The challenge in learning abstract concepts from images in an unsupervised fashion lies in the required integration of visual perception and generalizable relational reasoning. Moreover, the unsupervised nature of this task makes it necessary for human users to be able to understand a model's learnt concepts and potentially revise false behaviours. To tackle both the generalizability and interpretability constraints of visual concept learning, we propose Pix2Code, a framework that extends program synthesis to visual relational reasoning by utilizing the abilities of both explicit, compositional symbolic and implicit neural representations. This is achieved by retrieving object representations from images and synthesizing relational concepts as lambda-calculus programs. We evaluate the diverse properties of Pix2Code on the challenging reasoning domains, Kandinsky Patterns and CURI, thereby testing its ability to identify compositional visual concepts that generalize to novel data and concept configurations. Particularly, in stark contrast to neural approaches, we show that Pix2Code's representations remain human interpretable and can be easily revised for improved performance.

Translating design into code can be tedious and not particularly thought-provoking--which also happens to be the criteria that makes a task ripe for automation. The Copenhagen-based startup UIzard Technologies is already on it: The company has trained a neural network to take a screenshot of a graphic interface and translate it into lines of code, effectively eliminating that part of the web design process for developers. Impressively, the same model works across iOS, Android, and web-based interfaces, and at this early point in the research the algorithm works with 77% accuracy. Last week, Tony Beltramelli, the founder and CEO of UIzard Technologies published a research paper on how the model, called Pix2Code works. The gist is this: Like all machine learning, the researchers had to train the model on examples of the task at hand.