Deep networks have shown outstanding scaling properties both in terms of data and model sizes: larger performs better. Unfortunately, the computational cost of current state-of-the-art methods is prohibitive. A number of new techniques have recently arisen to address and improve this fundamental quality-cost trade-off. Methods like conditional computation, adaptive computation, dynamic model sparsification, and early-exit approaches aim to address the above mentioned quality-cost trade off. This workshop explores such exciting and practically-relevant research avenues. As part of contributed content we will invite high-quality papers on the following topics: dynamic routing, mixture-of-experts models, early-exit methods, conditional computations, capsules and object-oriented learning, reusable components, online network growing and pruning, online neural architecture search and applications of dynamic networks (continual learning, wireless/embedded devices and similar topics).
The 1st Dynamic Neural Networks workshop will be a hybrid workshop at ICML 2022 on July 22, 2022. Our goal is to advance the general discussion of the topic by highlighting contributions proposing innovative approaches regarding dynamic neural networks.
- Workshop schedule is announced!
- Accepted papers (poster and oral presentations) are announced. Congratulations to all authors!
- Microsoft CMT Submission portal is now open!
Submission Deadline: May 31, 2022 (Anywhere on Earth)
Author Notification: June 13, 2022
Video Deadline: June 28th, 2022
Camera Ready Deadline: July 9, 2022
Workshop Day: July 22, 2022
Speakers (More Info)
Sapienza University (Rome)
Franklin & Marshall College
Warsaw University of Technology & Jagiellonian University & Tooploox
- Canwen Xu, UC San Diego
- Yigitcan Kaya, University of Maryland
- Maciej Wolczyk, Jagiellonian University
- Bartosz Wojcik, Jagiellonian University
- Yoshitomo Matsubara, Amazon Alexa AI
- Thomas Verelst, KU Leuven