分组密码算法作为对称密码算法的一个重要分支，在计算机通信和信息系统安全领域具有广泛应用。虽然经过近三十年的发展，分组密码分析理论已渐趋成熟，但仍存在一些未解决的问题：一方面，密码研究人员一直朝着“构建不依赖理想假设的新模型，从而完全彻底地掌握算法安全性”的终极目标不断努力，但差距仍存；另一方面，人工智能在各行各业对传统计算方式的冲击使得我们开始思考，其中算法能否为密码分析带来突破性进展。本项目围绕上述问题开展研究。首先，将密钥生成方案纳入模型构建过程，构造在轮密钥相关条件下仍然适用的线性壳相关度分布评估模型，更新已有的线性和差分线性分析理论框架。其次，探索轻量级分组密码算法在人工智能时代下的安全性，使用机器学习算法挖掘潜在的非随机统计特征，构建新型分析模型。最后，将数学问题求解器与深度学习相结合，开发功能更强的分组密码算法自动化分析工具，在大分组、长轮数搜索任务中取得显著成效。

As a crucial branch of symmetric-key ciphers, the block cipher has extensive use in the field of intercomputer communication and information system security. Although the past three decades witnessed that the cryptanalyses of block ciphers have come of age, there remain some unsolved issues. On the one hand, the ultimate goal in cryptanalysis is to construct new models which are independent of ideal assumptions and get a thorough understanding regarding the security of the cipher. Even though the researchers keep striving towards this goal, the gap is still striking. On the other hand, given the impact of artificial intelligence (AI) on traditional calculation methods in almost all walks of life, we start to consider if the algorithms in AI can bring about a breakthrough in cryptanalysis. This project will carry out researches centred on these problems. Firstly, based on our previous work, we will integrate the key-schedule into the constructing phase of the new model for the correlation distribution of linear hull. The new model is valid even when the round keys are dependent and will be utilised to update the current frameworks of linear and differential-linear cryptanalyses. Secondly, we will explore the security of lightweight block cipher in the era of AI. Novel cryptanalytic models will be constructed with the nonrandom statistical feature identified by machine learning algorithms. At last, we will combine mathematical solvers with deep learning methods to develop a more powerful automatic tool for block ciphers. The new tool will make remarkable progress in the searching tasks with large block sizes and a large number of rounds.