Co-organised by JOANNEUM RESEARCH, the Joint Workshop on On-Device Machine Learning & Compact Deep Neural Network Representations took place on June 14th, 2019, as part of the 36th International Conference on Machine Learning in Long Beach, CA, USA, one of the world's leading scientific events in the field of artificial intelligence and machine learning.
Similar to the workshop organised at NeurIPS 2019, this workshop brought together more than 150 researchers and practitioners working on the compression of neural network and optimised architectures for hardware acceleration of neural networks. With the rapidly growing number of applications for neural networks, there is an increased need for scalability, and for enabling the usage of these technologies on resource-constrained devices such as mobile phones, smart cameras or embedded processors in vehicles. In addition, in content analysis applications like the ones used in MARCONI, local processing can solve privacy issues, as the input data does not need to be sent to a central server.
The workshop hosted invited talks with speakers from MIT, Google, IBM and NVIDIA, five oral presentations, a poster session and a panel discussion on the research challenges that lie ahead. The participants remarked that there is a growing number of target hardware platforms (e.g., specialised tensor processing units) that come with their own specific toolkits for optimisation, which raises interoperability challenges. Thus standardisation activities on exchange formats and representation of compressed networks are required. One example is the standardisation activity in MPEG, to which JOANNEUM RESEARCH is contributing.
The workshop has been live streamed, and recorded talks are available via this link.