Microsoft have recently announced the public preview of Anomaly Detector and general availability of Custom Vision. With both services, Microsoft further expands its Cognitive Services offering for its customers.
Anomaly Detector is a new service within the Azure Cognitive Services family allowing users to detect unusual patterns or rare events in their data that could translate to identifying problems. The Anomaly Detector stems from the Machine Learning Anomaly Detection API, and Microsoft itself relies on this service as Anand Raman, chief of staff, Data Group at Microsoft, states in a blog post:
Today, over 200 teams across Azure and other core Microsoft products rely on Anomaly Detector to boost the reliability of their systems by detecting irregularities in real-time and accelerating troubleshooting. Through a single API, developers can easily embed anomaly detection capabilities into their applications to ensure high data accuracy, and automatically surface incidents as soon as they happen.
With this service, the financial industry, for instance, can leverage the service to detect anomalies more accurately in credit card transactions or large money transfers. Older approaches by financial institutes were on, for example, rule engines to determine a deviation in a transaction – leading to false positives. However, according to Eric Ogren, an analyst at 451 Research, in a TechTarget article:
Anomaly detection usually requires a person, not a machine learning algorithm, to ultimately pass a verdict on whether something is indeed a problem.
The Azure Anomaly Detector is currently only available in West U.S. 2 and West Europe regions, and Microsoft plans to release the service in all regions upon general availability. Furthermore, the pricing in the standard tier is 0.157 dollars per 1000 transactions, while in the free tier customers are allowed 20000 transactions per month. Note that the costs will double on general availability.
In addition to the preview of Anomaly Detector, Microsoft has graduated their Custom Vision service to general availability. With this machine learning-powered service developers can quickly build, deploy, and improve custom image classifiers to recognize content in images. As Raman states in the same blog post:
Developers can train their own classifier to recognize what matters most in their scenarios, or export these custom classifiers to run them offline and in real time on iOS (in CoreML), Android (in TensorFlow), and many other devices on the edge. The exported models are optimized for the constraints of a mobile device providing incredible throughput while still maintaining high accuracy.
With the GA release of Custom Vision, developers can also benefit from several enhancements, such as:
- Easy integration of computer vision capabilities into applications with 3.0 REST APIs and SDKs
- The ability to export classifiers to support Azure Resource Manager (ARM) for Raspberry Pi 3 and the Vision AI Dev Kit
- An advanced training feature to specify a compute time budget, which allows the service to identify the best training and augmentation settings experimentally
An example use case of Custom Vision is from Minsur, a mining company in Peru, that uses the service for the detection of foam levels in water being treated for livestock and agriculture. The company uses the combination of Custom Vision and Azure video analytics to replace a highly manual process so that employees can focus on more strategic projects within the operation.
Lastly, the Custom Vision is available in various US, Europe, and Asia regions, and the pricing details are available on the pricing page.
