Squashing activation functions in benchmark tests: towards eXplainable Artificial Intelligence using continuous-valued logic

Open in new window