Efficiency to Design and Implementation of Signal Processing Applications
Jani Boutellier is working as an assistant professor at the laboratory of Pervasive Computing. He joined TUT in August 2016 and has been working on his Academy of Finland post-doctoral research project that aims to bring together dataflow programming and GPUs.
Besides this postdoc project, Jani Boutellier has been studying the problem of fitting convolutional neural networks to small-scale embedded devices. This line of work got a significant boost in September with the start of the Academy of Finland CoefNet project that is running in collaboration with Aalto University, the TUT signal processing lab and our FidiPro Shuvra S. Bhattacharyya. Besides CoefNet, also the European COMPACT project has similar objectives, but is carried out together with four Finnish companies and a dozen of European collaborators. All of these projects have an emphasis on model-based design that aims to reduce the effort of application design without sacrificing efficiency.