Counter-cryptanalysis uses cryptanalytic techniques to detect cryptanalytic attacks. It was introduced by Stevens with a collision detection algorithm that detects whether a message is one of a colliding message pair constructed using a collision attack. Later, Stevens and Shumow improved the collision detection against SHA-1 by using unavoidable conditions. However, there are no results improving collision detection against MD5 due to its weak diffusion properties. In this paper, an improved collision detection algorithm against MD5 is proposed by using the 14-bit sufficient condition combinations. This leads to the dividing the 223 classes into four sets. Each element, belonging to the first two sets, holds the same sufficient condition combination. Our new algorithm can classify 126 classes efficiently. The runtime is 28.6% of the previous collision detection method.

中文翻译：

---

使用充分条件组合改进 MD5 碰撞检测

反密码分析使用密码分析技术来检测密码分析攻击。Stevens 引入了一种碰撞检测算法，该算法检测消息是否是使用碰撞攻击构造的碰撞消息对之一。后来，Stevens 和 Shumow 通过使用不可避免的条件改进了针对 SHA-1 的碰撞检测。然而，由于 MD5 的扩散特性较弱，因此没有任何结果可以改善对 MD5 的碰撞检测。本文利用14位充分条件组合提出了一种改进的针对MD5的碰撞检测算法。这导致将 223 个类分为四组。属于前两组的每个元素都具有相同的充分条件组合。我们的新算法可以有效地对 126 个类别进行分类。运行时间为 28。