Fighting cyber-criminals is not an easy task, and staying ahead of them is becoming increasingly more difficult. Using traditional methodologies is no longer the ideal strategy. Cyber-criminals are devising new methods to keep track of digital trails, making it difficult for even updated security tools to detect their activities. The harsh reality, as witnessed in the past, is that organizations can no longer fully defend themselves against cyber-criminals – but they can certainly make themselves difficult targets by using the right tools and practices.
Following this example is PayPal, who is looking at the “Deep Learning” approach to fight against criminals who attempt to exploit the online payment platform, as highlighted by Gigaom. The Deep Learning approach has found effective usage in pattern-recognition tasks such as stock market prediction, sales pipeline analysis, and fraud detection. Many are calling it a new approach in the arena of machine learning and artificial intelligence. It has already seen adoption by companies such as Google, Facebook, Microsoft, and Baidu.
From a technology standpoint, the Deep Learning systems use artificial neural network algorithms. These systems effectively gather data insights and recognize patterns. They have found many uses in computer vision, speech recognition, text analysis, and more. Now these systems are turning out to be effective in recognizing the patterns and characteristics of cybercrime and online fraud as well. PayPal had already switched from machine-learning-based pattern recognition to deep learning techniques a few years back.
PayPal’s efforts towards utilizing Deep Learning systems can be seen actively in its current anti-fraud systems. The Deep Learning systems have been effective in analyzing factors such as timelines, location, etc. as part of payment transactions. PayPal is also utilizing this Deep Learning methodology to determine which fraud-detection models will be implemented. PayPal hopes to take this Deep Learning approach forward and one day also generate data driven insights in real time to curb fraud.
Source: PayPal, Gigaom
The above illustration highlights PayPal’s fraud management options for developers.
Why Deep Learning?
- Helps to unearth low-level complex abstractions
- Helps to learn complex highly varying functions not present in the training examples
- Widely employed for image, video processing and object recognition