Mavrick use of ConvNets

I am exploring the use of Conv Nets for other problems (not only image recognition and NLP) and I was wondering if anyone has come across any resources or papers where ConvNets are used to solve other issues. The idea is that ConvNets are able to leverage certain properties of the data that normal neural networks can’t do as well. In particular the architecture of the data. I am having difficulty finding resources and just iteratively trying stuff takes a lot of time. Any help would be appreciated.