We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Artificial Intelligence

Title: Exceeding the Limits of Visual-Linguistic Multi-Task Learning

Abstract: By leveraging large amounts of product data collected across hundreds of live e-commerce websites, we construct 1000 unique classification tasks that share similarly-structured input data, comprised of both text and images. These classification tasks focus on learning the product hierarchy of different e-commerce websites, causing many of them to be correlated. Adopting a multi-modal transformer model, we solve these tasks in unison using multi-task learning (MTL). Extensive experiments are presented over an initial 100-task dataset to reveal best practices for "large-scale MTL" (i.e., MTL with more than 100 tasks). From these experiments, a final, unified methodology is derived, which is composed of both best practices and new proposals such as DyPa, a simple heuristic for automatically allocating task-specific parameters to tasks that could benefit from extra capacity. Using our large-scale MTL methodology, we successfully train a single model across all 1000 tasks in our dataset while using minimal task specific parameters, thereby showing that it is possible to extend several orders of magnitude beyond current efforts in MTL.
Comments: 10 pages, 7 figures
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
MSC classes: 68T07
ACM classes: I.2.6; I.2.7; I.2.10
Cite as: arXiv:2107.13054 [cs.AI]
  (or arXiv:2107.13054v1 [cs.AI] for this version)

Submission history

From: Cameron R. Wolfe [view email]
[v1] Tue, 27 Jul 2021 19:42:14 GMT (1408kb,D)

Link back to: arXiv, form interface, contact.