Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2109.11678

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Machine Learning

arXiv:2109.11678 (cs)
[Submitted on 21 Sep 2021 (v1), last revised 18 Jul 2022 (this version, v3)]

Title:Improved optimization strategies for deep Multi-Task Networks

Authors:Lucas Pascal, Pietro Michiardi, Xavier Bost, Benoit Huet, Maria A. Zuluaga
View a PDF of the paper titled Improved optimization strategies for deep Multi-Task Networks, by Lucas Pascal and Pietro Michiardi and Xavier Bost and Benoit Huet and Maria A. Zuluaga
View PDF
Abstract:In Multi-Task Learning (MTL), it is a common practice to train multi-task networks by optimizing an objective function, which is a weighted average of the task-specific objective functions. Although the computational advantages of this strategy are clear, the complexity of the resulting loss landscape has not been studied in the literature. Arguably, its optimization may be more difficult than a separate optimization of the constituting task-specific objectives. In this work, we investigate the benefits of such an alternative, by alternating independent gradient descent steps on the different task-specific objective functions and we formulate a novel way to combine this approach with state-of-the-art optimizers. As the separation of task-specific objectives comes at the cost of increased computational time, we propose a random task grouping as a trade-off between better optimization and computational efficiency. Experimental results over three well-known visual MTL datasets show better overall absolute performance on losses and standard metrics compared to an averaged objective function and other state-of-the-art MTL methods. In particular, our method shows the most benefits when dealing with tasks of different nature and it enables a wider exploration of the shared parameter space. We also show that our random grouping strategy allows to trade-off between these benefits and computational efficiency.
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Cite as: arXiv:2109.11678 [cs.LG]
  (or arXiv:2109.11678v3 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2109.11678
arXiv-issued DOI via DataCite

Submission history

From: Lucas Pascal [view email]
[v1] Tue, 21 Sep 2021 09:34:14 UTC (2,669 KB)
[v2] Mon, 4 Oct 2021 15:37:46 UTC (2,667 KB)
[v3] Mon, 18 Jul 2022 15:54:39 UTC (2,669 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Improved optimization strategies for deep Multi-Task Networks, by Lucas Pascal and Pietro Michiardi and Xavier Bost and Benoit Huet and Maria A. Zuluaga
  • View PDF
  • TeX Source
  • Other Formats
license icon view license
Current browse context:
cs.LG
< prev   |   next >
new | recent | 2021-09
Change to browse by:
cs
cs.AI

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar

DBLP - CS Bibliography

listing | bibtex
Pietro Michiardi
Xavier Bost
Benoit Huet
Maria A. Zuluaga
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack