Call For Papers
Constraints introduced by low resource environments such as developing countries (Africa, South East Asia, South America etc.) can necessitate alternative approaches to conventional ML problems. We encourage papers that address (but are not limited to) the following topic areas.
Algorithms and Methods
- Methods for collecting and generating training data within data scarce (limited labeled data) settings (such as weak labels, model-based pre-labeling, teacher-student models, and transfer learning).
- Machine learning techniques applied to limited data (e.g. active learning, few-shot and zero-shot learning).
- Approaches to training and inference on resource constrained devices (such as model quantization, model compression, model distillation, low precision training, model pruning methods, and generalized model optimizations).
- Alternative learning methods coupled with deep models targeted for low resources settings.
- Automated techniques to stratify and valuate data in order to increase throughput in low-resource settings.
- Analyse models in the perspective of fairness, explainability, etc.
Industry Experience and Applications
- Data science and engineering practices that help balance accuracy/latency tradeoffs while scaling ML models in low resource environments.
- Measuring success or impact that goes beyond algorithmic metrics (such as accuracy or F1 score).
- Data-driven techniques that support public institutions (government transparency, healthcare, education etc).
Social and Policy Topics
- Successful AI solution implementation stories which work at a small scale (e.g. local institution, city) that could be applied at larger scale.
- Connecting skilled professionals with the organizations that deeply understand the local problems.
- Securing funding for proof-of-concept (POC) projects or for scaling existing POCs.
- Building effective research and implementation teams, with a focus on challenges specific to developing regions such as countries in Africa.
- When machine learning is NOT a viable option.
- Strategies and policies enabling or enhancing AI/ML adoptions for developing countries.
Instructions
Submission types
- Short papers and position pieces (up to 5 pages)
-
Problem statements and abstracts (up to 2 page)
- In addition, ongoing work and papers that have appeared in a non-archival venue (workshops, arXiv, etc) are welcome. You may also submit work that has appeared in an archival venue in 2020 or 2021. Please note that submissions are non-archival.
Accepted Submissions
- 5-page submissions will be eligible for oral or poster presentation.
- 2-page submissions will be presented as posters. In both cases, page limit applies to content only, excluding references.
- PML4DC will arrange registration/attendance sponsorships for authors of accepted submissions.
Submission Format and Link
Contributions should be anonymized and submitted using the ICLR template via CMT. https://cmt3.research.microsoft.com/PML4DC2022. We have also included a brief guide on how to use the CMT template.
Important Dates
- Submissions deadline (Extended):
March 6, 2022 - Notification:
March 25th, 2022 - Camera ready:
April 4th, 2022 - Workshop: April 29th, 2022
Deadlines are at 11:59 PM (Eastern Time, ET). ET is UTC-05:00.