Multi-Task Learning via Conic Programming

By Tsuyoshi Kato, Hisashi Kashima, Masashi Sugiyama, Kiyoshi Asai. 2007

113 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): “For example, the methods in [73], [74], [75], [76], [77], which utilize a priori information on task relations, directly learn W and b by defining g(Σ) = 0, some task clustering methods [65], [69] identify task clusters by assuming that Σ has a block structure, several task relation learning methods”
A Convex Formulation for Learning Task Relationships in Multi-Task Learning (2010, methodology): “The objective functions of the methods in [12, 11, 17, 16] are all of the following form which is similar to that of problem (7): with different choices for the formulation of 𝑓 ( ⋅ ) .”
A Comprehensive Survey on Transfer Learning (2019, background): “In light of the manifold assumption [30] and the graph-based regularizer [117], [118], fast-DAM designs a domain-dependent regularizer.”
Multi-Task Learning and Algorithmic Stability (2015, methodology): “On the other hand, problem (8) has been studied by many existing works, such as (Evgeniou and Pontil 2004; Evgeniou, Micchelli, and Pontil 2005; Kato et al. 2007) which assume that Ω is given as a priori information and (Jacob, Bach, and Vert 2008; Zhang and Yeung 2010a; Solnon, Arlot, and Bach 2012”
Parallel Multi-task Learning (2015, methodology): “We classify regularized multi-task algorithms into two main categories: learning with feature covariance [5], [7], [6], [8], [9], [10], [11], [19], [20], [21], [22] and learning with task relations [2], [3], [4], [13], [15], [14], [17], [18], [16].”
$\ell_{p}-\ell_{q}$ Penalty for Sparse Linear and Sparse Multiple Kernel Multitask Learning (2011, background): “Several works have provided empirical evidence on the benefit of such a framework [9], [16], [ 24 ], [38].”
Revisiting Stein's paradox: multi-task averaging (2014, background): “The most closely related work is that of Sheldon (2008) and Kato et al. (2008), where the regularizer or constraint, respectively, is T∑ r=1 T∑ s=1 Ars‖βr − βs‖22, which is the MTA regularizer if applied to mean estimation.”
Towards Universal Object Detection by Domain Attention (2019, background): “However, the sharing is not always beneficial, sometimes hurting performance [7, 22].”
Action2Activity: Recognizing Complex Activities from Sensor Data (2015, methodology): “As an improvement over a uniform or pre-deﬁned relatedness model [ Kato et al. , 2008 ] , we adaptively learn the relatedness among tasks.”
Graphical Multi-Task Learning (2008, methodology): “also proposed a similar method for multi-task learning with a network of tasks [7]; their graph-based regularization is very similar to GMTL except that it penalizes the maximum of a number of terms (one for each edge) instead of the sum.”
Classifier learning with hidden information (2015, background): “The task relationships are generally characterized as model parameter relationships, which can be manually specified [18] or automatically learned from data [19, 20, 21].”
A Regularization Approach to Learning Task Relationships in Multitask Learning (2014, methodology): “Relationships with Existing Regularized Multitask Learning Methods Some existing multitask learning [Evgeniou and Pontil 2004; Evgeniou et al. 2005; Kato et al. 2008; Jacob et al. 2008] also model the relationships between tasks under the regularization framework.”
Ordinal Regression Methods: Survey and Experimental Study (2016, methodology): “existing proposals, MTL via conic programming [78], was validated in the context of ordinal regression, showing promising results.”
Multisubject Learning for Common Spatial Patterns in Motor-Imagery BCI (2011, methodology): “To this end, we will use some ideas of multisubject learning, an active topic in machine learning [10, 11].”
Convex Multitask Learning with Flexible Task Clusters (2012, background): “This can be further extended to the case where task relationships are represented in the form of a network (Kato et al., 2007).”
Domain adaptation from multiple sources via auxiliary classifiers (2009, background): “The regularizers de.ned in our DAM and MTL are different in the following aspects: 1) MTL simultaneously learns two functions fs and fT (see (4)) and the two functions are compared in the same RKHS H.”