We invite you to submit open challenges in deploying and monitoring ML systems which you feel are only partially, or are wholly unsolved at the present time. These problems might be algorithmic, computational, programmatic or hardware-based in nature. However, in all cases they should represent a significant blocker in the continual use of a machine learning system.
High level topics submitted problems might relate to include:
Updating machine learning systems, or managing the life-cycle of machine learning systems, whether manually or automatically (potentially using machine learning itself)
Machine learning systems for resource-limited hardware, e.g. industrial devices or low memory/low power devices
Detecting, and responding to changes in the data, e.g. data drift, or adversarial attacks
Detecting and responding to unexpected events, e.g. oddball or missing data
Ensuring an ML system is and continues to be compliant with any existing regulations, e.g. with respect to regulations about data privacy
Measuring the impact of deploying an ML system, e.g. how to measure the carbon-footprint of a deployed ML system
If possible, in addition to describing the problem, you should give examples where your submitted problem has proved a blocker to success deployment or monitoring of an ML system.
After the submission deadline, we will collate all submissions, and publish a register of problems on the workshop’s website. We will then invite the authors of the submissions representing the most important and challenging problems to participate in a live broadcast panel discussion during the workshop.
--> SUBMISSIONS CLOSED <--
We invite submissions of full papers on machine learning systems, deployment and testing, including:
MLOps for deployed ML systems
ethics around deploying ML systems;
fairness, trust and transparency of ML systems;
providing privacy and security on ML Systems;
useful tools and programming languages for deploying ML systems;
deploying reinforcement learning in ML systems
performing continual learning and providing continual delivery in ML systems;
data challenges for deployed ML systems
All papers submissions will be made in PDF format, with a limit of four pages, including figures and tables, excluding references and appendices.
Formatting instructions are provided in the ICML website. The reviewing process will be blind and the workshop allows re-submissions of already published work and double submission.
Paper submissions can be made through the EasyChair system (you must create an easychair account in case you do not have one already).
--> SUBMISSIONS CLOSED <--
Paper submission deadline EXTENDED: 17th of June, 2020
Open problems deadline: 30th of June 2020
Acceptance notification: 30th of June, 2020
Workshop: 17th of July 2020
All submissions will be made in PDF format:
Papers are limited to four pages, including figures and tables, excluding references and appendices.
Open problems are limited to tow pages.
Formatting instructions are provided in the ICML website. The reviewing process will be blind.
The workshop allows re-submissions of already published work and double submission.
Accepted papers (plus the optional supplementary material) will be made available on the workshop website as non-archival reports.
All accepted papers will be presented at the workshop during the virtual poster sessions. A selected number of accepted papers will be presented during the oral session in the format of pre-recorded videos. The remaining accepted papers will be allocated a slot during the poster spotlight session. The accepted open problems will be presented during an allocated panel discussion.
Please note that this is a virtual event. Please refer to the ICML virtual schedule plan. More information about the format of the presentations will be released soon.