MPP 2021: 10th Workshop on Models for Parallel Programming Portland, OR, United States, May 22, 2021 |
Conference website | http://www.mpp-conf.org/ |
Submission link | https://easychair.org/conferences/?conf=mpp20210 |
Abstract registration deadline | March 21, 2021 |
Submission deadline | March 21, 2021 |
Current trends in Computer Architecture and Parallel Programming point towards the importance of accelerating Machine Learning (ML) Algorithms. The ubiquity of ML, the amount of data treated by ML and the adoption of much more complex Artificial Neural Networks (so called Deep Learning) reinforce the importance of tackling ML problems from a performance point-of-view. Therefore, this issue of the Workshop on Parallel Programming Models (MPP) will mainly focus on works that provide acceleration to ML systems, but as usual will also welcome papers in any topic related to parallelism/acceleration.MPP is a workshop designed to explore parallel programming models, architectures, and runtime systems to enable developers to deal with these trade-offs. MPP has been held each year since 2012, co-locating with prestigious conferences such as WSCAD, SBAD-PAC, and IPDPS. It has attracted industry sponsorship (Maxeler, LG, Microsoft, NGD Systems) and top-tier keynotes (Arvind - MIT, Michael Flynn - Stanford and Jesus Labarta - BSC). MPP 2021 will be focused on Machine Learning Performance and has the potential of attracting high-quality papers and audience for fruitful discussions.When addressing the performance aspect of Machine Learning, there is also the issue of the amount of data used for training deep-learning models. In the case of Big Data, the application of in-memory computing (which was the main topic in MPP 2019) can be essential to reduce the gap between data and the ML model, in terms of latency. MPP 2021 aims at bringing together researchers interested in presenting contributions to the evolution of existing models or in proposing novel ones, considering the trends on Machine Learning, In-Memory Computing and Security. MPP 2021 will be held in conjunction with The 35th IEEE International Parallel and Distributed Processing Symposium (IPDPS 2021), in Hilton Portland Downtown, Portland, Oregon, United States, on Friday, May 21.Submission GuidelinesMPP invites authors to submit unpublished full (8 pages maximum) or short (4 pages maximum) papers on the subject. Submitted manuscripts must be single-spaced double-column pages using 10-point size font on 8.5x11 inch pages (IEEE conference style), including figures, tables, and references. The submitted manuscripts should include author names and affiliations. Papers must be submitted by Feb, 18, 2020, in the following url: easychair.org/conferences/?conf=mpp2021
List of TopicsTopics of interest include (but are not limited to):
- Compression of Deep-Learning Models;
- Tools for ML Model design;
- Hardware specifically designed for Machine Learning;
- In-Memory Computing;
- Novel Deep Neural Networks architectures;
- Error Detection/Recovery in ML systems;
- Robust Neural Networks;
- Privacy of data in ML systems;
- Robustness of decision making ML systems;
- Neural networks inference and training on IoT, Fog, Edge and cloud environments;
- Machine Learning for Parallel Applications and IoT.
The proceedings of MPP 2021 will be distributed at IPDPS 2021 and will be submitted for inclusion in the IEEE Xplore after the conference.