What type of vulnerabilities do statistical attacks often exploit?

Statistical attacks primarily exploit vulnerabilities related to the mathematical and algorithmic aspects of software, which includes floating-point errors and issues in random number generation. These types of vulnerabilities can lead to unexpected behavior and vulnerabilities that attackers can leverage to gain unauthorized access or extract sensitive information.

For instance, in the context of cryptography, if a system generates cryptographic keys using poor random number generation techniques, an attacker may be able to predict or uncover the keys through statistical analysis. Similarly, floating-point errors can arise when computations don't produce expected results due to precision issues, which can be exploited to manipulate outcomes or reveal secret information inadvertently.

The other choices focus on different areas of vulnerability. Poor usability in software relates more to user experience and interface issues rather than statistical exploitation. Weak user passwords involve human factors and authentication methods rather than mathematical weaknesses. Network configuration settings pertain to the communication and architecture of systems, which do not directly relate to statistical vulnerabilities in the manner that floating-point errors and random number generation do.