Online Multi-task Learning with Hard Constraints

By Gábor Lugosi, Omiros Papaspiliopoulos, Gilles Stoltz. 2009

52 citations recorded by the source corpus.

We discuss multi-task online learning when a decision maker has to deal simultaneously with M tasks. The tasks are related, which is modeled by imposing that the M-tuple of actions taken by the decision maker needs to satisfy certain constraints. We give natural examples of such restrictions and then discuss a general class of tractable constraints, for which we introduce computationally efficient ways of selecting actions, essentially by reducing to an on-line shortest path problem. We briefly discuss "tracking" and "bandit" versions of the problem and extend the model in various ways, including non-additive global losses and uncountably infinite sets of tasks.

Subfields: Artificial Intelligence, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Management Science and Operations Research

Citations in the corpus

An Overview of Multi-task Learning (2018, neutral): “In[76],the proposedonlineMTLalgorithmsmodeltaskrela-tionsbyplacingconstraintsonactionstakenforall thetasks.”
A Survey on Multi-Task Learning (2017, methodology): “In [154], online MTL algorithms are devised when the relatedness of all the m tasks is modeled by constraining that the m-tuple of actions for tasks needs to satisfy some hard constraints.”
Robust Online Multi-Task Learning with Correlative and Personalized Structures (2017, methodology): “As a result, online learning algorithms are efficient and scalable, and have been successfully applied to a number of MTL applications [7], [8], [9], [10], [11], [12].”
Distributed multi-task classification: a decentralized online learning approach (2017, background): “Other OMTL literature can be found in Li et al. (2014) where a collaborative model is proposed and Lugosi et al. (2009) where OMTL with hard constraints is studied.”
Online Multitask Relative Similarity Learning (2017, unknown): “However, the performances of these algorithms were usually limited due to the adopted underlying learning scheme (i.e., Per-ceptron) [Ammar et al. , 2014; Lugosi et al. , 2009].”
Sequential Transfer in Multi-armed Bandit with Finite Set of Models (2013, background): “A number of works (Dekelt al., 2006; Saha et al., 2011; Cavallanti et al., 2010; Lugosi et al., 2009) focused on this setting and showed how the samples coming from different tasks can be used to perform multi-task learning and improve the worst-case performance of an online learning algorithm…”
Adaptive Smoothed Online Multi-Task Learning (2016, methodology): “al [7] imposed a hard constraint on the K simultaneous actions taken by the learner in the expert setting, Agarwal et.”
Online Learning of Multiple Tasks and Their Relationships (2011, background): “Another work [Lugosi et al., 2009] along similar lines extended the notion of experts to the set of decisions the forecaster is allowed to take.”