@inproceedings{Dan2020,Author = {Daniel Lünemann and Maher Fakih and Kim Grüttner},Title = {Capturing Neural-Networks as Synchronous Dataflow Graphs},Year = {2020},Booktitle = {Methoden und Beschreibungssprachen zur Modellierung und Verifikation von Schaltungen und Systemen (MBMV) 2020},type = {inproceedings},Abstract = {Machine learning (ML) algorithms have recently become a more and more common and powerful tool in technology and science alike. Capturing ML algorithms in a Model of Computation (MoC), like the synchronous data-flow (SDF), allows the application of system level design methods. This in turn, enables a Design Space Exploration (DSE) process and the synthesis of such applications as software or hardware components on heterogeneous target systems. In this paper, the translation of neural network (NN) classifiers, most commonly used in the machine learning domain, to synchronous dataflow graphs (SDFGs) is presented. Experiments showed that our translation is correct, where all automatically generated SDF classifiers have correctly predicted the same results as the original implementations inthe python development framework.}}@COMMENT{Bibtex file generated on }