Development of Sample-Efficient Algorithms for Training of Neural Networks
grant ID: PT01229
Project leader: Rafał Wolniak
Implementers:
- Rafał Wolniak
Politechnika Gdańska
Wydział Elektroniki, Telekomunikacji i Informatyki
Gdańsk
Start date: 2025-02-21
Planned end date: 2028-02-21
Project summary
The study focuses on the development of algorithms, that require less interactions with training data. At this point of the study, the approach is focused on estimating more accurate influence on loss of potential updates of trainable parameters of a model. Having the knowledge of the influence on loss in terms of each model parameter, the weight update can be more relevant. The approach may enable to maintain appropriate direction of model updates for high values of learning rate, increasing sample efficiency of training.
CONTACT
Our consultants help future and novice users of specialized software installed on High Performance Computers (KDM) at the TASK IT Center.
Contact for High Performance Computers, software / licenses, computing grants, reports:
Administrators reply to e-mails on working days between 8:00 am – 3:00 pm.