Credit: Dreamstime
Amazon Web Services (AWS) has placed its machine learning (ML)-based fraud detection managed service, Amazon Fraud Detector, into preview for its Sydney region, among others.
Made to detect potentially fraudulent activities online, Fraud Detector uses ML science and over 20 years of fraud detection experience from both the Amazon marketplace and AWS to create fraud detection models.
Specifically, the service can detect fraud relating to potentially high risk new account creations, keep tabs on guest checkout orders and accounts that may abuse ‘try before you buy’ policies. Additionally, online payment fraud detection has been flagged as coming soon, being able to flag suspicious only payment transactions before payments are processed and fulfilled.
The service is available in a free trial for the first two months as well as a paid version. In the trial, users get access to 50 compute hours for model training per month, 500 compute hours for model hosting per month, 30,000 real-time online fraud insight predictions per month and 30,000 real-time rules-based fraud predictions per month.
In the paid version, model training is priced at US$0.39 per hour and model hosting is priced at US$0.06 per hour.
The pricing for both real-time online fraud insight predictions and rules-based fraud predictions are based on the number of predictions generated per month. The first 400,000 real-time online fraud insight predictions for the month are valued at US$0.030 per prediction. The next 800,000 are then priced at US$0.015 per prediction and then every prediction past this is US$0.0075.
Meanwhile, the real-time rules-based fraud prediction monthly pricing starts off at US$0.005 per prediction up to the first 400,000, then the next 800,000 predictions are priced at US$0.0025 per prediction and every prediction after that is valued at US$0.00125.
In addition to being available in the Sydney region, the Fraud Detector is also available in preview in Singapore in the Asia Pacific region, North Virginia and Ohio and Oregon in the US East and West regions, respectively, and Ireland in the Europe region.
Users can sign up for the preview via the AWS website.
