Using sparse semantic embeddings learned from multimodal text and image data to model human conceptual knowledge (2018)