Before we understand this, we would know few words about Machine learning Algorithms, Machine learning algorithms can be used in various learning algorithms, such as supervised learning, unsupervised learning, or both. Supervised learning algorithms use labeled data to learn patterns and make predictions, while unsupervised learning algorithms identify anomalies or clusters with the data without pre-existing labels. Multiple models can be trained simultaneously to capture different aspects of suspicious activity.

Binance Machine learning Engineers typically use two types of pipelines. They are Batch and Streaming.

Batch: is used for large volumes of data

Streaming: Mean Data in real time as they’re collected. This makes situations that require a near-instant response, like detecting Hacker before withdrawing funds from any account.

Above both pipelines very important. Batch are best at handling large amounts of data, while Streaming is better for providing real-time response.

Suppose of fraud prevention, it needs to prioritize real-time data to avoid a situation called “Model Staleness.”

Impact of Staleness

If people don’t stay updated with the latest information or techniques, machine learning models can also become less accurate. According to this situation I would prefer everyone to keep yourself always updated with information or techniques.

Account Takeover (ATO) Model.

ATO Model train to identify accounts that illegitimate user hijacked with malicious objective. Then this model measures the number of transactions which are made in the past minute.

Hackers follow these steps.

1.     Sequential Pattern

2.     High number of operations (Withdrawals in a short period of time)

In this condition, Binance System calculate this feature as soon as possible in case of potential threats. It means minimizing delays between user action and user activity data is processed through this model.

For Further Information visit

https://engineering.linkedin.com/blog/2022/near-real-time-features-for-near-real-time-personalization

Batch Computing Role:

Importance of feature staleness can depend on the model. Some features, for instance, are relatively stable. In the ATO case mentioned above, it would also need to retrieve data on the user’s withdrawals in the past 30 days to calculate a ratio based on their most recent transactions.

In this situation, Batch computing over longer time periods, such as daily or hourly intervals, is acceptable despite the higher staleness resulting from waiting for data to arrive in data warehouses and for batch jobs to run periodically.

In this article some data fetched from Binance Blog, therefore if you want to know more in details just visit Binance Blog. #azuki #pepe #crypto2023 #DYOR