Adaptive Smoothed Online Multi-Task Learning

By Keerthiram Murugesan, Hanxiao Liu, Jaime G. Carbonell, Yiming Yang. 2016

39 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

A Survey on Multi-Task Learning (2017, methodology): “An online multi-task learning algorithm is proposed in [182] to jointly learn the per-task model and the task relations”
Benchmark Environments for Multitask Learning in Continuous Domains (2017, background): “Additionally, by training on multiple tasks, the agent can exploit common traits to gain efﬁciency and generalize to unseen tasks (Caruana, 1998; Murugesan et al., 2016; Finn et al., 2016).”
Confidence Weighted Multitask Learning (2019, methodology): “Different from ﬁrst-order techniques (Murugesan et al. 2016), this algorithm captures the second-order information by exploiting the conﬁdence of weight parameters over Gaussian distribution.”
Distributed multi-task classification: a decentralized online learning approach (2017, background): “Since the relationship between tasks may not be static in the long run, we also propose a dynamic relationship matrix B with a smooth updating rule [22] 2, B−1 i,t = αB −1 0 + (1− α)Pi,t, (12) where B−1 0 refers to an initial matrix that can be set to be I or any fixed stochastic matrix and Pt is th”
Asynchronous Online Federated Learning for Edge Devices with Non-IID Data (2019, methodology): “To better model task relationships, Lugosi et al. [25] impose a hard constraint on K simultaneous actions taken by the learner in the expert setting; Agarwal et al. [26] use matrix regularization and Murugesan et al. [27] learn task relationship matrix automatically from the data.”
Asynchronous Online Federated Learning for Edge Devices (2019, methodology): “…model task relationships, Lugosi et al. [17] imposed a hard constraint on the K simultaneous actions taken by the learner in the expert setting, Agarwal et al. [31] used matrix regularization and Murugesan et al. [26] proposed a method to learn task relationship matrix automatically from the data.”
Towards more Reliable Transfer Learning (2018, background): “It is crucial to learn task relationships in multi-task learning problems, either used as priors [5] or learned adaptively [19].”
Robust Online Multi-Task Learning with Correlative and Personalized Structures (2017, methodology): “We also observed that ROMCO-NuCL/LogD are slightly slower than CW-COL, OMTLVON/LOG and OSMTL-e.”
Online Learning: A Comprehensive Survey (2018, methodology): “Many later works improve this by considering more complex inter-task relationships [151].”