• Classified by Topic • Classified by Publication Type • Sorted by Date • Sorted by First Author Last Name • Classified by Funding Source •
Fast Adaptive Multitask Optimization.
Bo Liu, Yihao Feng, Peter
Stone, and Qiang Liu.
In Neural Information Processing Systems Foundation, July 2023.
One of the grand enduring goals of AI is to create generalist agents that canlearn multiple different tasks from diverse data via multitask learning (MTL).However, in practice, applying gradient descent (GD) on the average loss acrossall tasks may yield poor multitask performance due to severe under-optimizationof certain tasks. Previous approaches that manipulate task gradients for a morebalanced loss decrease require storing and computing all task gradients (O(k)space and time where k is the number of tasks), limiting their use in large-scalescenarios. In this work, we introduce Fast Adaptive Multitask Optimization(FAMO), a dynamic weighting method that decreases task losses in a balanced wayusing O(1) space and time. We conduct an extensive set of experiments coveringmulti-task supervised and reinforcement learning problems. Our results indicatethat FAMO achieves comparable or superior performance to state-of-the-artgradient manipulation techniques while offering significant improvements in spaceand computational efficiency. Code is available athttps://github.com/Cranial-XIX/FAMO.
@InProceedings{bo_liu_neurips_2023,
author = {Bo Liu and Yihao Feng and Peter Stone and Qiang Liu},
title = {Fast Adaptive Multitask Optimization},
booktitle = {Neural Information Processing Systems Foundation},
year = {2023},
month = {July},
location = {New Orleans, United States},
abstract = {One of the grand enduring goals of AI is to create generalist agents that can
learn multiple different tasks from diverse data via multitask learning (MTL).
However, in practice, applying gradient descent (GD) on the average loss across
all tasks may yield poor multitask performance due to severe under-optimization
of certain tasks. Previous approaches that manipulate task gradients for a more
balanced loss decrease require storing and computing all task gradients (O(k)
space and time where k is the number of tasks), limiting their use in large-scale
scenarios. In this work, we introduce Fast Adaptive Multitask Optimization
(FAMO), a dynamic weighting method that decreases task losses in a balanced way
using O(1) space and time. We conduct an extensive set of experiments covering
multi-task supervised and reinforcement learning problems. Our results indicate
that FAMO achieves comparable or superior performance to state-of-the-art
gradient manipulation techniques while offering significant improvements in space
and computational efficiency. Code is available at
https://github.com/Cranial-XIX/FAMO.
},
}
Generated by bib2html.pl (written by Patrick Riley ) on Fri Jul 12, 2024 23:00:34