Data Scientist

POS terminals smart anomaly detector

Allan Costa,

All CloudWalk customers have an out-of-the-box artificial intelligence algorithm learning their transaction pattern for a sole purpose: to notify when something deviates too much from what it expects.

How can you benefit from such an algorithm? One of the main practical uses of our algorithm is to catch bugs in your backend endpoint or other infrastructure problems that reflect in your POS messages. For a concrete scenario, suppose your intended application flow is to send a sell payload to your backend endpoint with the selling details and to receive back an approval payload. If for some reason your database is unavailable and the POS receives an error message, our algorithm will probably flag this transaction as anomalous and you will be notified of it.

Data Clustering

It just works

Our machine learning algorithm works completely unsupervised, which means neither you nor we have to do anything. It just works. Another nice feature is that it adapts as it sees more transactions. In fact, we have the same algorithm running for dozens of customers and from each one it learns their specific patterns automatically.

Sounds good? So go ahead and set up your notifications at CloudWalk Manager. You can add an email to receive notifications or, if you are a Slack user, we can send the notifications directly to a Slack channel of your choice. For more information, please refer to the tutorial in our docs.

Coming soon

If you liked this service, keep in touch as it will get even better. Right now we are polishing our latest machine learning algorithm that builds upon state-of-the-art Recurrent Neural Networks to learn even better how to detect transactions deviating from your pattern. It will automatically start to learn from your transactions as soon as we release it!

About the author

Allan is our Machine Learning expert and is passionate about what he does. His time is divided between studying and applying new algorithms and the writing of his Master’s dissertation in Mathematics. He still finds time to play the piano, read science-fiction and study Physics for fun.