MacPaw Unveils Real-Time Phishing Detection System for Apple Mac Devices

TapTechNews July 13th news, the network security company MacPaw demonstrated a latest research result, a real-time phishing detection system that can run locally on Apple Mac devices, enhancing the network security of Mac users.

MacPaw senior research engineer Ivan Petrukha attended the 14th International Symposium on Security in Sociotechnical Systems on July 12th and will announce and introduce the research results of this system.

This system was developed by the company's Moonlock department, based on a local AI machine learning model, and uses a visual content analysis method to detect phishing websites in real time.

Studies claim that this system is optimized for the Apple macOS system. On the M1 chip, the CPU occupancy rate is tested at 16%, and the memory occupancy rate is 84 MB. In the data set test containing 50,000 pages, the accuracy (Precision) rate is 95.7%, and the recall rate is 87.7%.

TapTechNews queries public information to briefly explain the difference between accuracy and recall rate:

The formula for precision is P = \frac{TP}{TP + FP}, which calculates the ratio of all correctly retrieved items (TP) to all actually retrieved (TP + FP). In plain language, it is how many of the ones you predicted as positive examples are correct, and the base of precision is the retrieved samples.

2. The formula for recall is R = \frac{TP}{TP + FN}, which calculates the ratio of all correctly retrieved items (TP) to all items that should be retrieved (TP + FN). In plain language, it is how many of the positive examples in the sample your prediction is correct, and the base of recall is all positive samples in the sample.

3. The accuracy is for all our original samples, which indicates how many of all samples are accurately predicted.