Lightweight Cryptography is about developing and analyzing algorithms that provably obtain security goals, such as Confidentiality and Integrity, in resource-constrained environments. In these environments, computing devices are restricted in their memory and computational capabilities, which prevents the operation of well-established cryptographic algorithms, such as the Advanced Encryption Standard. The Internet of Things is a prominent instance of such an environment that largely consists of low-cost, smallscale computing devices. In the last decade, the increasing importance of Lightweight Encryption has been reflected in the NSA publishing two Lightweight Cipher families and NIST initiating a standardization process for Lightweight Ciphers. Considering the efficient performance of these Ciphers, their practical security needs the be closely examined by employing Cryptanalysis. One type of Cryptanalysis – Differential Cryptanalysis – recently was combined with Deep Learning, which yielded powerful attacks against Lightweight (Block) Ciphers. Further, it started an active field of research that combines recent advances in Machine Learning research with cryptanalytic attacks. Due to the variety and amount of recently published ideas in the field of Machine Learning-based Cryptanalysis, it is challenging to conceive the current state of knowledge. Hence, we summarize, implement, and compare recent work in a transparent and reproducible manner. Moreover, we extend the research by successfully applying new Machine Learning models to cryptanalytic attacks and by developing a new training setup, which impacts how the models can be utilized in a partial (Round-)Key Recovery, as proposed in previous research. Further, we have implemented a lightweight SMT-solver-based search for well-suited input differences: an essential component of cryptanalytic attacks. In all of our experiments, we trained artificial neural networks on a large set of training data in a supervised setting, for which we used the GPUs provided by the Lichtenberg cluster.