Our Snippets Of Knowledge Weekly Blog Series
As a growing suite of tools, Microsoft Cognitive Services, now renamed to Azure AI Services, offers a pre-packaged set of capabilities that supports a wide range of data science use cases. With different pre-trained Microsoft models and endpoints invoked according to the requirements. The tools can be used out-of-the-box or customised to support specific requirements.
The advantage to this approach is that the resource wrappers offer rapid development for common operations like speech to text recognition, Open AI, vision, and language translation, amongst many other applications. This offering abstraction means that Microsoft handle the heavy lifting in terms of underlying model training and expose only the required APIs for inference by the downstream solution. However, this can come with high run costs and may limit some applications to only row-by-row results, compared to large batch data scoring of results if the equivalent model was created/trained with in-house datasets and deployed from scratch.
See MS Learn for more information on this Resource here.
We hope you found this knowledge snippet helpful.
Check out all our posts in this series here.
Comments