Machine-learning techniques have been considered in many application domains. The upcoming year, my project will focus on applications in IoT (Federated Learning), education, and efficient matrix multiplication. This includes: 1. Making Federated Learning more privacy-preserving, by resisting model inference and/or extraction attacks. This project ties in with previous work on personalized Federated Learning. 2. Automated curriculum development for cybersecurity MSc students (a continuation of work done in last year's project). The end goal is to release an Open Access tool that can be used by European Cybersecurity MSc students or curriculum developers to compose a cybersecurity curriculum that fulfills needs expressed by the workforce. 3. Reducing the number of additions in Strassen Matrix Multiplication schemes using Reinforcement learning (a continuation of work done in last year's project). Naive solutions perform Greedy Search or other heuristic search methods, resulting in non-optimal matrix multiplication schemes. We expect that Monte-Carlo Tree Search methods will outperform naive solutions.