In the recent times, automatic detection of human personality traits has received a lot of attention. Specifically, multimodal personality trait prediction has emerged as a hot topic within the field of affective computing. In this paper, we give an overview of the advances in machine learning based automated personality detection with an emphasis on deep learning techniques. We compare various popular approaches in this field based on input modality, the computational datasets available and discuss potential industrial applications. We also discuss the state-of-the-art machine learning models for different modalities of input such as text, audio, visual and multimodal. Personality detection is a very broad topic and this literature survey focuses mainly on machine learning techniques rather than the psychological aspect of personality detection.