5 Keys to Using AI and Machine Learning in Fraud Detection

Views:
 
Category: Entertainment
     
 

Presentation Description

No description available.

Comments

Presentation Transcript

slide 1:

10/21/2019 5 Keys to Using AI and Machine Learning in Fraud Detection https://medium.com/fugenxmobileapp/5-keys-to-using-ai-and-machine-learning-in-fraud-detection-7356ee310396 1/4 5 Keys to Using AI and Machine Learning in Fraud Detection Fugenx technology Oct 21 · 4 min read Payment Fraud is a flexible use case for machine learning services and artificial intelligence AI and has a long track record of successful use. When customers receive a call text email or in-app message from their card issuer to verify a transaction or notify them of fraud on their card they may not even suspect that this excellent customer service is behind them. A set of algorithms. Recently however there has been much hype about the use of AI and machine learning in fraud detection making it very difficult for many people to separate myths from reality. At times you may come to the conclusion that AI and machine learning have just been invented or that it can be applied to payment fraud for the first time In this blog series I am going to explore five keys to using AI and machine learning in fraud detection. The insights here are based on FICO’s 25+ years in the field protecting billions of cards worldwide and my own experience in fraud management for the past 23 years. Shall we see Machine learning and artificial intelligence in fraud Before we get to Key 1 here is a brief definition of what we are talking about as the terms machine learning and AI are also misused. Machine learning refers to analytical methods that “learn” patterns in datasets without being guided by a human analyst. AI refers to the wider application of specific types of analytics to accomplish tasks ranging from driving a car to yes detecting a fraudulent

slide 2:

10/21/2019 5 Keys to Using AI and Machine Learning in Fraud Detection https://medium.com/fugenxmobileapp/5-keys-to-using-ai-and-machine-learning-in-fraud-detection-7356ee310396 2/4 transaction. For our purposes consider machine learning as a way to create analytical models and AI as the use of those models. Machine learning data can help scientists effectively identify which transactions are fraudulent while significantly reducing the false positives. These methods are very effective in fraud prevention and detection because they allow you to automatically detect samples in large quantities of streaming transactions. If done correctly machine learning clearly differentiates legitimate and fraudulent behaviors but over time adapts to new previously unseen deceptive strategies. Understanding the patterns in the data and applying data science to improve the ability to distinguish normal behavior from abnormal behavior is becoming increasingly difficult. This requires thousands of calculations to be done in milliseconds. Without a proper understanding of the domain as well as fraud-specific data science techniques you can easily use machine learning algorithms that learn the wrong thing resulting in a costly mistake. Just as people can learn bad habits so too can a poorly engineered machine learning model. Key 1 — Integrating supervised and unsupervised AI models into a coordinated strategy Because organized crime schemes are so sophisticated and quickly adaptable defensive strategies based on any one one-size-fits-all analytical technique yield sub-results. Each use case should be supported by specialist-designed disorder detection techniques which are suitable for the problem at hand. As a result unsupervised and unsupervised models play an important role in detecting fraud and are woven into comprehensive next-generation fraud strategies. The supervised model is the most common form of machine learning across all disciplines and it is a model trained on the greatest set of “tagged” transactions. Each transaction is tagged as fraud or non-fraud. Models are trained by taking huge amounts of tagged transaction details to find patterns that best reflect legitimate behaviors. When developing a supervised model the amount of clean relevant training data is directly related to model accuracy.

slide 3:

10/21/2019 5 Keys to Using AI and Machine Learning in Fraud Detection https://medium.com/fugenxmobileapp/5-keys-to-using-ai-and-machine-learning-in-fraud-detection-7356ee310396 3/4 Unsupervised models are designed to detect disorderly behavior in cases where tagged transaction data is relatively thin or non-existent. In these cases some kind of self- learning should be applied to surface models in the data that are not visible to other types of analysis. Chart with four boxes of different types of fraud analytics. Unsupervised models are designed to detect outliers that indicate previously unseen frauds. The majority of these AI-based methods identify behavioral disorders by identifying transactions that are incompatible. For accuracy these differences are assessed at the individual level and by sophisticated peer group comparison. By selecting the right mix of unsupervised and unsupervised AI methods you can quickly detect forms of suspicious behavior that you have never seen before but more

slide 4:

10/21/2019 5 Keys to Using AI and Machine Learning in Fraud Detection https://medium.com/fugenxmobileapp/5-keys-to-using-ai-and-machine-learning-in-fraud-detection-7356ee310396 4/4 subtle patterns of fraud that have previously been observed in billions of accounts. A good example of this is on our FICO Falcon platform with its Cognitive Fraud Analytics. Want to know more AI services then have a free visit for USM systems Articial Intelligence Machine Learning Ai Solution Ml In Frauddetection Ai Services About Help Legal

authorStream Live Help