognition is a good example because language training models have become very large,” Pavlik said. “We take that [training burden] off the user’s plate by providing models that are already trained for common recognition tasks. You can then customize it for your specific domain through transfer learning techniques.”

Users can also apply data labels, an important element in model training to extend functionality. For example, a base training model that recognizes cars and trucks can be extended to identify certain makes and models.

The six available modules include the following:

OCI Language performs text analysis at scale to understand unstructured text in documents, customer feedback interactions, support tickets, and social media.

OCI Speech uses prebuilt recognition models trained with thousands of speakers in multiple languages.

OCI Vision is used for image recognition and document analysis. It can be extended to specific use cases such as scene monitoring, defect detection and custom document processing as well as to spot anomalies and extract text from forms.

OCI Anomaly Detection includes representational state transfer or REST APIs and software development kits supporting several languages that can be used to integrate anomaly detection models into business applications like equipment monitoring and fraud detection.

OCI Forecasting provides explainable forecasts of business metrics based on machine learning and statistical algorithms.

OCI Data Labeling allows labeled datasets to be used for model training.

Oracle didn’t specify pricing but said the services would be volume-based and are “not meant to be unreasonable,” Pavlik said. The company’s objective is to encourage broader use of its AI services rather than to build a profitable business.

“We want to make AI boring,” said Elad Ziklik, vice president of software development at Oracle Cloud Infrastructure. “We want to infuse AI into everything you do.”

Photo: SiliconANGLE