Machine learning is technology where computers identify patterns in data. It has revolutionized areas like spam detection, voice recognition, and digital advertising. Credit card companies also use machine learning to determine when a card may have been stolen. It uses data from different sources to train computers to recognize patterns and correlate information. The goal of machine learning is to solve problems more efficiently, accurately, and objectively than a human could. 

The gold-standard technology for fraud detection, machine learning is a critical part of the fraud-detection systems at ecommerce pioneers like Amazon and Google though it was previously out of reach for other e-commerce companies. Sift strives to make fraud detection powered by large-scale machine learning accessible to all. 

We've also seen machine-learning called "self-learning" and "predictive analytics." 

You can read more about machine learning on our blog (Machine Learning for Poets) or in our e-book The Future of Fraud Fighting.

Why should it be done at a large-scale? 

• Millions of fraud patterns:  Sift’s fraud library has millions of relevant fraud patterns, and that number grows every day. 
• Unconventional rules: Sift's large-scale infrastructure allows us to surface granular, client-specific fraud signals, such as the specific product sold or the text of a gift message 
• Difficult to evade: Some fraudsters evade conventional fraud checks by disabling javascript or even disguising their IP to match the stolen billing address. Sift's large-scale machine learning lets us find other signals. In fact, these acts of evasion become signals into themselves.