My master thesis with title “Symbol Grounding in Deep Reinforcement Learning agents for non-Markovian rewards” is available here.